A subscription app marketing strategy is the chain of decisions that turns ad spend into renewing subscribers, and it is built backwards. You start with net revenue per install, which is set by your paywall conversion, your price after store commission and how long subscribers renew for. That number tells you the most you can afford to pay for an install. Everything upstream, the channel, the optimisation event, the creative, exists to buy installs below that ceiling at volume. Apps that grow profitably in 2026 are usually not the ones with the cheapest installs. They are the ones that know their ceiling and can produce enough creative to keep buying under it.
That framing matters because the market has stopped rewarding effort evenly. RevenueCat's State of Subscription Apps 2026, built on more than 115,000 apps and over $16 billion in revenue, found the top quartile of apps grew monthly recurring revenue by 80% or more year on year while the bottom quartile shrank by more than a third. The gap is not mostly a media buying gap. It is a gap between teams who treat acquisition and monetisation as one system and teams who run them as separate departments with separate dashboards.
This is the operator's version of that system. It covers the funnel and what the current data actually says about each stage, why two credible benchmark reports can look like they contradict each other, how to convert your own funnel into a spend ceiling, and where to intervene when a number slips. It assumes Meta is your main paid channel, because for most consumer subscription apps it is.
The five decisions a strategy has to make
Before any of the benchmarks are useful, it is worth being clear about what a strategy is deciding. There are five, and most app teams have only consciously made two of them.
- Who you are buying. Not a demographic. A person with a specific problem, at a specific level of awareness, who your onboarding can satisfy within one session.
- What you ask the platform to optimise for. Install, trial start or purchase. This single setting changes who Meta delivers to more than most targeting decisions do.
- What your monetisation model is. Hard paywall or freemium, trial or no trial, weekly, monthly or annual. This decides the shape of every conversion rate you will ever benchmark yourself against.
- What you are willing to pay, and by when. A spend ceiling with a payback window attached. Without the window, the ceiling is a guess.
- How much creative you can produce. This is a strategic constraint, not a production detail. It sets the ceiling on how much you can learn per month, and therefore on how fast the other four decisions can improve.
The funnel, and what 2026 data says about each stage
The table below is the current benchmark picture from the two largest public subscription datasets. Read the source and the definition column before you read the number. Every figure here has a different denominator, and comparing them as though they measure the same thing is the most common analytical error in this category.
| Funnel stage | What it measures | 2026 benchmark | Source | What controls it |
|---|---|---|---|---|
| Install to trial | Share of installs starting a free trial | 10.9% global average | Adapty 2026 | Ad promise, onboarding, paywall placement |
| Trial to paid | Share of trial starts converting to a paid subscription | 25.6% global average; 25.5% median on trials of 4 days or under, 42.5% on 17 to 32 day trials | Adapty 2026; RevenueCat 2026 | Trial length, product value delivered in trial, billing setup |
| Download to paid | Share of installs paying within 35 days | 2.0% global median; 2.9% health and fitness, 1.0% gaming; 10.7% hard paywall against 2.1% freemium | RevenueCat 2026 | The whole funnel, and your monetisation model |
| Revenue per install | Total revenue divided by total installs at day 60 | $0.34 global median; $0.66 health and fitness, $0.14 gaming; $0.55 North America, $0.11 India and South East Asia | RevenueCat 2026 | Everything above, plus price and geography mix |
| Renewal and retention | Subscribers still paying after the first cycles | Involuntary churn differs by store: 31% of Google Play cancellations trace to billing errors against 14% on the App Store | RevenueCat 2026 | Product value, billing recovery, plan length |
Two things in that table deserve more attention than they usually get. The first is the spread between health and fitness at 2.9% download-to-paid and gaming at 1.0%. If you are a fitness app benchmarking against a blended global median of 2.0%, you are congratulating yourself for being below your own category. The second is revenue per install. At a $0.34 global median at day 60, most subscription apps cannot profitably buy an install at a dollar inside two months, which is why payback windows and renewal assumptions do so much of the work in this business.
Why two credible benchmark reports look like they disagree
This is worth a section on its own, because it wastes an enormous amount of time in growth meetings.
RevenueCat's most quoted 2026 figure is that hard paywall apps convert at a median 10.7% against 2.1% for freemium. That number is download-to-paid within 35 days of install: the share of installs that result in at least one paid subscription inside that window. It is not a trial-to-paid rate, although a number of articles summarising the report have described it as one. If you take 10.7% into a meeting as your trial conversion target, you will conclude your paywall is broken when it may be performing normally.
