This week I spoke at MicroConf, breaking down how AI changed onboarding and activation. One thing I noticed is that the absolute majority of those changes lead to adding an AI chat to the product .

When I was screenshotting solutions for the research, I had a feeling that all our products started to look the same. It’s not only because we use a lot of the same brand systems, especially small startups. It’s also because the same solutions keep getting adopted.

So I started wondering, does designing everything around the AI chat take away your product’s individuality? This is something I would love to dig into today.

THE CHALLENGE

Should you add an AI chat to your product, or keep your product unique?

When I first started working with UX, one of the first pieces of advice I got was that if a specific element sits in a specific position in all the major products, you should put it in exactly the same place. For example, if the search icon is always in the top right corner, you should put the search icon in your product in the top right corner as well, because this is where people will expect to see it.

For years products did look different from one another, but the core elements were universal and recognizable. And it’s not because no one has fantasy. It’s because it’s important. It bridges the expectation gap and helps people navigate and scan the product quicker, so they spend less time learning the rules and more time actually using the product.

And I think right now we are about to be in the situation where it gets to the absolute extreme, because every product is about to look like an AI chat. Quite complex products add AI chat as their core interface decision, as their core input point.

A lot of teams hesitate about this decision because they’re afraid they have to rebuild their whole product and replace it with the chat. But this is not what this is about. And here’s why it can do more good than damage if you approach it right.

THE GAP

Researchers showed forty people screenshots of real websites they had never seen, and asked them to spot one thing as fast as they could, like the login on an online shop. On some sites it sat where most sites put it. On others it sat somewhere else. An eye tracker followed every look.

When the login sat in the usual spot, people found it in under a second. When it didn’t, it took them three and a half seconds and about 14 looks.

This is exactly what happens with the AI chat interface right now. If we don’t have a thing we can just talk to, it opens an expectation gap.

You can argue with me and say, “Hey, Anastasia, we’ve had search engines for years, and it’s not like they shaped our perception of how the product should look.”

Well. Hold on.

Almost every product on earth had and still has a search bar where you can go and search for something, because this is such a typical behavior. All of us have been doing it multiple times a day for years. And right now, that search bar turned from searching for information into an AI chat. A thing I can simply ask to do stuff for me, to save myself some time and cognitive effort.

I remember many years ago I was working on a media project, and users were begging us to add a search function to the website because we didn’t have it. But after we added it, less than 1% of people were actually using it, and they weren’t using it very often. But the absence of it was so annoying to people that they could not let it go. They wanted everything to look the same. They wanted to know that if they need it, they have this capability.

HOW TO WORK WITH IT

It doesn’t mean you have to erase your whole product and replace it with a chat. But it might be a good idea to use the chat as a door into your product.

Here’s how teams do it.

1. Make the chat the main way in.

Linear made its agent the default view. The whole home screen is one box, “Ask Linear,” with a few examples underneath. You can change the look, but if you don’t, the chat is the main way you work with the product.

2. Keep it always there, but subtle.

Miro puts a small chat bar at the bottom of every board, “Where shall we start?”, with a few suggested prompts above it. It’s easy to ignore, and when you click it, it expands. The canvas and tools stay the main thing.

I especially love that one of the suggested prompts is “Learn Miro AI.” People come to products like Miro with very different knowledge of how to use them, very different intentions, and very different images of the result they want in the first session. Sometimes figuring out what the product can do is the result. So the chat isn’t designed only around things that produce a direct output, but also around learning the product, if this is what you’re here for, so you can plan your work around it in the future.

3. Put it next to other ways in.

Zapier does something in between. The chat is the obvious option at the top of the home page, “What would you like to automate?” Below it sit other places to start, from scratch or from a recommended template, which help you get oriented if you’d rather click than ask.

N.B. There’s a concern that if people learn how to use something through the AI chat, they won’t learn the actual journey in the product. Well, maybe they don’t have to? Just make sure that if I can use some functionality through the chat, I can use it again and again, and it performs well every time, not only during onboarding. If you let me see the magic during onboarding, but then it’s impossible to recreate it in the product, that’s bad.

ONE THING TO REMEMBER

Don’t try to be unique at the cost of being inconvenient. You can follow the new UX habits without losing what makes your product special.

Recently I got a weird hyperfixation on a reality show, Clarkson’s Farm. Jeremy Clarkson hosted Top Gear for many years, and now he has a show about his farming. I won’t steal your time with the details, but it’s wildly entertaining and I really enjoy it.

What I wanted to tell you is that at some point he opened a farm shop to sell the produce he makes on the farm. And what fascinates me, he makes way less profit from the shop than when he sells his produce to other vendors, especially in bulk. can’t even be compared. But it doesn’t stop him. He still invests a lot in the shop, and he loves having it.

It got me thinking that we do the same with our products. We want to bring everything to our territory and control it. Even when it’s not necessarily the best strategy.

So today I wanted to dig into this.

THE CHALLENGE

The challenge a lot of teams have right now is should we bring users to our product, or go where they already are?

We want to own the user and anchor them to our product. There’s this belief that the more you do that, the better the retention will be. And it’s completely true. But building a new habit and a new ecosystem is way more difficult than getting embedded into something that already exists in people’s lives organically.

We are not new to integrations, duh. Everyone wanted to integrate with Slack, Google Calendar, etc. But right now we have completely new ecosystems, and they are so personalized and so easy to use that switching from them to somewhere else seems like an incredible chore.

Whether we like it or not, AI chats just put a completely new bar for every product that exists. People will judge your product against Claude, Grok and ChatGPT, against their convenience and their knowledge of context.

And I feel like we’re entering a space where integrating with these new super ecosystems won’t be a choice anymore, but a requirement.

THE GAP

So why is the chat so hard to compete with, even when your product is better at the job?

I think it’s because habits stick to places. When we do something in the same place over and over, the place starts making the decision for us. There’s a study that shows this really well.

Researchers handed out free popcorn to people watching movie trailers. Some boxes were fresh. Some were a week old, and everyone could tell it was bad. Half the people watched in a cinema. The other half watched in a plain meeting room.

So who ate the stale popcorn?

People who always eat popcorn at the movies ate it in the cinema just like the fresh one, about 63% of the box either way. The same kind of people in the meeting room ate half as much stale as fresh. They knew it tasted bad in both rooms. But the cinema setting just made them ignore it.

The AI chat is the cinema now. People already go there for everything, so inside it they ask and act without thinking twice. When you bring them to your product instead, you move them to the meeting room. Now they weigh every bite, and next to a chat that already knows them, your product has to win every single visit.

HOW TO WORK WITH IT

Claude and ChatGPT even made me adopt products that struggled to convert me before. I never adopted Notion, but right now, it’s the place where I put my release notes and all the context of my PRs before I hand them to the tech team. I do it through Claude; I don’t really do it myself, but guess what, I’m a happy customer.

So I started paying more and more attention to how products show that they can be embedded in my current workflow. There are different ways to do it, and here are three I really like.

1. Amplitude comes to you

Amplitude runs a whole new positioning about building self-improving products. Essentially they’re saying you won’t have to do anything, we will tell you what’s going on. And instead of highlighting their dashboards and how you can operate within the product, they highlight how they integrate with your Slack or Claude. They come to you, and you operate their product through the platforms you’re already using.

2. Granola moves the notes to where they’re needed

Granola is positioned as an AI note taker, and they do a very good job of being embedded in your current workspaces. I love that they have the MCP connection, because they understand that within the AI ecosystem the handoff is very important. Meeting notes on their own don’t matter much. We need to share them somehow, and no one wants to go and copy them. Everyone wants to integrate them, for example with an AI chat that’s connected to their task manager or sends updates to Slack.

Questions? Leave a comment, I’ll answer.

Leave a comment

3. PostHog lets you set it up with AI

PostHog just highlights that you can install and set up their system with AI. I think it’s very important for people who are AI native right now. They don’t want to dig through the documentation or go through the onboarding. They just want to paste the code and operate the new product through their current ecosystem.

And what’s important here: it must be a cohesive positioning that is connected from your website to your product. Admit that people are using products in new ways and show them how your tool won’t be an additional old-time headache.

Leave a comment

ONE THING TO REMEMBER

Don't teach people your way when you can match theirs. Get embedded in the journey they already have, and I promise you'll get better adoption and happier users.

I’m sure, like all of you, I’m using Claude heavily with all the tools I’m integrated with, and one of them is Amplitude. This week I was updating an event taxonomy system for one of my current clients, and I wanted to build the dashboards based on that new taxonomy.

Claude in Amplitude could build the dashboard, but it could not change it. So the first version was there, but it wasn’t what I needed yet. To get there, I needed small adjustments, for example changing the timing of a funnel. Instead, I had to either delete the dashboard and rebuild it, or go and do it manually. And when you have 20 plus charts on different dashboards, it’s a tedious thing that I would love to ask Claude to do. I know that Amplitude has its own AI, but for me, honestly, it works worse, and I want to do everything from the same place.

Btw, if you want to find a better way to leverage your analytics, my ex-colleague Mario Araujo, a data wizard from Flosum and ProductLed, recently launched his own course. Feel free to check it out here: https://maven.com/s/course/a49ee14042

Anyway, this whole thing got me thinking that even when the zero to one output is good, it’s still just the first step. And when I have no control over the next steps, I get frustrated. It’s like, here’s my final output, take it or leave it. Have you ever felt this way?

Leave a comment

So today I wanted to dig into this topic.

THE CHALLENGE

“Should we make the result effortless?” Sounds like a stupid question, right?

Actually, not always.

Let’s take a look.

There is a difference between the output and the final result. When we talk about output, we usually think about the first thing we receive after the initial input. But come on, everyone who works with AI knows that it’s actually a step towards the final result.

Recently, for one of my clients, I’ve been interviewing users who are looking to buy expensive property soon. Because we are building an agentic workflow, I asked them, are there any decisions you would never delegate to AI? And they actually named a couple. Some of them weren’t ready to delegate the financial analysis. Some weren’t ready to let it sign an NDA. Quite often, they weren’t ready to delegate the questions where they had specific requirements. And of course, no one was ready to delegate the final decision.

And that’s the thing. I think it’s not only because AI cannot do a good job with some of them. It’s mainly about our own sense of control. The more important the job the product does, the more significant it is in our life, the more we want to have that sense of control.

THE GAP

The expectation gap here happens between the output and the final result. And how big it is depends on the job. If you’re generating a QR code, or changing the format of a document, I’m sure the gap between the output and the result is very small. But if you’re designing a website or a prototype, the gap between the first generated version and the final result you’ll be satisfied with is pretty large.

To get from the output to the result, I need to steer. Seeing that I get exactly what I ask for helps me generate the feeling of control. But when the changes don’t correlate with what I asked, and I have a feeling that the system behaves uncontrollably, I feel insecure, I feel threatened, and I feel like I cannot trust it.

But does it actually have psychological grounds, or is it just my personal perception?

It actually does. Research from 2018 shows that people use an imperfect model way more often if they can affect the final result. When people couldn’t change the model’s answers, 32% chose to use it. When they could adjust them, 76% did. And it held even when they could only adjust each answer by a tiny bit.

So we get the result we want with higher probability, and it also gives us the sense of control, the sense of collaboration. That’s how we get a relationship we can trust and rely on. It’s a reliability issue, essentially.

