The paywall audit: three levers, and the five things people get wrong

The paywall is one of your highest-leverage surfaces for LTV:CAC, and it runs on three levers rather than one: how well it converts, where it is placed and exposed after value has landed, and the LTV of the plan people end up on. Win one and lose the other two and you get a paywall that looks like it is working. Below are the five things I see missed most, in the order they cost you, each with its publisher and its caveat.

The paywall setup with your best day-one conversion can be the one with your worst twelve-month LTV. Adapty has found exactly that: the configuration with the lowest Day 0 conversion often produces the highest twelve-month value.

Which is why "how is our paywall converting" is a third of the right question.

The three levers

One: how well it converts. The screen itself. Framing, defaults, what it promises, what it does with an objection.

Two: where it is placed, and how it is exposed after value has landed. A paywall is not just a screen, it is a moment. The same screen at a different moment is a different product.

Three: the LTV of the plan people end up on. Two paywalls can convert identically and be worth very different amounts, because one of them lands people on a plan they keep and the other lands them on a plan they cancel.

Almost every paywall conversation I get pulled into is about lever one. Most of the money is in levers two and three, and lever three is the one nobody has looked at.

The five things people get wrong

1. The paywall arrives after intent has already decayed

89.4% of all trial starts happen on Day 0, and 82% of purchases happen in the first session (Adapty 2026). Onboarding-placed paywalls with trials convert at 1.78% against 0.89% for in-app placement. Rootd reported a 5x revenue increase after moving theirs to the start of onboarding, and a plant-care app went from 3% to 15% on sign-up-to-trial.

Download intent decays fast, and users who delay a trial rarely come back to start one. That is lever two, and it is usually the cheapest of the three to move because it requires no new design at all.

The exception is real and it is Education. Only 71.3% of Education trials start on Day 0, and 23.5% start on Day 31 or later. People in that category genuinely browse, leave and come back with intent, so hard-gating at the end of onboarding cuts off the return trip.

2. Annual is priced as a lump sum

Re-anchoring the annual price to a monthly equivalent lifted new revenue per paywall impression by 45% in Brazil, 26% in Mexico and 8% in the USA, at zero price change (Mojo, first-party, reported via RevenueCat). Per-day pricing display lifts annual adoption by 20 to 40%, though that figure comes from industry reports rather than a single controlled study, so read it as directional.

Look at the spread across those three markets, because it tells you what drives the effect: it is biggest where a lump sum feels prohibitive relative to income.

3. The default plan is wrong, or there is no default at all

Switching the default from monthly to annual inverts the plan mix, roughly 37% yearly to 63% yearly. Among users who did not close the paywall, 82% simply took whatever was pre-selected. One app pre-selecting annual saw a 70% increase in annual revenue. That is Superwall network data, observational rather than randomized, so treat it as a strong prior rather than a proven causal effect.

This is lever three hiding in plain sight, and it is the element nobody argued about.

Almost every element on that screen was decided deliberately. The price. The headline. The badge, the artwork, the order of the benefit list. Someone argued about each one, probably in a meeting with a designer present. And then there is the toggle, which loads on whatever it loaded on the day it was built.

That toggle is doing more work than the artwork. A lot of paywall performance is decided by the options you never framed as decisions, which is why redesigns so often move nothing: they change everything that was already argued about and leave the one thing that was not.

4. It lists features when it should promise an outcome

Features answer "what do I get." Outcomes answer "what happens to me." Only one of those is why anybody downloaded.

One meditation app replaced a features carousel with a preview of the first week's program and moved paywall conversion from 20% to 30%, a 50% lift. That figure comes from a secondary source, so treat it as medium confidence. The principle underneath it is well established and is not in doubt.

5. Nothing catches the people who say no, and nothing reassures the people who are scared of the trial

The single biggest objection at a trial paywall is not price. It is the fear of forgetting to cancel.

Blinkist's transparent trial timeline, showing the start day, the reminder day and the charge day, produced a 23% conversion lift, 55% fewer complaints, and push opt-in going from 6% to 74%. That one was A/B tested and reported first-party by their Head of Product, which makes it one of the better-evidenced numbers in this whole area.

