Most monetization advice for subscription apps is about getting someone to pay. Far less is said about what happens right after: why new subscribers cancel within days, whether trials still make sense, and whether people who leave can be won back.
To cover everything around that first payment, FunnelFox hosted a roundtable with three growth practitioners:
- Phil Carter, founder and CEO of Elemental Growth, a growth advisory firm for subscription apps, and host of the Subversive podcast. Previously, he worked at Faire, Quizlet, and Ibotta.
- Jacob Rushfinn, co-founder of Botsi, a personalized monetization platform that helps apps show the right offer to each user, and author of Retention.Blog.
- Natalia Burakova, an independent product and growth advisor to apps, who has worked on more than 50 mobile apps.
Below are the main takeaways: why subscribers cancel so early, how onboarding prevents it, which plans and trials to offer, whether win-back is worth building, where to find more revenue per user, and when a new app should add a paywall.
Why do subscribers cancel within days of paying?
FunnelFox data shows how early subscription churn starts. For a typical app, about a third of monthly subscribers who cancel turn off auto-renewal within three days of paying. And once a subscriber turns auto-renewal off, fewer than 2% ever pay again.
The real window is even shorter than three days. Most of those cancellations happen within the first few minutes, which makes the first session the deciding moment.
“It’s all about that first session, those first few minutes in your app. Can you get users to experience value? Can you communicate what the point of using your app is?”
People usually cancel for one of two reasons. Either they aren’t sure the app is right for them and turn off renewal just in case, or they are confused and don’t like what they see.
“Most users cancel just in case. They’re not sure the product fits their needs, and they don’t want to get billed for it. They may forget about the subscription in the next five minutes because a phone call distracts them.”
Some early cancellations don’t mean anything went wrong. Younger users turn off auto-renewal at a higher rate: they know how subscriptions work and don’t want anything to convert automatically, even if they love the product. They want renewal to be their own choice. The rate generally goes down with age.
Tactics like the Blinkist-style trial timeline on the paywall and trial reminders help reduce early cancellations. The main fix, though, is the product experience new users get.
“Ultimately, it’s about creating an engaging, compelling experience for new users, so they think about the value they can get from your product, and not ‘Am I going to have to pay for this? Am I going to forget about it?'”
How do you keep new subscribers from canceling?
Since early churn is decided in the first session, the work starts before the first payment, in app onboarding.
Sell the result users came for
Take someone who clicks a weight loss ad and lands in a web funnel. They want to lose weight. They weren’t looking for a new subscription. So the paywall, on web or in the app, shouldn’t promote a specific feature or plan. It should sell the idea that this product will help the user reach their goal, at a price that makes sense for that promise.
That’s why onboardings in health, fitness, and weight loss are long: they need time to show how the app’s tools, programs, and rituals will get the user where they want to be. By the paywall, the user should be fully convinced.
The stakes are high. By most stats, around 90% of all conversions happen during onboarding. If the app doesn’t sell the product there, it rarely gets another chance: either users try the free features and find them enough, or a hard paywall blocks everything and pushes them into a plan they aren’t ready for, which they then cancel.
The value needs to be explained several times, in different ways: through the copy, visuals, and social proof, such as examples of other users.
“Finish the onboarding with the paywall, because there is no second chance there.”
Get users to the aha moment in the first session
The aha moment, as Reforge defines it, is the point when the user understands what makes your product unique and worth paying for. With AI, the bar for that moment keeps rising. User expectations are fundamentally different from what they were before ChatGPT went mainstream, so apps need an “automagical” first-time user experience.
Timing matters as much as quality. Data shows that over 80% of trial starts often happen within the first 24 hours, which usually means the first session.
“More and more consumers aren’t going to read anything you put in front of them. They just want to experience the product.”
Three apps show how this works:
- Duolingo doesn’t explain why it’s a more fun way to learn a language. It asks a few questions about your language and goals, then drops you into the first lesson, and you’re learning within 60 seconds.
- Wispr Flow turns voice dictation, a mundane category, into one of the most engaging onboardings Phil has seen. It shows its accuracy and low latency right away and gets users nodding, laughing, and raising their eyebrows. That quality is why the company is now valued at $2 billion.
- Tolan, an AI guide through life, has users answer a few questions from “the oracle” and then matches them with an AI companion. It builds an emotional connection with new users faster than any other product Phil has used.
The impression these apps make in the first two to three minutes affects trial starts, trial conversion, and long-term retention.
The benchmark for that impression isn’t set by your category.
“Consumer expectations of a great product experience change based on everything they see and interact with. You can’t compare yourself to just your competitors. You have to compare yourself to the entire world.”