HOW TO WORK WITH IT

There is no silver bullet here, because all products are different. But let me try to draft some possible tactics based on how habit-driven your product already is.

So the larger the gap, the more we should allow our users to steer the wheel, and we should not pretend that we can ignore it. Ignoring it will just result in a lot of bad outputs that are still so far away from the result that the result is never achievable.

1. Let me select the thing I want to work with.

When Magic Patterns just started, you could only ask for changes in the chat and hope it would find the right element. Right now they have a visual editor where you select the specific element you want to work on.

2. Let me edit it myself.

You can actually go and edit the thing yourself and just save the changes, or confirm them with AI

3. Let me ask AI to change just that part.

You can select a specific element and ask an additional question about it, without touching the full output. It will save me credits, and it will save you credits. And it will give me the sense of control that will likely motivate me to come back later.

Leave a comment

ONE THING TO REMEMBER

Output isn’t always the result, and it might take additional work to create something users will want to return to.

I feel like there is a shift in PLG happening right now. For a very long time, we’ve all been obsessing about the first session, and don’t get me wrong, it’s been very important. But I feel like there are more and more questions linked to habit formation and retention.

And I have bad news for you.

Resolving them actually takes way more time and way more trial and error than working within the first session, which is still hard.

So today I wanted to dive into this challenge and look at a couple of strategies that actually help you work with it.

THE CHALLENGE

The first session is incredibly important, and it’s hard enough to deliver the value within it and produce good results. But if you look at the overall health of the product, compounding growth, recurring upgrades instead of upgrades within the first session, satisfaction is not enough.

Everyone keeps saying how activation is better with AI and how much faster we can deliver the value. But according to ChartMogul’s retention report, AI-native products consistently have much higher churn.

When we see something like this, we drift to the question: is the output good enough? And don’t get me wrong, sometimes it is the problem.

But way more often, we are so obsessing about delivering the first value as fast as possible that we forget to establish the returning loop.

On the landing page we see “transcribe the meeting in seconds”, “build the presentation in seconds”. People sign up, get that exact result, and the loop is closed. And with so many products on the market for every single solution, there is no guarantee they will even remember us the next time the problem comes back.

So, what should we do when a good result isn’t good enough anymore?

THE GAP

In 2005 marketing researchers asked kind of the same question: do satisfied customers buy more?

Everyone assumes yes. So they tested it.

They took the customers of a retailer and asked them two things. How satisfied are you? And are you planning to come back?

Sneaky enough, they didn’t stop at what people said. They followed what the same customers actually did over the following months: did they come back and buy again, or not.

What do you think happened?

With what people said, it worked exactly like we all expect. The happier the customer, the more they said they’d come back.

But with what people actually did, satisfaction on its own didn’t predict anything. Happy customers didn’t come back just because they were happy. It only turned into real visits when other things lined up, like how convenient it was to come back and how much competition was around.

So essentially this is the gap between "I know exactly when and why I will use it next time" and where they actually are: they just received everything they needed, and they're ready to stop thinking about this product forever.

HOW TO WORK WITH IT

There is no silver bullet here, because all products are different. But let me try to draft some possible tactics based on how habit-driven your product already is.

1. The trigger already exists in their week.

Some products will inevitably be linked to environmental triggers, and in this case our job is just to connect them to that existing process.

Granola, for example, kind of gets embedded into the system that you have already. Instead of going with some one-time solution, like record this meeting right now, transcribe this meeting right now, they get embedded into your current flow. And they establish it from the landing page. It doesn’t lead with anything quick or one-time, it kind of establishes that they’re selling the system for the process that runs on a daily basis.

So after you sign up and connect your calendar, you actually see all of your upcoming meetings put together, so Granola would be able to join them.

So from the first screen, it's not a one-time solution, it's something that will be interactive with your system.

This is also how Claude and other tools are trying to drive agent adoption right now: they prompt you to schedule the task that they saw you repeating.

2. There’s no outside trigger, but the work is ongoing.

If we talk about the usage-based thing, then credits renewal can be really helpful for a free tier. This is what Lovable does, and a lot of other AI services: you have the updated limit every day, so you have a reason to come back every day.

I even know people who build just within the free limits, and if it takes them a couple of months, it will take them a couple of months, but they're dedicated to build within the free limits.

3. There’s no outside trigger, and the result is one-time.

For example, you build a prototype once to show your point, and it’s not clear when you will need it again. So make it easy to take the next step with the result.

Sometimes the next step is sharing it and getting feedback. Sometimes it’s taking the result and applying it somewhere else. If that next step happens through your product, they have a reason to come back.

For example, the reason why for a very long time I was using Magic Patterns instead of Claude Code was because it was easier to share the designs. And what I really loved, my clients could go and leave comments directly on the designs, and I could add my comments, clarifying some of the suggestions and giving the context. So you go from "I created a design" to "I created a collaborative ground for our future work". And in this case, I would be prompted more to come back and keep working there, because it was so easy to access for everyone.

Gamma and Loom send you analytics updates when someone watched your video/presentation, so you have a reason to come back and check who it was and what happened there.

ONE THING TO REMEMBER

Delivering first value is important but not nearly enough for long-term growth.

More embarrassing, they actually have a need for them, but they solve it with other methods, from spreadsheets to ai chats.

Ohhh my god.

THE CHALLENGE

We all talk a lot about how to onboard new users, because this is usually where the conversion is, and where the growth team spends its energy. But the underestimated move is expansion: the revenue we get from existing accounts as they grow into more of the product over time.

Whenever we enrich our product with something new we hope that since these users already know our product, they’ll discover the new value themselves.

But they don’t.

And they especially don’t when it comes to AI, because AI implementation quite often doesn’t add a new feature. It reimagines how you use the features you already have, and it reimagines the established ways of working that your users have spent months or years building.

You might be thinking: hey, Anastasia, it doesn’t look so bad.

But it is.

If users don’t think they have space to grow in your product (solve more tasks, or solve them in a better way), they will leave it for something where it looks obviously possible.

And here’s why.

THE GAP

In 2008 researchers put expert chess players in front of a board with two ways to win.

1. One was a familiar five-move checkmate every strong player has seen.

2. And the other was a less familiar three-move mate that was better.

They had to find the shortest one, and eye-tracking recorded what they looked at.

Most of them spotted the familiar solution quickly, and then said they were still searching for something faster.

But….

Their eyes kept going back to the squares tied to the solution they already had, and not to the ones that would have revealed the better move. They believed they were open to a better option, but their attention had already locked onto the familiar one.

On Ghosted, I usually write about the expectation gap that we need to close. But here, expectation actually plays a wrong trick with us.

People are getting driven to the familiar option because it’s the closest thing to what they expect from our product. They don’t expect us to launch new features by default, because they don’t really think about the product between sessions (I know, hurts every time).

So even when we introduce something better, if it’s not presented in the right way, they’ll still fall back on the old workarounds.

HOW TO WORK WITH IT

Amplitude does two things I really love, and both are worth stealing.

1. Trigger on the behaviour, not on a schedule.

Honestly, a lot of people right now are saying very bad things about tooltips, but guess what, they actually get the job done if they're not static. If you set them up to react to behaviour and suggest a natural next step, why not use them? Fashion aside, almost every PLG’s darling uses them, just in more organic formats than they existed years ago.

Here's an example I really like. The Amplitude bubble shows up in an account I've been actually using, and it appears specifically when I'm doing something manually that could be done in a smarter way. I'm checking a dashboard (I switched to a different tab to hide the data hehe), and it just suggests how I could do the same thing differently and how it can save me time. It's tied to what I'm looking at in that moment, not sent to everyone on a Tuesday morning.

2. Pre-open the new feature so it doesn’t require intent to find.

When I open the dashboard, they also open the AI chat by default, and they start it with a very universal question that I can ask about this dashboard. It’s very easy to start, because it’s right there, and it doesn’t require any additional effort from me. Putting the new option on the surface I already opened solves the attention problem before it happens.

If you’re not ready to build any of this, at least stop treating lifecycle emails as the whole answer. Emails are fine as an add-on, but if you still have a product that people use manually, either through your MCP or by logging into the platform, you should be treating the advanced journey the same way you treated the first one, just for more advanced tasks, so you can use any other instruments at your disposal.

An important note: don’t launch an advanced sequence for someone who hasn’t pushed through the initial steps. I saw many times how two of those collapse because of the wrong triggers, and the user ends up with multiple checklists, guides, etc.

ONE THING TO REMEMBER

If someone loves one feature, they don’t automatically become an expert on the whole product. Use cases end, competitors come, and ongoing onboarding can bring as much growth (if not more) as the one within the first session.

How to get back someone we already lost?

Because every product loses users. And it’s actually completely normal. But churn comes in many shapes, and not all of them are hopeless if we approach it the right way.

THE CHALLENGE

Every product gets to reactivation eventually, because nobody’s churn is getting smaller, and least of all in AI products. And when we get there, the whole thing tends to land on lifecycle marketing, whose job is to earn the click and find the trigger that brings people back.

Somewhere along the way we decide that getting them back into the product is enough, and that they’ll figure out the rest from there, because after all they’ve used this before.

So we pretend they never left and drop them right where they stopped the last time. At a half-created project, mid-way integration, unfinished design, etc. We even lead with it in the email, because they made that thing and people care about things they made, and asking them to redo the setup they already started a year ago would be insulting.

But what happens to most of them after they log in?

They leave.

Again.

And that is why reactivation campaigns often have bad rep.

Even when the campaign works and people click, the ones who come back tend to activate worse and pay worse than someone who’d never heard of you. Which is strange, because these are the cheapest users you will ever address. You already paid to acquire them and they already understand what you do.

So, why can’t we meet their expectations by dropping them right where they left off?

THE GAP

Here’s the thing about a reactivated user: you kind of acquire them twice. If we are talking about big breaks in usage, which is my main focus in this issue, 2 complitely different triggers could have brought the same user to your product the first time and the second time.

Imagine.

You were building a website in Canva last year, then the project got deprioritized, and you forgot about it, didn’t have another design task for a long time. Today you’re back because you saw the new poster templates somewhere in an Instagram ad and decided that you want to try it. But after logging in, you are being dropped back into your messy, half-baked website flow and have to dig yourself out of that situation, find the poster flow, find the right template, etc.

For this case, it’s almost worse than doing nothing, because now you are being forced into something that isn’t relevant to you, and honestly? You wouldn’t have picked this path if you were given a choice.

There’s a study from 1978 where people were organized in pairs to do puzzles. The first person in each pair chose which three puzzles they wanted and decided how many minutes to give each one.

The second person got handed three puzzles and told how long to spend, and it was the exact same three their partner had chosen, for the exact same minutes. So both of them sat there doing identical work, and the only thing separating them was that one had decided it and the other had been told.

Then the researcher announced the experiment was over and walked out, leaving the puzzles on the table with nobody watching.

You know where I’m going with this, right?

The people who had chosen went back to the puzzles in that empty room, and when they were asked whether they’d want to come and do this again another day, they said yes far more often than the partners who’d been handed the identical thing.

So when we automatically force “reactivated” (by that I mean actually just “relogged”) users into their old setup, we're kind of picking the puzzle for them and making the expectation gap as large as the Grand Canyon. And I’m not saying that this journey is fundamentally wrong. If the goal remains the same, there are true benefits to it.