Then there is the other end. Exit offers contribute 15 to 20% of total revenue (Superwall), post-paywall welcome offers produce a 10 to 15% ARPU uplift (Adapty), and roughly 50% of users who start an in-app purchase never complete it. That last number is the one worth sitting with: half the people who got as far as tapping buy are still leaving, and most paywalls have nothing at all built for them.

"Always offer a trial" is wrong for about half of app categories

This belongs to lever three and almost nobody checks it.

Adapty's category analysis has trials actively hurting LTV in four of eight categories: Productivity at -13.7%, Lifestyle at -21.2%, plus Entertainment and Graphics & Design. Utilities sees the largest premium at +85.1% and Health & Fitness +63.6%.

Our own paywall library calls this the single most important nuance in the paywall benchmarking space, and it is the clearest example of why lever three exists as a separate lever. A trial can raise your conversion rate and lower what a customer is worth. Both things are true at once, and only one of them shows up in the dashboard people look at.

The four questions

So when someone asks me to look at their paywall, I do not start with the design. I ask four things, and I ask them in this order because the order follows how a user actually meets the screen.

<circle cx="120" cy="256" r="11" fill="hsl(var(--chart-1))"/>
<text x="120" y="261" text-anchor="middle" font-size="12" font-weight="800" fill="hsl(var(--chart-1-foreground))">1</text>
<circle cx="280" cy="300" r="11" fill="hsl(var(--chart-1))"/>
<text x="280" y="305" text-anchor="middle" font-size="12" font-weight="800" fill="hsl(var(--chart-1-foreground))">2</text>
<circle cx="175" cy="396" r="11" fill="hsl(var(--chart-1))"/>
<text x="175" y="401" text-anchor="middle" font-size="12" font-weight="800" fill="hsl(var(--chart-1-foreground))">3</text>
<circle cx="175" cy="70" r="11" fill="hsl(var(--chart-2))"/>
<text x="175" y="75" text-anchor="middle" font-size="12" font-weight="800" fill="hsl(var(--chart-2-foreground))">4</text>

<text x="360" y="112" font-size="11.5" font-weight="700" fill="hsl(var(--muted-foreground))" letter-spacing="0.08em">FOUR QUESTIONS, IN THIS ORDER</text>
<text x="360" y="168" font-size="13" font-weight="800" fill="hsl(var(--foreground))">1. The toggle</text>
<text x="360" y="188" font-size="11.5" fill="hsl(var(--muted-foreground))">What does it load on, and who decided that?</text>
<text x="360" y="248" font-size="13" font-weight="800" fill="hsl(var(--foreground))">2. The annual price</text>
<text x="360" y="268" font-size="11.5" fill="hsl(var(--muted-foreground))">Doing the mental math, or asking for it?</text>
<text x="360" y="328" font-size="13" font-weight="800" fill="hsl(var(--foreground))">3. The billing date</text>
<text x="360" y="348" font-size="11.5" fill="hsl(var(--muted-foreground))">Can they see when they will be charged?</text>
<text x="360" y="408" font-size="13" font-weight="800" fill="hsl(var(--foreground))">4. The screen before this one</text>
<text x="360" y="428" font-size="11.5" fill="hsl(var(--muted-foreground))">Half of what is wrong is not on the paywall.</text>

What does the toggle load on, and who decided that?

The answer is almost always "it has always been that way."

That is a fossil sitting on your highest-intent screen. It was set by whoever built the first version, usually before the pricing was final, and it has survived every redesign since because a pre-selected state does not look like a decision when you are reviewing a mockup.

Ask who chose it. If nobody can name a person or a reason, you have found a test.

Is the annual price doing the mental math for the user, or asking them to?

A yearly figure presented as a lump sum asks the buyer to divide, at the exact moment they are least willing to do arithmetic.

Most of them will not do it, and the ones who do are doing it while deciding whether to trust you. The monthly-equivalent framing costs nothing to implement and changes nothing about what you collect.

Can they see when they will be charged?

The fear at the decision point is not "is this worth it." It is "will I forget to cancel."

If the screen does not answer that question, the user answers it themselves, and the safe answer is to leave. Showing the start date, the reminder, and the billing date turns an open-ended commitment into a bounded one. That is why making cancellation legible tends to make starting easier rather than harder.

What happens on the screen right before this one?

Half of what is wrong with a paywall is not on the paywall.