AI also opens a window for apps in categories that haven’t changed in years. When Duolingo appeared in 2012–2013, Rosetta Stone was the gold standard for learning a language. Duolingo used mobile, gamification, bite-sized lessons, and personalization to reinvent the category and made Rosetta Stone irrelevant. AI gives apps the same chance today, and voice dictation is one example.
How long should app onboarding be?
It depends on the product. There are two signals that a longer onboarding can work:
- Traffic from social ads. People arriving from Meta were scrolling and doing something else, so the onboarding has to build up their intent to buy.
- An emotional product. When the motivation is self-improvement, like losing weight, improving mental health, or learning a skill or a language, a longer flow makes sense. For casual, day-to-day products, a long onboarding can do more harm than good.
Noom and AllTrails show the two ends of the spectrum.
Noom’s onboarding is famously long, and it works. People in weight loss have often tried other apps, diets, recipes, and exercise that didn’t work for them. Noom asks enough questions to make users feel heard and to raise their confidence that this time it will work. The length also creates a sunk cost: by the paywall, users have invested so much time that leaving means starting over.
AllTrails built its business on hyperlocal SEO. Anyone can find a trail on Google without the app, so people download it to record their activity, often while standing at the trailhead. A long quiz would only frustrate them. AllTrails lets users register with SSO, shows one paywall in case they’re ready to pay, and then lets them start recording right away.
“It comes back to what problem you’re solving. Build the onboarding experience around that.”
What if your product takes months to deliver results?
In fitness, weight loss, or job search, the end result takes time. The way to keep users until then is to give them an early win in the first session:
- A job search app can ask a few questions and show several personalized matches before asking for money.
- A fitness app can show projected progress: based on the user’s answers, how much weight they could lose in seven days.
- An AI journaling app can create a personalized journal entry during onboarding, so the value of a months-long self-care habit is clear right away.
“There’s almost always some way for someone to understand and experience the value immediately and go, ‘Oh, I get it.’ That’s why it’s called an aha moment. If you can get someone to go ‘aha’ out loud, you’ve won.”
After that, small daily mechanics keep users coming back, as Duolingo does with push notifications and streaks. A simple task users can check off every day gives them enough dopamine to return tomorrow.
The aha moment can be vague, so it helps to define it with an activation metric:
- Brainstorm six to twelve candidate activation metrics.
- Run a regression analysis to see how each one correlates with three- or six-month subscriber retention.
- Pick the metric with the highest R-squared that still happens early enough to be a leading indicator.
Elemental Growth uses this approach with a weight loss client. The team can’t wait months for subscribers to lose weight, so it picked metrics that happen far earlier, such as cooking the right recipes, and correlate with long-term retention.
Which plans and trials should a subscription app offer?
The plans an app offers shape who subscribes and how long they stay.
Where weekly plans fit
Weekly plans get a lot of attention, but they don’t fit every app. A health and fitness app is built around a habit that takes months, so plan length should match it. That’s why very few apps in this category use weekly plans and skew toward monthly, quarterly, and annual ones.
Weekly plans work best as a substitute for a trial in apps with a one-off use case. Imagine someone who needs to send a fax once, for a car registration. With a free trial, they would use the app and cancel right away. With a weekly plan at $8.99, they pay for the week and maybe stay a few more. AI photo and video apps fit the same pattern: people are excited to create something, but they don’t have a long-term use case.
In most other cases, weekly plans come with very high churn.
“People should be more worried about weekly plans than excited. You’re moving revenue upfront at the sacrifice of building a long-term, sustainable subscription business. As a subscription business, you want to build compounding cohorts over time.”
Free trial, freemium, or hard paywall
In FunnelFox data, most apps pick one extreme: they either start almost every new subscriber on a free trial or don’t offer a trial at all. The right choice depends on two things.
How the app grows. Apps that grow mostly through paid ads tend to be conservative with free trials and freemium features. Cash fuels their growth loop: they buy users on Meta or Google, aim for day-8 ROAS above 100% (day 8 because of the seven-day trial), and reinvest the money into more ads. For them, a hard paywall or a very limited trial usually works better.
Apps that grow organically can be more generous. At Quizlet, where Phil Carter was a growth leader, most users came through word of mouth between teachers and students and through user-generated content that drove SEO. Free users bring value beyond the money they pay, by recommending the product or creating content that helps it grow.
How fast users reach the aha moment. If users understand the core value quickly, an app can skip the free trial or freemium tier. If it takes longer, the app needs to give away more value for free.