The good news is: instead of assuming something by default, we can always ask.

HOW TO CLOSE IT

I love love love how Duolingo does it, so let’s just steal it. These 3 steps can be adjusted to your product of course, they are pretty simple and do help you truly reactivate users who are coming back after a long break.

1. Tell me you noticed.

Come back after a long absence and the first thing the owl does is say it knows who you are and that you’ve been gone. It costs nothing and it asks nothing, and it lands, because it’s like walking into a bar where everybody knows your name.

2. Ask what I’m here for today before you decide for me.

Then it asks. Keep going with German, take a different course, or try the brand new chess thing. Which matters, because who knows what actually brought me back? It could have been an ad about a new feature, an upcoming trip, or a major change in my life, and any one of those could point somewhere other than where I stopped.

3. Update the context.

Once I’ve said German it asks how much I still remember, why I’m learning (now), and then builds a starting point out of those answers rather than out of the one I abandoned. Instead of returning me to my old plan, it’s making a new one for the person I am today.

If you’re B2B and this feels very consumer, the translation is not complicated. Ask whether they’re rejoining an old team or starting a new one. Ask whether they’re picking up the old project or beginning something else. Show them what changed while they were away. Keep what they built but clarify the context.

None of this applies if they’ve only been gone a fortnight, and it can be modified if they told you why they came before logging in. My point is the same as always: our product plays a very small role in our users’ lives and exists within the context of specific tasks. Without knowing it, we can’t provide value as well as we’d like and meet their expectations, which only keep growing.

ONE THING TO REMEMBER

Life happens between when somebody left and when they came back, and the setup they left behind was built for a version of them that has already expired. Make them feel welcome and put them on the track that makes sense today.

What I do find fascinating is the way AI adoption, and even simply talks about AI, change our (and users’) behaviour. Because this is something I have to solve every day. I don’t know if AI will kill SaaS but it is already changing how people interact with SaaS and really want to explore that.

And wherever you look, there’s one problem that's gotten bigger than others.

I’d call it the expectations gap.

Because as heartbreaking as it is, users don’t evaluate our product by some absolute scale. They compare it against other options, former experience and finally, the impression they got from our promises, even if they were absolutely subjective. And the scaling AI adoption created a whole new bar for those expectations.

So I truly believe that the expectation gap, aka the difference between what we expect and what we get at different steps, is the key to figuring out how to create a great user experience and build a healthy business.

In the next issues, I want to dive into challenges that we see that were created by this gap (which is actually way more complex than it sounds) and see how we can close it.

THE CHALLENGE

You saw all of these AI complaints. How bad the output is, how sloppy it is, how stupid the automated outreach is. Maybe I’m wrong, but I don’t remember people being this publicly dissatisfied with any other technology.

I’m not bringing it up because I’m worried about AI’s feelings (although maybe I should be, dear overlords). I’m bringing it up because the quality of that output is a growth metric now. And it’s relative and subjective as hell.

If people think what came out is good enough, they think the product is worth carrying on with. If they don’t like it, they don’t like it, and that’s the end of it.

And guess what, AI products often have horrible retention. Every other question I get from AI products is related to churn. Activation looks great, because everyone is eager to try the magic button. But then they ghost.

And I have a hypothesis why.

THE GAP

Officially, the expectation gap comes from mismatched results. All the AI positioning makes us believe that it will magically solve our problem in the best way, but somehow it always seems worse than the real thing.

But…

I don’t think we’re judging the output at all. We’re judging something we had no part in making, and we are merciless about those.

I can bet you’ve given feedback to an AI that you would never say to a person’s face. You’d definitely never say it to yourself (even though the thing you’re tearing apart got built from your prompt, but it doesn’t automatically feel like yours, does it?).

There’s a study (way more than one) that shows how far this goes. Michael Norton, Daniel Mochon and Dan Ariely sat people down with written origami instructions and had them fold a frog or a crane, which mostly came out badly, because folding origami off an instruction sheet is genuinely hard. Then they asked what they’d pay to keep the thing they’d just made. Separately, they took a group who hadn’t folded anything, put the same amateur origami in front of them, and asked the same question. A third group was shown origami made by actual experts.

You know where I’m getting with it, right?

The builders bid 23 cents for their own crumpled paper. The people who hadn’t made it bid 5 cents for those exact same objects, so near enough worthless. The expert origami went for 27 cents, which puts the builders’ own bad frogs within four cents of professional work. And the builders assumed everyone else would agree with them.

So whenever we work with AI output, what we see is somebody else’s crumpled paper. It’s never good enough, because it happened without our effort.

HOW TO CLOSE IT

I found a couple of fixes that actually don’t mess with your output at all but work with other steps to adjust the framing.

1. Make me work.

Ask users for something but don’t be greedy. Just one real decision that changes what comes out the other end.

I noticed this in Lovable recently. Instead of building the magical app for you right away, it stops you midway and asks you to pick a theme. It generates a couple of designs and asks which style you prefer before it goes on. They're asking for additional input to make you part of the process, not a passive observer.

2. Flatter me.

I've been working with a team where I'm constantly pushing commits I wrote with Claude, and I noticed on GitHub that the commit is submitted by “Anastasia and Claude”. Which is absolutely ridiculous, because in my mind I did so little of the job that I don't even deserve to be mentioned.

But it made me feel so good…….

You push something big, and you go, oh my god, mom, look at me, I pushed some code. I am not a technical person, so it empowers me so much. It makes me feel like a collaborator instead of a passive client, and I value the output more, because I feel like I invested a lot into it.

ONE THING TO REMEMBER

Next time you hear output complaints, don’t rush rebuilding the whole system behind closed doors. Find a way to bring users into the process in a way that not only improves the input and the context, but also shows that they’re collaborators. Because the quality is rarely objective, and if people don’t value what you produce, it will never be good enough.

Anyway, what are your thoughts on this?

Leave a comment

I’ve lost count of how many times a founder has told me their product can’t have guided onboarding. Too many use cases, too many features inside each use case, everyone shows up for something different. So how would you even pick a path for them?

And I get it, it’s a real problem. But the conclusion people draw from it is wrong.

Because what usually happens next is they show everyone the same empty product, hope people figure it out on their own. Also the word “explore” comes to mind, reminding us more of hunting in a jungle than solving a small task.

On top of that, trying to stay on top of all the possible ways to activate users in the product, teams build these complex PQLs that at the end make it impossible to understand what’s going on (especially when the product is young and you really need to find that successful path).

In all of this wonderful complexity, trying to accommodate everyone’s interests and feelings, we forget one thing:

Within the first session, all our users need is just to get started, not to implement an annual strategy.

THE FIND

Have you heard of Claude? It’s this… I’m kidding. You know what Claude is. I bet you also know that it's about the most horizontal product on earth. Literally anything you can type into it is a use case.

I started working with a new team this week, was joining their Claude with a new account, and saw this during the onboarding:

Before they asked me about my role and my interests (which was about me, not about the product, so I liked it more). But this is the first question about the product, and look how small and effortless it is.

The word doing all the work here is "first." And the subheading calls them starter ideas, which makes the whole thing feel like nothing. I'm picking what to do in the next few minutes, I'm not deciding anything about myself.

And that one step actually does the heavy lifting in your activation. Let’s see why.

WHY IT WORKS

Dear ambitious product people, you probably heard it from your therapists (I did), and you’ll hear it from me: small goals are always better. Because they actually get people working.

In 1981 some researchers took a group of kids who were struggling with maths and gave them all the same pile of material and seven sessions to get through it. One group was told to do six pages per session. Another group was told to do all forty-two by the end. A third group got no target at all.

The ones going session by session did clearly better on the test at the end. The ones aiming at the whole pile did about the same as the kids with no target whatsoever.

An overly ambitious goal performed like no goal at all, which I find kind of funny given how much of onboarding is built on asking people about their goals, but as you can see, not all the goals are created equal.

But to hell with the first session and onboarding; it goes further than that.

When those kids were later left alone with a bunch of things they could do, the ones who’d been hitting small targets went and picked maths on their own. Nobody made them. Finishing something small and seeing it work made them want more of it.

For our product-focused minds, it means that just one question can affect not only our first session and activation, but long-term engagement as well.

Someone who finished one real thing has proof the product works for them, and that proof is what makes them open the tab tomorrow. Someone who got handed the whole product and told to go achieve their goals ends the session with nothing except a vague feeling that they… khm.. failed.

And nobody upgrades after they failed. Nobody even stays.

THE PLAYBOOK

When to use it:

Any product where the first screen could go in ten different directions and all of them are fine. Automation tools, AI products, analytics, design tools, anything where you open it and there's no obvious single next step.

How to do it:

1. Ask where they want to start, not what their overall goal is.

“What do you want to try first” and “what will you use this for” look like the same question and they really aren’t. One of them a stranger can answer in two seconds. The other one makes them think about what their boss said, their end-of-the-year KPIs, and all that stuff they wrote in the strategy deck.

You get a better input out of it too. A starting point is something people actually know, because this is why they signed up. Even if it is “I need to see what it can do before I take it to my boss”, that’s a clear next step too.

2. Every option has to be finishable in one sitting.

“Plan milestones for a creative project” ends somewhere. “Explore integrations” doesn’t end anywhere, it’s the forty-two pages again with friendlier wording.

Think about it like joining a gym. If the guy at the front desk asks what your fitness goals are, you make something up on the spot and then don’t come back. If he says come in Tuesday at seven, we’re doing legs, you show up on Tuesday.

Remember and don’t lie to yourself: any, ANY task can be broken down into smaller steps.

3. Always let them type their own answer, and then actually go read those, group them, and add popular ones to options.

Anyone who skips your options and writes their own thing came in with something specific already in their head. That’s free positioning research, and an alignment check; don’t waste it.

Advanced step: feed the answer forward past onboarding. What they picked should decide the empty states, the checklist, the first week of emails, not just the first screen.

When it backfires:

When you ask and then do nothing with the answer.

Someone tells you they want to build a report, and you drop them on the same empty dashboard everybody else gets. Now they’ve got a stated goal, eleven tabs and no idea which one gets them there. Asking and then abandoning is worse than never asking, because users feel tricked and ignored.

So sort out the handoff before you ship the question. Best case: you pre-start the task for them (open the editor with the template already loaded, the workflow scaffolded, the first field filled in). If you can’t build that yet, at least land them on the right tab with a short checklist that matches what they picked. And if you can’t do either, hold the question until you can.

If you forget everything, remember this:

You don’t always need to know why someone signed up. You need them to finish one small thing today, because that’s what makes them come back tomorrow. Ask what they want to start with, keep it small enough to finish, and then go start it for them.

🎉 Woow, you finished the issue, that’s awesome!

Hi, I’m Anastasia Kudrow, and I write Ghosted.

I also help SaaS teams use psychology and PLG to build profitable products that people love using.

Come find me on LinkedIn: https://www.linkedin.com/in/anastasia-kudrow/

Or check out my site, maybe we end up working together:

https://www.ghosted.consulting/

See you next week!

Have you ever heard that mobile converts worse than web? It was the case almost on every product I’ve touched. And everyone reaches for the same excuse: mobile users are flaky.

They signed up just because they saw an ad, couldn’t be bothered to finish, so we’ll email them later when they’re back at a laptop.

Yes and no. Honestly, seeing half of my friends having their whole lives in their phones, I don’t buy it, not most of the time.