If the user arrives without having felt the product do anything, the screen is being asked to sell a promise instead of confirming an experience. No amount of framing fixes that, because the problem happened before the paywall loaded. This is the question that most often redirects the whole engagement, and it is the one people are most surprised I ask.

Why defaults survive one redesign after another

Usually at least two of those four have never been examined. That is not sloppiness.

Defaults do not announce themselves as choices. When you review a paywall mockup, you see a screen with a plan already selected, and your eye goes to the things that vary between versions: the headline, the artwork, the button copy. The selected state reads as part of the furniture, not as a variable, so it does not enter the conversation and it survives.

The same thing protects the missing billing date. Nobody reviewing a screen notices the sentence that is not on it.

The rule I would hand you

Anything on that screen you cannot give a reason for is a test you have not run yet.

Walk the paywall element by element and try to say out loud why each one is the way it is. The elements with real answers are decisions, and decisions can be defended or revisited on purpose. The ones that produce a shrug are your backlog, ranked by how much traffic passes through them.

The toggle is usually at the top of that list, because it has the most traffic and the least deliberation of anything on the screen.

What a paywall audit actually hands you

Since this is what I do, here is the shape of it rather than a pitch.

The body is short on purpose: our goal, a two-to-three sentence read, a rating stated against the achievable ceiling of the fixes rather than against any observed conversion number, "do these now" grouped by paywall element (plan cards, billing toggle, trial CTA), and the numbers that matter with one number per metric. When no funnel data was supplied, that section becomes "here are the bars" and says so explicitly rather than implying a measurement nobody took.

The appendix carries the benchmark table in full form with its spreads and read-as caveats intact, what we observed, the full ICE-ranked backlog, and paywall comparables when a competitor set exists.

Minimal input is paywall screenshots plus price points, or a public pricing URL. Without screenshots or a price the run does not start, because there is no version of this built from a description of a paywall instead of the paywall.

Run against a public pricing page with zero funnel data, the top five moves in one sample were all framing: express the annual price monthly, restate the saving in real money, flip the default toggle, split the plan grid by who is buying, and put a trial timeline under the CTA. Not one of them touched the price. That was a worked sample on a public company rather than a client engagement, which is the honest way to describe it.

The takeaway

Three levers, and most paywall work only touches one.

How well it converts, where it sits relative to the moment value lands, and the LTV of the plan people end up on. Before you brief a redesign, run the four questions: what the toggle loads on and who chose it, whether the annual price is doing the arithmetic or outsourcing it, whether the billing date is visible, and what the user experienced on the screen immediately before.

Most teams find at least two answers they cannot defend, and those two are cheaper to change than anything a redesign would touch.

Common questions

What are the three levers on a paywall?

How well it converts, where it is placed and exposed relative to the moment value lands, and the LTV of the plan people end up on. Most paywall work only touches the first. Two paywalls can convert identically and be worth very different amounts, because one lands people on a plan they keep and the other on a plan they cancel.

Should every subscription app offer a free trial?

No, and this is the least-checked decision in the category. Adapty's category analysis has trials hurting LTV in four of eight categories, Productivity at -13.7% and Lifestyle at -21.2%, against Utilities at +85.1% and Health & Fitness at +63.6%. A trial can raise your conversion rate and lower what a customer is worth at the same time.

Why is my paywall not converting even after a redesign?

Because redesigns change the elements that were already deliberate and leave the ones that were not. The plan toggle's default state, the absence of a visible billing date, and the framing of the annual price are rarely revisited in a design review, since none of them read as a variable when you are comparing mockups. Superwall's data has 82% of non-closing users taking whatever plan was pre-selected, so the default is carrying more conversion weight than the visual design it survived.

What should I check on my paywall before redesigning it?

Four things, in order: what the plan toggle loads on and who decided that, whether the annual price is expressed as a monthly equivalent or as a lump sum, whether the user can see the date they will be charged, and what the screen immediately before the paywall did for them. Each is cheap to change and none requires new design work.

Does showing the billing date on a paywall hurt conversion?

It tends to help. The hesitation at the purchase decision is usually about forgetting to cancel rather than about the value of the product, so a visible start date, reminder date, and billing date removes the specific unknown that causes people to postpone. Products that add the timeline generally see both conversion rise and billing complaints fall.