AI has changed the math on top of that. The cost of serving a digital subscriber used to be close to zero. AI features run on LLMs with real inference costs, so apps are moving from seven-day to three-day trials and putting more features behind the paywall.
“You have to maintain healthy unit economics, or you don’t have a sustainable business. Those costs have to be factored into your monetization strategy, which means being less generous with trials and free offers.”
Why do web funnels use paid intro offers instead of free trials?
On web, apps often replace the free trial with a paid intro offer, such as a seven-day plan for a dollar or two. The goal is the same as a free trial, but it filters out chargebacks and failed payments. In the App Store, Apple handles both. On web, the app has to deal with them itself.
There is also an ad platform reason. Most web funnels skip free trials so they can send Meta the right signals and get the right users in return.
That changes what short plans mean on web. A weekly or monthly plan becomes a de facto trial, and apps should expect cancellation rates at the level of a trial.
“People use your cheapest plan as a safe trial option. Maybe the product will work, and that’s great. If it doesn’t, $8.99 is not a big deal. They can live with it.”
The “choose your trial” paywall
One newer approach shows two annual options side by side: the regular annual plan with a free seven-day trial, and an annual plan with a longer paid trial, for example 30 days for $3.99.
Most apps want as many users as possible on annual plans, but showing only an annual plan leaves users no choice. Two annual options give them one while keeping everyone on annual. For most apps, very few people buy the longer paid trial. They look at it and pick the free one.
“You’re moving the choice from ‘Should I buy?’ to ‘What should I buy?’ This is a super important concept in choice psychology.”
This paywall has a catch. By showing only annual options, it removes some price-sensitive users. If an app had a monthly plan with strong renewals, those monthly subscribers would have stayed for a long time, so the switch can raise upfront revenue at the cost of long-term revenue.
A simple check: how good are your monthly renewals? For most consumer apps, they aren’t great, and this paywall is often a good option. If yours are decent, compare the LTV of monthly subscribers with the upfront annual revenue before switching.
Is it worth winning back subscribers who cancel?
In FunnelFox data, fewer than 2% of customers who turn off auto-renewal ever pay again. Part of the reason this number is so low is that many apps never try any win-back mechanics at all. Well-designed win-back campaigns can bring back around 10% of overall revenue.
Still, win-back isn’t where to start. Changes to onboarding, the paywall, and subscription plans have a far bigger impact and a much higher ROI for the time invested. Win-back campaigns, cancellation flows, and a CRM system to follow up with churned users make sense once the base is in place: onboarding performs well, plans cover both users willing to pay more and price-sensitive ones, and the first session is engaging.
“When all of this is set up, win-back is definitely a thing to do. But I would never say that it’s the first priority.”
What happens to users who take a discount
A common second offer is a discount shown when a user closes the paywall. Some reports claim it adds 15–20% in revenue, but for most apps, that’s unrealistic. A more realistic goal for all second offers combined, including win-back, discounts, and renewal offers, is around 10% of revenue. Around 5% is more common.
Users who convert on a discount will likely renew at a lower rate. They are more price-sensitive: they weren’t sold by the product’s value at first, and what changed their mind was a discount. Even so, the result is almost always net positive.
“If you don’t show a second offer while they’re in your app, you’ve lost them forever. The churn is so high that your only chance is now.”
The exception is a strong freemium product with good retention. There, many free users would convert over time anyway, so discounting upfront can sacrifice long-term revenue. This case is rare.
There is also a group of users who never pay full price on principle. Users today are more tech-savvy and know how app subscriptions work, so some of them always look for a discount, before the trial or after canceling it. They may renew for a shorter period or pay a little less over the same lifetime, but they still bring revenue.
“At the end of the day, they will still bring you significantly more revenue than zero, and any revenue is always better than zero.”
Where to find 20% more revenue per user
Optimize your web checkout
In the US, Apple now lets apps add a web checkout button inside the app. App-to-web checkout flows have worked well for many companies over the past year. Right after the Apple vs. Epic ruling, results were weaker: consumer expectations hadn’t shifted yet, and companies hadn’t learned to streamline the flows.
There have been setbacks. Cal AI was temporarily removed from the App Store for breaking Apple’s developer guidelines. But as long as apps stay within those guidelines, web checkout is fairly safe, at least while the ruling holds. Apple is appealing it.
Web checkout is a quick optimization that claws back some of the margin from the app stores.
Look beyond web checkout
More strategic levers can unlock more value.