What gets someone to set up and actually pay is the same on mobile as it is anywhere. The rules don’t change. Mobile just punishes you way harder when you break them.

A bloated setup on a laptop is annoying but people push through anyway because the environment is different.

With that same setup on a phone, everything’s working against you. They’re distracted, half-watching something else, doing this standing in a line somewhere.

They have less time to focus, and a mobile screen just takes less physical space in front of them; therefore, as ridiculous as it sounds, it can be perceived as less important.

THE FIND

I was setting up a new event type in Calendly on my phone this week. You know what Calendly is, right? It’s packed with settings, even on the free plan.

But when I want to create a new meeting, I just see this :

Three things my page can’t exist without, and a Create button. That’s the whole screen. Everything else sits under “More options,” one tap away, for whenever I actually want it.

To be honest, this isn’t a mobile-only trick (even though it could have been). Calendly does the exact same setup on web, I checked. I’d just never noticed it there.

On a big screen a clean setup is nice and completely forgettable. On my phone it jumped out, because I’ve been trained to expect the opposite: open a product, get buried in a wall of settings, half of which have nothing to do with me.

WHY IT WORKS

You’ve probably seen the jam study and know about choice overload.

But it’s not the whole story, and the bit that actually matters here is effort.

In 2010 some researchers let people freely pick between two tasks that were basically the same, except one made you switch back and forth more, so it took a little more mental effort. People kept going for the easier one.

And here’s the part that gets me: they couldn’t even tell you which task had been the harder one. We dodge mental effort on autopilot, before we’ve consciously decided anything.

That’s what a maxed-out first screen is up against. Nobody looks at it and thinks “eh, too much, I’m out.” They just feel the weight of it and drift to the back button, and if you asked them why, they’d struggle to tell you.

Now stick that person on a phone.

When we get pulled onto a new thing before we’ve finished the last one, part of our attention stays behind, and we’re slower and sloppier at the new thing for a while. That’s just the default state of someone on mobile. They opened your setup in the middle of something else, running on half a tank.

So the same screen that’s totally fine in a focused laptop session is landing on someone with way less to give it.

So, here it is. On mobile, you have so many forces literally working against your commitment that, of course, converting is harder. But it’s still doable.

THE PLAYBOOK

When to use it:

Any product where people have to set something up before it does anything useful. Scheduling, site builders, workflow tools, dashboards. It matters most when you’ve got real mobile traffic and an audience of small teams or solo people, the ones doing this between five other things on their phone. They’re not less patient than your desktop users but they’ve got more distractions.

How to do it:

1. Give them the smallest version that actually works, and let them launch from it.

Calendly gives you 3 must-have steps to publish a real, working booking page. It doesn’t mean people will publish it and keep right in that setup. But they can preview it, get a general feeling of the result, and come back to editing later.

This is how trust and motivation are built.

Think about it as trying a new dish for the first time. You have no idea if you’ll like it, but if you read just the main ingredients (not all of them, with cooking instructions), you can get a good enough idea if it will work for you.

Same goes for the first step: give me highlights that create the main palette, not every tiny toggle.

2. Hide the rest, don’t cut it.

Every advanced setting Calendly has is still there, one tap under “More options.” A power user goes and digs it out. A newbie never has to look at it. You haven’t taken anything away; you’ve just decided what people see first, so you’re not scaring off your serious users to save your new ones.

Advanced step: if you have different use cases that genuinely appreciate very different core features, customize the first step. Add a profiling question to teh onboarding and link it to the setup people see.

3. Keep the irrelevant, narrow, specific stuff off that first screen.

A setting that has nothing to do with someone’s use case can throw people off: “wait, is this even for me, or for some other kind of user?” And that’s a really expensive question to plant in someone’s head thirty seconds in.

When it backfires:

if your product genuinely does nothing until it’s heavily set up, and there’s no honest minimal version. In this case, don’t try to fake it. Make it scannable, let people play in a sandbox so they can see the potential result. Seriously, take a page out of Amplitude’s book, they are crushing it.

If you forget everything, remember this:

People aren’t there to learn your product, validate it, or be impressed. They also don’t care if you structured all of your features in a very cohesive way that makes sense to developers. They just need the stuff to get started with. Give them one small result to build trust, and they will want to go deeper naturally.

🎉 Woow, you finished the issue, that’s awesome!

Hi, I’m Anastasia Kudrow, and I write Ghosted.

I also help SaaS teams use psychology and PLG to build profitable products that people love using.

Come find me on LinkedIn: https://www.linkedin.com/in/anastasia-kudrow/

Or check out my site, maybe we end up working together:

https://www.ghosted.consulting/

See you next week!

Hey everyone!

We keep building our guide to referral programs, and this is the one I've been most excited to write.

Where we've been so far:

Why reward doesn’t matter that much

When to ask

Who to ask

Aaand, today we will talk about how to ask, without even asking.

Almost every referral program is built around what the referrer gets. Credits, cash, a discount, something for them and something for their friend. That's fine, it works, nobody's mad about free credits.

The problem is that the reward isn't what makes someone want to recommend you. Nobody wants to feel like they're running ads for your product in their group chat. What people will happily share is something they'd send anyway that happens to have your product inside it.

Some products are built to hand people exactly that. I'll be honest about who can pull it off and who can't by the end. Let's dive in.

submit a product you are curious about

THE FIND

I was clicking through Amplitude's template gallery this week. Amplitude is product analytics (you know it) that I ADORE, and their gallery is full of ready-made dashboards you can drop into your own account.

Here's what I always loved: many dashboards have a real person's name and face on them.

These are built by known product and growth people, and each one is credited to its creator. Amplitude obviously reached out to them, but look at what they offered instead of a referral link. A format that lets these people showcase their own expertise. The creator gets a polished artifact with their name on it that makes them look sharp to their audience. Amplitude gets a dashboard that only works once you're inside Amplitude.

Nobody here is running a "sign up with my code" campaign. They're publishing something that makes them look good, and the product is baked into it. It is a referral program, hidden inside the affiliate / advocate / endorsement / partnership / whatever you call it campaign.

Canva does the same thing at scale with its Canvassador program, and Miro has their famous Miroverse.

Does the thing your users would share make them look good, or does it just make you look good?

Reply to everything. Edit nothing.

Your inbox is full. Slack is piling up. Client messages need a response yesterday. Typing thoughtful replies to all of it takes hours you don't have.

Wispr Flow turns your voice into clean, professional text you can send the moment you stop talking. Speak like you would to a colleague — tangents and all — and get polished output. Emails, Slack, LinkedIn, WhatsApp, whatever's open.

89% of messages sent with zero edits. Used by teams at OpenAI, Vercel, and Clay. Works on Mac, Windows, and iPhone.

Try Wispr Flow free

PSYCHOLOGY BEHIND

People share what makes them look smart (like this issue, for example). That's the whole engine, and there's solid research under it.

Jonah Berger and Katherine Milkman went through around 7,000 New York Times articles over three months to see what landed on the "most emailed" list. One of the strongest predictors: content that made the sharer look good to send. Useful, surprising, impressive. We forward the things that reflect well on us, and skip the things that don't.

So the format matters more than the reward. Anyone who's done sales knows you open with value, not with a pitch about yourself. A credited dashboard is value first, something the other person actually wants in their hands. A "20% off with my code" is a pitch, and you're handing it to a friend to deliver for you.

The discount has a second problem. It only moves people who were already circling your product, the ones who are problem-aware, solution-aware, and weighing the buy. For them it's the final nudge that gets them over the line (cool). For a cold person who's never heard of you, it's noise, because there's no trust yet for a discount to land on (not so cool).

A referral exists to reach exactly those cold people, which makes it an onboarding problem more than a rewards problem. Someone who arrives through a shared artifact is a brand-new user who happened to come in through a friend, and you onboard them the way you onboard anyone: value first, offer later.

THE PLAYBOOK

When it works:

Your product makes something a user would be proud to put their name on. A design, a dashboard, a template, a teardown, a report. Ideally, it’s not only shareable, but reusable. If that's you, this is the strongest referral loop you'll ever build, and most teams sit on it without noticing.

How to use it:

1. Put the creator's name on it.

Duh, right? But let’s not ignore it. Attribution is the whole difference between "here's a template" and "here's this person's template."

2. Make sure the artifact means something in the sharer's world.

For example, a Miro template is valuable for Product / Project managers and useless for a lot of people who don’t work in digital. Same targeting logic as the earlier issues: you want people whose audience actually cares about this thing. I know it sounds boring, but you do have to know your ICP.

3. Build the signup into using it, not next to it.

Cloning the dashboard puts you in Amplitude. Opening the design puts you in Canva. If your creator shares a PDF with a "sign up here" button at the bottom, you've split the value from the product and dropped the person back into salesman mode.

When it backfires:

Be honest about your product.

If it makes no shareable artifact but people use it together, run the loop through team invites instead, where the payoff is getting your actual team in, so the thing finally works.

If it's not collaborative either, you're down to external branding: "powered by," a badge, some swag. That's not the most high-performing tier, unless your UI is one of the strongest competitive advantages (like in fashion) or you have truly massive regular exposure to a new audience.

Not every product is built to go viral, and it's better to know which one you are than to run a creator program nobody would sign their name to.

If you forget everything, remember this:

People share what makes them look good, not what makes you look good. Build your referral so that using your product is the thing that makes them look good, put their name on it, and you stop asking for referrals and start collecting them.

🎉 Woow, you finished the issue, that’s awesome!

Hi, I’m Anastasia Kudrow, and I write Ghosted.

I also help SaaS teams get more upgrades by using psychology instead of cheap tricks. Because hype aside, people will be buying your product for many more years, AI or not AI.

Part 1 said the reward matters less than you think.

Part 2 said timing the ask beats sweetening it.

This is the one people skip and then wonder why nothing converts: you can only get an invite out of someone who actually has a person to invite. Duh.

Sounds obvious. Then you watch how these prompts usually get built, and it's "invite your whole team" fired at everyone who signs up, including the solo freelancer who's never had a team and never will. No reward and no timing trick fixes an empty room. There's just nobody on the other end of the ask.

Btw, don’t ignore this 👇

submit a product you are curious about

THE FIND

I was in Canva's team settings this week. Big product, most companies already have a login floating around somewhere.

What got my attention wasn't the invite prompt itself. It was who Canva decided to show it to, and why.

Canva looked at who I'd been sharing designs with and used that as the tell. I've been sending work to specific people, so it figures, correctly, that I've got people worth inviting, and it puts them right in front of me. If I'd never shared a thing with anyone, there'd be nobody to suggest, and the whole prompt would mean something different.

So, for your own product: when you push someone to invite their team, what tells you they've got one?

PSYCHOLOGY BEHIND

A prompt can't create willingness. It can only collect on willingness that's already there. And the clearest sign someone's willing to invite a coworker is that they're already behaving like they have one.

In 1966, Freedman and Fraser ran a study that's now in every persuasion textbook. They asked homeowners to put a big, ugly "Drive Carefully" billboard on their front lawn. Asked cold, fewer than one in five agreed. But when the researchers first asked for something small, sticking a little safe-driving sign in the window, then came back later with the billboard, agreement jumped past half. Same big request. The difference was one small step taken beforehand.

Sharing a design is that small step. Someone who's been sending work to a coworker has already pulled that person into how they use Canva. Asking them to make it official is just the next move for someone already headed that way. The user who's shared nothing is the homeowner asked cold, and no prompt talks them into a team they don't have.

The invite is how a free user turns into a paid team. Aim it at the people whose behavior already points to one, and the same prompt starts turning into seats.

THE PLAYBOOK

When to keep it in mind:

Any product that can see collaborative behavior. If you can track people sharing files, receiving comments, being added to something, you can read who's working with others and prompt off it. Whether someone has a team of two or two hundred doesn't matter, the behavior is what tells you they've got someone. In this case, this loop is quite often about the revenue expansion and engagement, not about acquiring new users (unless people can interact without signing up).

How to use it:

1. Find the behavior that points at someone not yet on the account.

Canva reads sharing. Find the action that means a user is working with someone outside their current team, sharing a design, a comment, a co-edit, then prompt off that instead of signup date. The target doesn't have to be new to the product. A free user or someone on another team is exactly who you want, because the win is expansion onto a paid team, not net-new signup necessarily.

2. Name the person, and say why.

The signal tells you who: the person on the other end of that share. Show them, and give the reason, "you already share designs with this person." Back in Part 2 I called every invite a reputation bet. That line shrinks the bet, because they've watched you use it already.

3. Match the push to the strength of the signal.

A recent, repeated share is a confident signal, lead with that name. A single old one is weak, ease off. Prompting a shaky signal as hard as a sure one is how you land on "invite your ex-manager" and lose trust you won't get back.

When it backfires:

Sensitive domains. If your product touches legal, health, HR, anything where who someone works with is confidential, surfacing collaborators exposes relationships the user never meant to show. Skip the auto-suggestions there and let people type.

If you forget everything, remember this:

An invite prompt can't manufacture someone to invite. It only pays off with users who already have one, and you find them by watching what they do in your product, and then acquiring more people like them at the top of the funnel.

🎉 Woow, you finished the issue, that’s awesome!

Hi, I’m Anastasia Kudrow, and I write Ghosted.

I also help SaaS teams get more upgrades by using psychology instead of cheap tricks. Because hype aside, people will be buying your product for many more years, AI or not AI.

Last week I argued that wanting has to come first, that no reward makes someone recommend a product they don't already love. A few of you replied, and one of you asked me to break down Magic Patterns' entire upgrade motion, which I'll do properly over the summer because there's a lot in there.

Today I'm staying on referrals and pulling out one small thing they get right: how you ask the people who already love you. As always, comment on Beehiiv, mention @Anastasia Kudrow for discussions on LinkedIn, or just hit reply.

Btw, don’t ignore this 👇

submit a product you are curious about

THE FIND

Magic Patterns is an AI design tool. You describe a screen or a flow, it generates real, working UI you can hand straight to engineers, and every generation spends credits. Go deep on a project and you burn through them fast (I’ve been there).

I use it constantly with clients. I'm not a designer, but I know what good looks like, and this lets me show it, an interactive version of the thing in my head, before it ever reaches a real design or dev review. Yesterday I was mid-task, burning through credits, getting close to my limit. Then I saw this.

It didn’t hit at the moment when my credits went dry; that was a place for a classic upgrade trigger. But this one showed up right when I was experiencing value, without being threatened by a potential usage block.

Most tools bury the referral link in a settings page nobody opens. Magic Patterns puts it on top of the editor, at the exact moment I need more credits, and the reward is credits, the thing I'm burning through, not a ten-dollar gift card I'd forget by lunch.

Then they fire the same offer again, the second I want to share a project with a teammate (again, a clear value indicator).

PSYCHOLOGY BEHIND

Wanting to do something and actually doing it are further apart than they look. The gap usually gets closed by something small: a prompt that shows up at the right moment and makes acting easy right now.

Psychologists call that a channel factor. The clearest study is Leventhal, Singer, and Jones, back in 1965. They handed Yale students a frightening pamphlet about tetanus. The students came away more scared and said they meant to get vaccinated, but almost none did, around 3%. A second group got the same pamphlet plus a campus map with the clinic circled and its hours. Vaccinations jumped past 30%, roughly a tenfold difference. Same fear, same stated intent, but the map did the work.

But the map only helps once the wanting is already there, so the real question is when that wanting shows up. A team invite is still a social bet: you're telling coworkers this tool is worth their time, and your name is on it. How early you'll make that bet depends on the risk.

When the product is a familiar category, or something the team has no real choice but to adopt, the risk is low and people invite from the first screen.

Push them toward yet another unproven tool and you're the one who dragged everyone in, which is exactly the social suicide nobody commits on faith. They want a happy moment first, proof it works, before they'll put their name on it.

THE PLAYBOOK

When to keep it in mind:

You can point to a moment where a user clearly succeeds: an output they came for, a result they can see, or a stretch of heavy, invested use. Workspace and credit- or seat-based products get the most from this, because the person you invite lands inside your account and becomes another seat.

How to use it:

1. Trigger on a result.

The strongest moment to ask is right after someone gets the thing they came for, a finished prototype, an exported file, the first report that actually works. They have proof in hand, and they're willing to put their name next to it. Ask here and the referral brings in someone with the same problem, the highest-fit user you'll ever get. It can depend on how complex your motion is and if one person can achieve any result without the team, of course. Which brings me to the second point.

2. Trigger on deep engagement.

Not every product has a single finish line. Some just get used, hard. The person on their fifteenth iteration, or the one burning through credits, is telling you they're invested, and that's its own signal of willingness. Fire there, and let the reward pay off the moment: Magic Patterns offers credits right as you're running out, so the ask reads as help rather than a pitch.

3. Stay inside the flow.

A success moment is fragile. Interrupt it with a modal that stops the user cold and you've turned a good feeling into friction. Keep the trigger next to the work, a small banner over the editor, an inline nudge on the screen they're already on, something they can act on or wave off without losing their place.

When it backfires:

Don’t reach out to those who experience problems with your product. To them, it will seem borderline offensive even if the offer is good.

If you forget everything, remember this:

We don’t need an additional trigger to share negative feedback. But we do need that one tap to share something that we love because otherwise, we just take it for granted.

🎉 Woow, you finished the issue, that’s awesome!

Hi, I’m Anastasia Kudrow, and I write Ghosted.

I also help SaaS teams get more upgrades by using psychology instead of cheap tricks. Because hype aside, people will be buying your product for many more years, AI or not AI.

I want to spend the next few issues on referral programs, because I don't know a single team that hasn't struggled with one. They're a pain to build, the math gets risky fast, and after all the cash you hand out, referrals still tend to land as some embarrassingly small line on your revenue.

And yet everyone keeps building them, because when they actually work, it’s magic.

So I want to take them apart properly.

If there's a referral question that's been bugging you, drop it in the comments or just hit reply, and I'll work the good ones into the series.

Today, let’s start with the question underneath all the others: why won't people share?

The incentive isn't the engine

Most referral programs are built on a weird assumption: that people will harass their friends and colleagues, and look like a sleazy salesman doing it, all for ten dollars. They won't. The money was never going to be the reason, so when nobody shares, making the reward bigger almost never fixes it.

Think about what you're actually asking for.

Go to someone whose opinion of you matters and vouch for a product. That's a bet on your own reputation. When researchers look at why people recommend things at all, money sits near the bottom of the list.

We do it to help someone we like, and to look good doing it (!), to be the one who knew about the good thing first, yes, it’s all about ego, everyone has one. The trigger is internal, and a reward can't manufacture it.

I see this from the other side all the time. I don't pay anyone to recommend Ghosted, so every time someone does it anyway, with nothing in it for them, it makes my whole day. But of course, I’ll come, like, and comment on it haha

Wait… are incentives useless?

Not really, even though we all could use a hot take for LinkedIn that something is dead again, right?

Seriously though, a referral program can absolutely have incentives, and a good one helps. You just have to be honest about the job it's doing.

But here’s the thing: the incentive doesn't make someone want to recommend you.

It lowers the friction once they already do. It's the nudge that gets a willing person to act today instead of next week, the little push to finally send the link they were already half-thinking about. It gives them a reason to do it now, and a reason to do it again. That's a real job, and it's worth designing well. It's just not the heavy lifting.

The wanting comes first, and the incentive rides on top of it. When the wanting isn't there, the incentive has nothing to work with, and it sits unused no matter how big you make it.

So the real question was never how much to pay (even though a “too small” cash incentive can absolutely hurt your flow). The real question is whether you've created the conditions where someone wants to share at all.

Okay, how to create those?

A referral only happens when a few things line up, and when they don't, no reward buys your way out.

1.The person has to actually love the product.

Not just "activated," but uses it and would be annoyed if it disappeared tomorrow. You can't vouch for something you've barely touched, and people won't. This is why emailing your whole user base a referral offer does so little. Most of them have nothing to vouch for yet, and a bigger reward won't hand them conviction they don't have.

2.They also have to know someone with the same problem.

Teams skip this one completely. A recommendation only makes sense between two people who share a need, the same job, the same headache, the same kind of work. If your user is a freelance designer, their referral is worth something to another freelance designer and nothing to their cousin who sells insurance. So your program leans on whether your best users run in circles full of more potential best users. When referrals stay flat, it's often not the reward, but the problem of who you are targeting (yes, at the very top of the funnel). Do they even have anyone to share it with?

3.And the ask has to make them look good, not you.

The second your program makes someone feel like your unpaid sales rep, you lose the exact people you wanted. Dropbox got this right, and we all can’t stop talking about it. Their referral line wasn't "earn by sharing," it was "get more space." Same action, but one makes the user your salesman and the other makes them someone getting more value from the product they already love and sharing the same with a friend.

But even when those are in place, you can fail, if…

… you simply forget to inform me that this opportunity exists. Sorry for this official tone, but I really wanted to highlight how serious this dumb problem is.

We ran a regular research with one of my clients to figure out why our referral program didn’t perform as well as we’d like. We bet that the offer wasn’t good enough. Turned out, 60%+ of users never even heard about it.

Placement does more work here than the size of the reward, especially if you have a high-volume product.

A referral link tucked in a settings menu gets close to zero use; the same link on the main screen gets 30%+. One Shopify brand tested it and watched participation climb from 0.8% with a footer link, to 4.2% with a dashboard embed, to 11.7% with an in-product prompt that fired at the right moment. That's a 14x swing from placement and timing alone, before you touch the reward.

So, two things:

Where it lives: put it in the path people already walk, the dashboard, the onboarding flow, the screen they land on after they get a result. Don’t make it annoying but make it present.

And when it fires: ask right after a win, not at signup before they've felt anything. The moment someone finishes a project, sends the first invoice, hits the result they came for, that's when they feel good about you and will actually pass it on.

3 things to make people want to share your product

1. Pick the segment before the reward.

Start with people who love the product and know others with the same problem. If those two things aren't true, no incentive saves you, because you're paying people to do something they have no reason to do.

2. Make it visible at the moment of value.

Put the ask in the path they already walk, and trigger it right after a win. A referral link nobody sees is the most common reason a program looks dead, and no reward fixes a link nobody finds.

3. Treat the reward as a nudge, not the main point.

It smooths a trigger that's already there, but a great experience and product adoption must come first. So when referrals are flat, don't reach for a bigger number. Go back and check the conditions and the placement first.