1. Build new products and features. This takes the most effort, but it creates more value for users and raises their willingness to pay. Jimmy Liu, one of the first PMs at DoorDash and the creator of DashPass, told Phil Carter on his podcast that DoorDash hasn’t raised the price of DashPass since its launch in 2018. The company wants pricing to be the last thing it optimizes.
“Increasing price isn’t adding value for customers. If you increase price, you’re actually making the product less valuable for them relative to the dollars they’re paying.”
2. Revisit pricing. Many companies go far too long before they change their price. Strava launched around 2009 and didn’t change its price for over a decade. When it finally did, users had gotten so anchored on the old price that there was a big backlash, partly because the communication was fumbled. Methods like Van Westendorp, Gabor-Granger, MaxDiff, and conjoint analysis help get pricing and packaging right.
3. Use strategic discounting. Avoid blanket sales. Target discounts at the segments that benefit most: lower-intent users or groups with less disposable income, like students or educators. This aligns pricing with each segment’s willingness to pay.
4. Optimize the paywall. Even after onboarding has warmed users up, the paywall is where they make the purchase decision. That’s where the biggest wins often come from: 5%, 10%, or 20% lifts in trial start or subscriber conversion rate. Tolan, for example, personalizes the value proposition on its paywall based on what users told the oracle during onboarding. Emphasizing longer plans, annual over monthly or monthly over weekly, is another tactic.
Test your prices more often
Price testing is one of the fastest levers, and many apps undercharge. Indie developers in particular are often afraid to raise prices.
“Please test your pricing. Don’t set the price to fit your competitors. Set the price that matches the value you put into your product. Just test it and be bold.”
In many of Natalia’s tests, raising the price improved results, and conversion didn’t drop until the price reached a certain level. Many products are undercharging simply because their teams don’t believe the product is worth more. Prices should be retested every couple of months.
Keep the paywall simple
The best consumer paywalls are simple. People mostly don’t read long text, and consumers are busy and fickle. Prosumer and B2B products are different: the buyer is a professional, the price is higher, and the purchase is more considered.
The right message depends on the category. Tolan users decide based on how the product makes them feel, so an emotional appeal works better than a full grid of features. A photo editing app may need to list the technical specs that photographers care about.
If a paywall has to be more complex, two formats help:
- A multi-step paywall, pioneered by Finch, splits the purchase decision into three to five steps, often including a Blinkist-style trial explainer, so users can click through it.
- A video paywall replaces a paragraph of text with a 15-second video for products that need more explanation.
“The paywall is capturing all the intent you’ve built before. The best paywalls get out of the way and just reinforce what you’ve already said.”
That’s why the onboarding and the paywall deserve at least as much attention as the rest of the product. Building features is more fun, but changes to the first-time experience reach every new user.
“You should be spending equal time on the onboarding and paywall of your app as on the entire product. Potentially more. This drives all your success.”
When a new app should add a paywall
From day one, as soon as the app is in the App Store. Without a paywall, there is no strong signal that people care about what you’re building.
“The best signal of product-market fit is someone willing to pay for the product, not just using it.”
A paywall also shows what paying users actually do in the app and get excited about. Even trial starts are a useful signal. Indie developers often launch without a paywall, see installs and what looks like traction, then add a paywall and watch everything fall apart. Without it, there’s no real visibility into product-market fit.
The moment to shift focus from the product to onboarding and the paywall comes early.
“When the app looks good to the point that you’re still ashamed to show it to your friends, but you already can, that’s the moment to stop focusing on the product itself and start investing in your onboarding and paywall.”
Product-market fit comes first
For a brand-new app, the only thing that matters at first is product-market fit. Optimizing the funnel is a waste of time until users get enough value from the product to stay.
The most important metric is long-term subscriber retention: when, and at what percentage, the cohort retention curve flattens out. That answer takes months or years, so a faster leading indicator helps: the Sean Ellis test, a one-question product-market fit survey created by one of Dropbox’s first marketers. It asks users how they would feel if they could no longer use the product: very disappointed, somewhat disappointed, or not disappointed. If 40% or more say “very disappointed,” it’s a strong sign of product-market fit.
Once the core product works, optimizing onboarding and the paywall brings more than building the next feature. A testing plan prepared in advance rarely survives launch. A better approach:
- Run the app for at least a week to a few months, depending on traffic and budget, to collect initial data.
- Break the funnel into parts and find where it performs best and worst.
- Build hypotheses for the biggest problem or the lowest-hanging fruit.
- While one test runs, start looking into the next part of the funnel: onboarding, then the paywall, then pricing, then back again.
“It’s more a loop than a structure. The key thing is to have clarity on your data and to see where you need to target your efforts right now.”