If you forget everything, remember this:

You can't pay someone enough to feel good about selling out their friends. If users don’t share your product, look into segmentation, exposure, and overall client satisfaction. If those components don’t work, a bigger offer won't change it.

🎉 Woow, you finished the issue, that’s awesome!

Hi, I’m Anastasia Kudrow, and I write Ghosted.

I also help SaaS teams get more upgrades by using psychology instead of cheap tricks. Because hype aside, people will be buying your product for many more years, AI or not AI.

I was on a call with a founder the other day. Users were struggling with activation, free-to-paid conversion was low, and we were going through what to do about it.

Somewhere in the middle of the conversation, they stopped me and said: "So basically, you're telling us we built a really smart product and now you want us to dumb it down."

I said yes.

And I've been thinking about that ever since, because they sounded offended. Like I was asking them to undo the best thing about what they'd built. But the more I sit with it, the more I think this is one of the most common traps in product: the team builds something genuinely sophisticated, genuinely powerful, and then can't figure out why users bounce off it like it's a brick wall.

So let me say it plainly: dumb products win. They win on activation. They win on conversion. They win on retention. Users are human, and humans gravitate toward things that feel easy because they make them feel smart.

No one wants to feel stupid

Your product doesn't exist in a vacuum. You know this intellectually, but it's worth saying out loud because it changes how you should think about everything you build.

The value of your product isn't what's on your feature list. It's what your users perceive. And perception is an emotional process first.

When someone opens your product and immediately gets what it does, finds the thing they need, and gets a result, they feel smart. That feeling attaches to your product. "This tool is great" is really "this tool made me feel competent," and competent feels good. There's a psych concept for this called processing fluency. When information is easy to process, we rate it as more trustworthy, more valuable, more appealing. We don't evaluate the information more carefully and then decide it's good. We feel the ease, and the ease becomes the evaluation.

But someone opens your product, sees seventeen options, isn't sure where to start, clicks something that opens a settings panel they weren't looking for, and now they're two layers deep in a UI they don't understand. They feel lost. That feeling also attaches to your product. And once it does, your feature list is irrelevant. I don't care that you have AI-powered workflows and 200 integrations. I feel dumb right now, and I want to close this tab.

This is the thing founders miss. The user is evaluating how the product makes them feel in the first minutes. And if that feeling is confusion, it doesn't matter what comes after.

What "dumb" actually means

I'm not talking about building something simple or stripped down or easy to clone. You can be incredibly sophisticated under the hood. The dumb part is the surface.

Three things make a product feel dumb in the right way:

The value is undeniable and easy to describe.

If your user can't explain what your product does in one sentence to a friend, you have a clarity problem that no feature can fix. "It automates my workflows" is fine. "It's an AI-powered cross-functional operational efficiency platform" is a product that's about to get ghosted.

The next step is painfully obvious.

At every point in the experience, I should know what to do next without thinking about it. Not "figure out which of these three paths applies to me." Not "read the tooltip to understand what this button does." The action should be so obvious it feels almost insulting. That's the target.

The interaction doesn't feel like homework.

If using your product for the first time feels like working through a thirty-page workbook, you've lost. It doesn't matter that the workbook is well-designed. Cognitive load is cognitive load. The brain doesn't distinguish between "this is hard because it's confusing" and "this is hard because there's a lot of it." Both feel like effort, and effort triggers avoidance.

Why we ignore the clever product

Here's where it gets interesting for anyone doing PLG.

You'd think that if you offered someone a full-featured product and a stripped-down demo, they'd pick the full version every time. More is more, right?

They don't. And it's consistent.

When you give someone access to everything, every feature, every setting, every workflow, you're also handing them a decision problem. What do I use first? What matters? What can I ignore? Every option is a tiny cognitive tax, and those taxes add up fast. The psych term is choice overload, and it doesn't just reduce satisfaction. It reduces action. People stare at the all-you-can-eat buffet and walk away hungry.

The limited version, the demo, the guided free tier, the "dumb" version with three features, removes all of that. Fewer decisions to make. The path is clear. And because the path is clear, the user actually walks it, gets a result, and now they have a real reason to upgrade based on something they experienced.

Self-efficacy matters here too. People don't just need to believe your product is valuable. They need to believe they personally can get value from it. A complex product with a steep learning curve quietly signals: you might not be able to pull this off. A dumb product says you got this. Look, you already did the thing.

That confidence is what drives upgrades. The quiet, internal belief that "I can use this and get what I need."

The sophistication trap

The founder I talked to wasn't wrong about their product being smart. It probably is. Most products that struggle with conversion aren't bad. They just make perfect sense to the people who built them and almost no sense to anyone else.

You built it, so the interface makes perfect sense to you. You know why that dropdown exists. You know what "configure workspace" means. You know that the three-step setup is actually fast once you understand the mental model. But your user doesn't have your mental model. They have forty seconds of attention and a vague sense of what they came here to do.

Building the dumb version means closing the gap between what you know about your product and what a first-time user can absorb in a single sitting. The sophistication can live underneath. Nobody's asking you to delete features. Just stop making every feature visible at once.

If you forget everything, remember this:

Dumb products make their users feel smart. Clever products make their tech team feel smart. Who are you building for?

🎉 Woow, you finished the issue, that’s awesome!

Hi, I’m Anastasia Kudrow, and I write Ghosted.

I also help SaaS teams get more upgrades by using psychology instead of cheap tricks. Because hype aside, people will be buying your product for many more years, AI or not AI.

But people are finally starting to talk about what comes after: the habit.

I think it's about time, because activation doesn't always bring you upgrades. Habit and long-term engagement do. And I see this constantly with my clients. This week was a perfect example: a client with a very strong top of funnel. Dedicated, curious audience that was actively looking for guidance. Barely anyone came back after the first session. Barely anyone upgraded without a demo.

The early value wasn't great either, but that wasn't the main problem. The main problem was habit building. Introducing the core features. Getting people to actually adopt them. This is one of the biggest challenges I see, especially with more complex B2B tools.

How do you form a habit that drives long-term revenue? And why are the rules so different from how we work with activation?

Want to read issues that solve your current challenges? 👇

submit a product you are curious about

The Find

I heavily use Magic Patterns. I honestly think it's the most convenient product for collaborative design and prototyping, especially for consultants who work across multiple client projects.

The other day when I opened the product, I saw this:

A suggestion to create a design system, with free credits gifted to me to accomplish this task.

Here's why I think this is genius for habit formation.

1. It specifically prompts an activated user to do a task that will signify a bigger commitment, improve the overall quality of their results, and create a sense of ownership over what they've built.

2. On top of that, it reduces the barrier. Creating a design system sounds like homework, but Magic Patterns invested in me so I could do it. Free credits, on them. Just go and do it, you will benefit from it, and the product will benefit from it too.

Now, I can see how this contradicts the activation logic that many people use. When we think about activation, we're trying to cut as much as we can. Reduce the number of steps, postpone the most complicated tasks. But big results don't come from small moves. You have to invest to actually feel dedicated. And when it comes to habit, it does make sense to onboard users into heavier use cases that contribute to long-term engagement and the overall experience from the product.

Why it works: psychology behind

Think about the last product you kept coming back to. The one where, at some point, you built something inside it. You uploaded your data. You configured your workspace. You set up templates that saved you time every week. You committed. 

Now think about whether you'd switch to a competitor. Probably not. And the reason isn't that the competitor is worse. It's that you already put work in. You'd have to redo all of it. That feeling of "I've already invested too much to leave" is doing more for your retention than any activation flow ever could.

Norton, Mochon, and Ariely studied this in 2012. They had people assemble IKEA boxes, fold origami, and build Lego sets. Participants consistently overvalued their own creations, rating their wobbly origami cranes nearly as high as expert-made ones. The researchers called it the IKEA effect: when you put labor into something, you value it more. You probably know that.

But there's a critical boundary condition. 

The effect only kicked in when participants successfully completed the task. When they built something and then had to destroy it, or couldn't finish, the increased valuation disappeared entirely.

This is exactly what Magic Patterns is engineering. 

A design system is meaningful labor. You're importing your brand, your components, your typography. When you finish, you've built something inside the product that's yours. You value it. You don't want to walk away from it.

Nir Eyal calls this the investment phase, the most overlooked step in his Hook Model. Users put something into the product that stores value for future use. Data, content, configurations. The more you invest, the more the product appreciates for you specifically, and the harder it becomes to switch. Your design system now lives in Magic Patterns. Every future generation matches your brand. Leaving would mean rebuilding all of that somewhere else.

The free credits solve the boundary condition problem. Creating a design system sounds like homework. Credits reframe it from "work you should probably do" to "a gift you'd be silly to waste." 

The Playbook

When to keep it in mind:

This move is for users who've already gotten some value from your product. They've used it, they've seen what it can do. If your setup and even activation look reasonable, but there are barely any self-serve upgrades after the trial, this is for you.

How to use it:

1. Find the investment action that raises switching costs.

Look for the task that, once completed, makes your product personalized in a way that's painful to replicate elsewhere. For Magic Patterns, it's the design system. For a project management tool, it could be importing your team's workflow templates. For an analytics platform, setting up custom dashboards / integrating with Slack or Notion. The key is that the action should produce something the user will reference repeatedly.

2. Subsidize the effort.

This is where most products fail. They know the deeper feature exists. They even nudge users toward it. But they treat it like any other feature adoption message: "hey, have you tried this?" Magic Patterns gave me free credits. They put skin in the game. Think about what your version of that looks like. Extended trial access to the feature, a free migration service, a setup wizard that cuts the manual work in half. The goal is to get users past the finish line, because the IKEA effect research is very clear on this: incomplete investment creates zero additional attachment. People need to finish the task for the magic to work.

3.Time the prompt to a moment of momentum.

Magic Patterns showed me this when I opened the product, when I was already in work mode, already engaged, already likely to say yes. Don't bury the investment prompt in an onboarding email three days after signup when the user has already mentally moved on. Show it when they're in the product, doing things, riding the energy of something they just accomplished.

When it backfires:

If your users haven't hit initial value yet, asking them to invest deeper feels like you're piling homework on top of homework. You're essentially asking someone to furnish a house they haven't decided to rent. I see this a lot with B2B tools that rush users into complex setup flows before anyone has seen a single result. If activation metrics are still weak, fix that first. Pushing investment too early accelerates churn.

Better move here: focus on a time-to-value sprint. Strip the first session down to one clear result. Then introduce the investment play on return visit two or three, when they've already decided the product does something useful for them.

If you forget everything, remember this:

Activation gets users in the door. But it's investment that makes them stay. The products with the strongest long-term retention are the ones that get you to build something inside them you don't want to leave behind. And the smartest ones make that building feel like a gift, not a chore.

🎉 Woow, you finished the issue, that’s awesome!

Hi, I’m Anastasia Kudrow, and I write Ghosted.

I also help SaaS teams get more upgrades by using psychology instead of cheap tricks. Because hype aside, people will be buying your product for many more years, AI or not AI.

Hey everyone! If you have thoughts on Ghosted, feedback, requests, you can comment right on Beehiiv or hit reply. I read all of it.

This week I spent way too long on a pricing page, and the product caught me.

Btw, don’t ignore this 👇

submit a product you are curious about

The Find

Riverside — recording platform for podcasters and video creators. Their pricing page has four plans, monthly and yearly billing, and feature lists that go on forever. It's a lot to take in.

I was looking through the plans and about five seconds in, a popup showed up 👇

So I took it. Eight questions about what I'm trying to do, how often I record, whether I need a team workspace, that kind of thing. Most of them were just two options 👇

Everything is coming into focus.

Join beehiiv live on July 16th at 1PM ET for a first look at the future of audience-led business.

This isn’t just another feature launch (though there will be plenty of those). It’s a look at a more connected future for creators and brands that are tired of juggling disconnected tools, platforms, and data.

If you care about building an audience online, this is worth your time.

Reserve your spot.

And at the end: one recommendation. "The Business plan is a great fit for you."

For the record: I think the quiz is far from perfect; you could tell it even by looking at the last screen. But the idea of going through pricing as a conversation felt completely different from squinting at four columns trying to figure out which features I actually care about.

Riverside's pricing is genuinely complex because their customers are genuinely different. Solo podcasters, video production teams, enterprise webinar operations. They can't collapse that into two clean tiers. So they built something that catches you when you're stuck and walks you through the decision.

Does your pricing page do anything when someone freezes?

Why it works: psychology behind

There’s that famous jam study in 2000. Tasting booth at a grocery store in Menlo Park. One Saturday, 24 varieties were on display. Another, just 6. More people stopped at the big display (60% vs 40%), but way more actually bought from the small one. 30% purchased from the 6-jam table, only 3% from the 24-jam one.

The obvious read is "fewer options = more sales." That advice falls apart when your product genuinely serves different audiences and you need those tiers. Riverside can't cut their plans down to two because a solo podcaster and an enterprise webinar team need completely different things.

What the quiz does is break one overwhelming comparison into a sequence of small, easy questions. Each answer eliminates paths you don't need. By question eight, you've already made the decision in pieces without realizing you were doing it. The complexity is still there, you just never had to hold all of it in your head at once.

Sometimes you can't escape a complex screen. What you actually can do is help people walk through it.

The Playbook

When to keep it in mind:

Don't count your plans. Look at your data. If people spend real time on your pricing page and then bounce without converting, that's them telling you they wanted to buy but couldn't navigate the decision. If your support team keeps getting "which plan should I pick?" tickets, same thing. If pricing conversion is healthy, a quiz just adds steps nobody asked for.

How to use it:

1. Show it when someone's stuck.

Riverside triggers the popup after about five seconds of scrolling. If someone lands and picks a plan right away, they don't need a quiz. It's for the people who keep scrolling, keep comparing, keep not clicking. Time-on-page and scroll depth are the simplest triggers. Cursor drifting toward the back / logo / buttons or top right corner works too.

2. Ask about their situation, not your features.

Riverside's early questions get this right: what are you creating, how often, team or solo. Those are questions about me. But by question seven, it asks me to choose from a list of advanced features like CRM integrations and brand governance. That's the pricing table repackaged as multiple choice. Keep every question in the user's world and map their answers to your tiers on the backend.

3.Land on one recommendation with a clear CTA and maybe even some social proof.

The result says "The Business plan is a great fit for you" and then a ton of text. Don’t do this. Highlight the features I said I cared about, add “upgrade” and “book a demo” CTAs, and throw in some review from a person who represents my use case.

When it backfires:

If your tiers are obviously segmented by role or company size, the segmentation already does the guiding. Adding a quiz creates friction where there's no confusion. Also watch quiz length. Riverside runs eight questions, and honestly, that's a lot for something labeled "1-minute." If users bail halfway through, you've just created a new drop-off point.

If you forget everything, remember this:

Self-serve still needs hand-holding sometimes. When a screen is genuinely complex, the worst thing you can do is leave people alone with it because you believe that demos and additional steps are horrible.

🎉 Woow, you finished the issue, that’s awesome!

Hi, I’m Anastasia Kudrow, and I write Ghosted.

I also help SaaS teams get more upgrades by using psychology instead of cheap tricks. Because hype aside, people will be buying your product for many more years, AI or not AI.

Hey everyone! This week I have a story about urgency, specifically, about how most products manufacture it badly, and what it looks like when it shows up on its own. I also panicked a little. You'll see.

Btw, don’t ignore this 👇

submit a product you are curious about

The Find

I use Inkless for my contracts. E-signing tool, free plan, does the job.

This week I was prepping a partnership agreement for a new client, and my head was clearly somewhere else, because I made a typo in my own email domain. The contract sends to me and the client for signatures. I only caught it when my copy didn't show up. Checked spam. Nothing. Went back to the document, looked at the email, and there it was. Wrong domain.

So now an actual legal document is sitting in some stranger's inbox. I was, not gonna lie, a little panicking.

Then I saw the tiny pencil icon next to the recipient. Omg, I can edit!!

I hit it immediately.

Upgrade wall.

Look, this upgrade page is not good. All three plans stacked, no context about what I actually need, completely generic. We've talked about this in the newsletter, if you do a feature-locked upgrade, it should be about that feature, not your entire pricing table.

But I barely noticed. Because I wasn't browsing features. I was mid-panic trying to fix a mistake. The trigger did more heavy lifting than any part of that pricing page.

Stop babysitting dashboards. Ship from Slack. Touch grass.

700+ teams have Viktor reading their Google Ads every morning.

Your media team opens Slack at 8am. There's a cross-platform brief in #growth: Google Ads spend vs. ROAS, Meta CPA by campaign, Stripe revenue by channel. Viktor posted it at 6am. Nobody asked for it.

Last week, one team's Viktor caught a spend spike at 2am on a broad match campaign and flagged it in Slack: "CPA up 340%. Recommend pausing and shifting budget to the top two performers." That would have burned $3K by morning. The media buyer woke up to a problem already handled.

Your strategist reviews spend trends. Your account manager checks revenue attribution. Same Slack channel, same colleague, before anyone's first coffee.

Google Ads, Meta, Stripe. One message. No Looker, no Data Studio. Anomaly detection runs around the clock. Cross-platform reporting runs on autopilot.

5,700+ teams. SOC 2 certified. Your data never trains models.

"Viktor is now an integral team member, and after weeks of use we still feel we haven't uncovered the full potential." — Patrick O'Doherty, Director, Yarra Web

Start free. $100 in credits →

Why it works: psychology behind

George Loewenstein published a paper in 1996 called "Out of Control" about what happens to decision-making when people are in what he called visceral states: fear, pain, urgency, hunger. The kind of states where your body is involved, not just your brain.

His finding: people in these hot states make fundamentally different choices than when they're calm. And, this is the part that matters for your product, people in calm states can't predict how they'll behave in hot ones. The gap between the two is large and consistent.

Every product team I've seen designs upgrade triggers in a calm state. Meeting room, Figma file, "let's A/B test the CTA." Meanwhile, the user on the other end might be mid-panic. And in that state, your carefully worded pricing page barely registers.

Now think about how most products manufacture urgency. Countdown timers. "Limited offer." "Only 3 spots left." Your users know it's fake — they've seen the same timer reset three times. No visceral state fires because there's nothing real behind it.

A typo in a live contract? That's real. Your user's body is involved. A mediocre upgrade page almost got me because nothing about that urgency was manufactured.

The Playbook

When to use it:

Any product where users hit moments of real stakes under time pressure they didn't choose. You're on a free plan of your tax software, and it's April 14th, the deadline is the government's, not the product's, but the upgrade trigger benefits all the same. You're screen-sharing a design to an important client, and there's a watermark on it or a time limit. Or, well, you sent a contract to the wrong email, and you need to fix it. It works when something real is on the line, and your users actually have an urgent problem to solve.

How to use it:

Three natural urgency moments to map:

1. "I broke something." Users don't comparison-shop when they're trying to undo an error, they buy the fix. Find where your users make mistakes and put an upgrade path right there.

2. "I'm about to lose something I built." Nobody made a mistake here, but the loss is concrete. Kahneman and Tversky showed that losses hurt roughly twice as hard as gains feel good. Speak to that.

3. "Someone is waiting and I can't deliver." This is an external factor, and you won’t know if it hits without testing the message, but I personally upgraded products multiple times, driven by this reason. A client just said yes and needs the proposal today. A prospect responded, and the conversation is hot. The user needs to move now, and the free plan is too slow or too limited. The urgency comes from someone else's shrinking attention window.

When it backfires:

Now, the part nobody plans for. These triggers convert well. But the user who upgrades mid-panic (reasons 1 and 3) didn't fall in love with your product. They bought an emergency fix.

Panic-upgraded users need their own onboarding. Show them what else they just unlocked. Get them using two or three paid features in the first week, the ones they didn't come for. Otherwise you turned urgency into a one-time payment and churn.

If you forget everything, remember this:

Stop building urgency with countdown timers that your users have learned to ignore. Map where urgency already lives in your product and offer relief.

🎉 Woow, you finished the issue, that’s awesome!

Hi, I’m Anastasia Kudrow, and I write Ghosted.

I also help SaaS teams get more upgrades by using psychology instead of cheap tricks. Because hype aside, people will be buying your product for many more years, AI or not AI.

About once a month, I'll sit down with a founder or growth expert and ask them three things: what frustrates them the most about user behavior, what mistakes teams keep making trying to fix it, and which products actually did a great job and converted them personally.

Pretty cool, huh? I know you don’t care about the textbook. You want to know what actually happens in those bloody growth trenches.

And the first issue is with Lisa Heiss — a UX designer and PLG expert. We've collaborated on projects at ProductLed and bonded over a pretty similar approach to growth.

She didn't give me polished answers, thank God, it’s pretty spicy.

Btw, don’t ignore this 👇

submit a product for a breakdown

The user behavior that kills our lovely funnel

I asked Lisa what drives her crazy about users right now. She went straight for it:

❝

The attention span. How bad we've gotten at paying attention to anything longer than two seconds. If I'm trying to market a product to you and you can't even read three bullet points, it gets very hard."

Lisa Heiss

And the landscape around this has shifted fast. Users aren't stuck with you anymore (bummer, I know). They find another tool by tomorrow. They build their own over the weekend (or they think they can and that’s enough).

❝

"If I can't solve it with your tool, either I build my own or I figure out another tool. There's not much dependency on what you do."

Lisa Heiss

So you've got users who won't read your copy, won't sit through your onboarding, won't give you more than a few seconds, and they have zero reason to try harder because the next alternative is one search away.

Now look at what most teams do about it.

How teams make it worse

Mistake one: explain harder.

Teams see users skipping their flows and think the problem is comprehension. So they add more. More onboarding steps. More tooltips. More checklists. Lisa fights this approach like hell:

❝

"Show how the product drives and changes behavior. That is your explanation. If you can do that with fewer screens, that's the thing we care about."

Lisa Heiss

Your explanation should be the product working, not words about the product working.

I never adopted Notion (almost a crime for a person in SaaS). Never paid for it. I keep forgetting it exists, it never built itself into how I already work. My husband downloaded it once. Back then, they had this typed-in checklist on the main page as their onboarding. He didn't even register it as an onboarding. He saw text sitting there and moved on. Had no idea he was supposed to do anything with it.

For every clever onboarding, there's a person who just wants things very clear, very simple, and right next to where they are. Clever that doesn't land is invisible. Invisible loses users before they've done a single thing.

Mistake two: trap them.

Some products skip the explanation entirely and go straight to making it impossible to leave. Lisa has a real grudge against Adobe (who doesn’t?):

❝

"I was tricked into what started as 20 bucks a month. It renewed for 60, and it turned out to be an annual contract I was paying monthly, tied in without knowing it. I will do anything to not use Photoshop. Any AI product that does the same job, I'll use it just because of that."

Lisa Heiss

I had my own. A transcription tool, I needed one file done, signed up for what looked like a weekly trial, tried to cancel. No button. Nothing. I dug through their knowledge base, found a hidden email address, and sent a pretty aggressive message threatening a fraud claim with my bank (I know, very dramatic). The product did its job fine. But I will never open it again.

No one loves an abuser. We leave abusers. It's 2026, for God's sake.

❝

"People aren't obligated to stay with you. So these techniques just mean you end up with no loyal customer base. Nobody loves you for doing that."

Lisa Heiss

Mistake three: stop at the aha moment.

This is the one nobody wants to hear. I asked Lisa what she used to believe about growth that she's completely changed her mind on. She went straight for the sacred cow:

❝

"I'm not saying the aha moment doesn't matter, but that is not it. If you stop there, it's as good as having a good first date. You won't make him your boyfriend because you had a good first date."

Lisa Heiss

I've had a chicken-and-egg problem with this for a long time. Does performing this action make users motivated? Or are the users who perform it already more interested, and that's why they do it? If you bully everyone into one activation event, what happens to users who would've found value a different way? Would you even see the same correlation between that event and long-term upgrades if it's the only option you give people?

The part almost nobody designs for is what comes after. The daily loop. The thing that brings someone back on Tuesday without a reminder. Lisa said it directly:

❝

"Building a habit is incredibly hard. Very few designers are educated about it. Very few companies think about it in enough detail to actually make the product sticky."

Lisa Heiss

How many good habits do any of us even have? I quit my diet this week because a friend brought wine, and I was like, well, farewell. That's how easily a habit you actually want breaks. Now imagine building one around your SaaS product, in a market where users have a hundred alternatives and zero reason to be loyal.

How to earn an upgrade from a professional

I asked Lisa which products she started using for free and ended up happily paying for. Three came up. They all did the same thing, but each in a different way.

Magic Patterns — an AI design/prototyping tool that I love love love.

Lisa was skeptical, tried it free, got into a creative flow, and ran out of credits. The upgrade prompt didn't slam a gate. It pitched what she'd gain:

❝

"The way you were asked to upgrade didn't feel intrusive. They were marketing benefits instead of taking something away, you'll get double the speed, you'll deliver your work faster. It felt like a soft switch. You could come back the next day with free credits. Nobody was pressuring you."

Lisa Heiss

I know that feeling from Lovable. You're building something, you're in the middle of it, credits run out, and you just aren't done. The upgrade doesn't feel like a wall. It feels like continuing. But I pushed Lisa, because I've hit that "not done" moment with plenty of products and didn't upgrade most of them. Same interruption, completely different reaction. The difference is whether the product earned enough momentum before it asked. Whether you already experienced enough value that paying feels like a continuation, not a transaction.

CleanMyMac — laptop cleaning software, the kind of product Lisa would normally never buy. But before it mentioned money, it ran a full diagnosis of her machine:

❝

"You felt like they were giving you something real, this is how your laptop looks right now, here's what's working versus what's not. I wasn't learning about the product. It was showing me how it can help. It just felt effortless."

Lisa Heiss

The diagnosis was the onboarding. Value first. Payment second. No explanation needed because the product already proved itself.

Granola — a note-taking app that took the same principle even further. It read her calendar, detected a meeting, and just started working. No setup, no bot joining the call, nothing to configure. Every step she would've had to take was already handled.

Three products. One pattern. None of them tried to teach Lisa how they work. They all fit into the way she already works:

❝

"Our job is not to educate the users. Our job is to match the way they already think right now."

Lisa Heiss, I couldn’t agree more.

And transparency counts just as much. If users know exactly what they'll get, if that gap between "I don't know what will happen" and "I know exactly what I'm paying for" is as short as possible, trust builds fast.

One piece of advice for every growth team

I asked Lisa for the one thing she'd want every founder and growth team to hear right now. She said the fear in the market, am I being replaced, is my product even relevant tomorrow, is paralyzing people who should be moving:

❝

"The shift isn't as black and white as the market makes it out to be. You're not an old SaaS that can't renew, and you're not a new AI product just hoping to survive. There's a very valid middle path: you have the distribution, you have a problem worth solving. Go innovate on that. Even if it's small pieces, even if the AI is for certain challenges only."

Lisa Heiss

She's right. I see this with teams I work with: they're so overwhelmed by the idea of rebuilding everything that they rebuild nothing. Meanwhile small, targeted improvements could already be making their product stickier.

Want to read more issues like this? Like, comment, and let me know. Have suggestions? Just reply to this email.

🎉 Woow, you finished the issue, that’s awesome!

Hi, I’m Anastasia Kudrow, and I write Ghosted.

I also help SaaS teams get more upgrades by using psychology instead of cheap tricks. Because hype aside, people will be buying your product for many more years, AI or not AI.

Some of you know I work with Wes Bush — author of Product-Led Growth, and a whole team of other great PLG people. Internally, we keep coming back to the same thing: underestimating the gap between acquisition and activation is one of the most common mistakes team make.

And lately we've been talking more and more about how, especially with AI, activation doesn't have to wait until people sign up. Your landing page should already be doing that work.

So this week I went looking for products that actually pull it off. Not one example — three. Three levels of "show them the value before they commit." Let me walk you through what I found and why it works.

Btw, don’t ignore this 👇

submit a product you are curious about

The Find

Quick context: all three of these are AI-ish products with complex value props. The kind where a hero section and a feature grid don't really explain what you're getting.

1. The end result.

Relay.app is workflow automation — Zapier meets AI agents. On their landing page, you can browse actual templates other people built. Click through, see the agent, see the workflow, see the output. Before you sign up, you already know what problems this thing solves and roughly how your result will look 👇

I don't know about you, but that's way more convincing than "automate your workflows with AI" in a headline. I can see how it will look like for me.

2. The daily experience.

Weave measures engineering performance and AI impact. Their landing page has these interactive, animated dashboard previews — not static screenshots, not marketing renders. You can see how data moves, how charts respond, what it would feel like to check this every morning 👇

The less cartoonish and polished it looks, the better. This feels like peeking into the actual product. My brain stops treating it as an ad and starts treating it as experience.

3. The start.

Lovable’s landing page is the product. There's a prompt field right there. You type what you want, and it starts building. No account, no email, nothing. By the time they ask you to sign up, you've already got the sense of progress you don't want to lose 👇

So: three products, three levels. Show the output. Simulate the experience. Let them do the thing. Does your landing page do any of these?

Why it works: psychology behind

There's a concept in consumer psychology called mental simulation — your ability to picture yourself using something before you commit to it.

Most landing pages force the user to do all the simulation work themselves. Read the headline, look at a screenshot, imagine what it means for your specific job. The problem? Imagination is lossy. Every gap you leave, the user fills with doubt.

Each of these three products closes that gap differently. Relay replaces imagination with real artifacts — you see the output, so you don't have to guess.

Weave replaces static images with motion — you see the interaction, so your brain treats it as experience rather than marketing.

Lovable removes the gap entirely — you're not simulating use, you're doing it, and now Bandura's self-efficacy kicks in. You don't wonder if you can use this product. You already proved you can.

The progression matters: the more concrete and interactive the preview, the less the user has to imagine, and the less they doubt.

The Playbook

When to keep it in mind:

Any product where the gap between "landing page" and "first real value" is long enough to lose people. AI products especially — because AI makes it cheap to generate previews, interactive demos, and real product experiences dynamically. If you're still relying on hero images and feature bullets, you're leaving activation on the table.

How to use it:

1. Start with the output.

Let people browse real results — templates, reports, projects that existing users built. Relay does this well. The cost is low: you're surfacing content that already exists. I'd push most teams to start here — it's the fastest win with the least engineering effort. If your product's value is visible in its artifacts, show them.

2. Add motion.

Go beyond static screenshots. Interactive previews, animated walkthroughs, GIFs that capture what daily use looks like. Weave does this with their dashboard demos. The key: it has to feel real, not produced. The moment it looks like a marketing render, the user's brain discounts it. Rough and honest beats slick and staged every time.

3.Drop the gate entirely.

Let the user start a real task before asking for credentials. Lovable nails this — type a prompt, watch your app get built(ish), sign up only after you have something to lose. Fair warning: this only works if your product can deliver a meaningful result fast or showcase the process at least. If you need config, data import, or team setup before value, Level 1 or 2 is the smarter move. But honestly? I’m sure some part can be delivered earlier.

When it backfires:

If the real product experience doesn't match the landing page preview. You've set an expectation now — and you have to deliver on it immediately after signup. An interactive demo that looks gorgeous, followed by an empty dashboard and a 12-step onboarding wizard, will hurt you more than a generic landing page ever would. The in-product experience has to pick up where the landing page left off and be a logical next step.

If you forget everything, remember this:

Your landing page doesn't exist to describe value. It exists to simulate, show it. The less your user has to imagine, the less they have to doubt — and the more likely they are to sign up and actually stick.

🎉 Woow, you finished the issue, that’s awesome!

Hi, I’m Anastasia Kudrow, and I write Ghosted.

I also help SaaS teams get more upgrades by using psychology instead of cheap tricks. Because hype aside, people will be buying your product for many more years, AI or not AI.

The last couple of months have been dense. I covered everything from free tier psychology to AI agents to why I personally rage-quit a payment screen. I know you don't read every issue (even though you should! Ghosty is watching you) so here are the five most popular ones from the past 2 months. Picked by your peers and yourself.

Btw, a reminder that you can always submit ANY product you are curious about for my breakdown ↓

submit a product you are curious about

#1 — How to drive advanced feature adoption

I didn't expect this one to land so hard, but it did. If you're still onboarding users through passive checklists — "install the extension," "connect your CRM" — this breaks down why that approach teaches nothing.

#2 — What kills a great upgrade trigger

Grammarly built one of the best upgrade triggers I've ever seen — and then killed it with the checkout flow. There's a template inside for how to design an upgrade trigger that doesn't reopen decisions your user already made. Save-worthy if you're touching anything on the payment side.

#3 — How an AI agent can make things worse

If you're adding an AI chat to your funnel — or thinking about it — read this before you ship. G2's bot drained a warm lead's motivation in under 20 minutes. Three redesigned chat mockups inside showing what "actually helpful" could look like.

#4 — 3 reasons I didn't pay Claude

I wanted to give Anthropic money. Like, actually wanted. And then their upgrade screen scared me off in four screens. This one's about how to design a payment moment that doesn't make your users do mental risk math for a purchase they've already decided to make.

#5 — Why users don't engage with your free plan

Most free users have never seen your pricing page. They saw "free," they signed up. So if your onboarding doesn't tell them what's included before setup — they won't set anything up, won't get a result, and won't upgrade. Relay.app nailed it with one screen. Three checks inside for designing a free tier that actually gets people to a result.

🎉 Thank you for reading Ghosted. I’ll see you next Sunday!

Hi, I’m Anastasia Kudrow, and I write Ghosted.

I also help SaaS teams get more upgrades by using psychology instead of cheap tricks. Because hype aside, people will be buying your product for many more years, AI or not AI.