Adapty's 2026 report, covering more than 16,000 apps and $3 billion in subscription revenue, gives a global average trial-to-paid of 25.6%. RevenueCat, segmenting the same broad stage by trial length, reports a median of 25.5% on trials of four days or under, 37.4% on five to nine day trials and 42.5% on trials of 17 to 32 days. The near-identical 25.6% and 25.5% are coincidence. One is an average across a whole dataset with its own mix of trial lengths and categories, the other is a median for a specific trial-length segment.
The practical rule: before you adopt any benchmark, write down its denominator, its measurement window and the dataset it came from. If you cannot state all three, you cannot use the number to make a decision. That discipline matters more as the reporting cycle gets noisier, and it is the same reason we insist on cohort reporting rather than platform-reported returns in our own accounts.
Stage one: acquisition, and what CPI is actually for
Cost per install is the most quoted and least useful number in subscription app marketing. It is useful for exactly one thing: explaining a change you have already observed further down the funnel. It is not a target, because it contains no information about whether the install was any good.
The failure mode is specific and common. A creative that leads with a curiosity hook, a trending audio, or a broad emotional appeal will usually win on install cost, because it attracts people who will tap out of mild interest. Those same people do not have the problem the app solves, so they do not start trials. Install cost falls, cost per trial rises, and the account looks like it is improving while the business gets worse. Our CPI benchmarks by vertical and geography are worth reading with that caveat attached, and lowering CPI is only ever a goal once you know downstream conversion is holding.
Geography compounds this. RevenueCat puts median revenue per install at day 60 at $0.55 in North America against $0.11 across India and South East Asia. Cheap installs in low-monetising markets are not a bargain, they are a different business. The same logic applies across platforms, which we cover in iOS versus Android cost per install.
Stage two: install to trial, which is rarely an ads problem
Adapty's 2026 data contains the single most useful operational fact in this article: 89.4% of trial starts happen on the day of install. Trial starts are effectively a first-session event. By category it ranges from 94.5% in entertainment down to 71.3% in education, but the direction is the same everywhere.
Three consequences follow. First, your onboarding and first paywall are doing almost all of the work, and any lifecycle campaign aimed at converting installs to trials on day three is fighting for a very small pool. Second, install-to-trial reads fast, so it is the earliest honest signal you have about creative quality. Third, when install-to-trial is weak, the ad account is usually not the thing to fix. The ad has made a promise that the first screen after install does not pay off, and the gap between those two things is where the users go.
That is the same problem we worked through with Steps & Beasts, where creative testing ran alongside onboarding changes rather than after them, and revenue rose 145% with active subscriptions up 118%. Treating the ad and the first session as one continuous piece of messaging is not a nice-to-have on a subscription app. It is the mechanism.
Stage three: trial to paid
Trial to paid is where product value and price meet, and it is the stage paid acquisition has least direct control over. The RevenueCat trial length data is striking: 25.5% median on trials of four days or under against 42.5% on trials of 17 to 32 days.
Resist the obvious conclusion. That correlation does not establish that lengthening your trial will raise your conversion rate. Apps that run long trials tend to be apps confident enough in their retention to offer one, often in categories where value takes weeks to become obvious. The trial length and the conversion rate may both be downstream of the same thing, which is how good the product is at proving itself. Longer trials also delay cash, which changes your payback maths and makes early return on ad spend harder to read.
What acquisition can control here is who arrives at the trial. A trial population recruited by an ad that named a specific problem converts differently from one recruited by an ad that promised novelty, even at identical trial length and paywall.
Stage four: renewal, which quietly sets your budget
Every additional renewal raises the amount you can rationally pay to acquire a subscriber. This is the least glamorous stage and the one with the most leverage over media budget.
One practical detail that is easy to miss: a meaningful share of churn is not a decision at all. RevenueCat found 31% of Google Play subscription cancellations trace to billing errors, against 14% on the App Store. On Android, recovering failed payments can be a larger lever than anything you do in the ad account, and it is a lever most paid teams never look at because it does not appear in their reporting.
Turning your funnel into a spend ceiling
Here is the calculation the whole strategy rests on. The numbers below are hypothetical and chosen for clarity, not taken from any client account. Substitute your own.
Take an app selling at £9.99 a month. Apple's standard terms return 70% of the subscription price during a subscriber's first year, rising to 85% after a year of paid service, or 85% immediately for App Store Small Business Programme members. Assume the standard first-year rate, so net revenue per payment is £6.99.
- Start with 1,000 installs.
- At a 10% install-to-trial rate, that is 100 trial starts.
- At a 35% trial-to-paid rate, that is 35 subscribers.
- Assume those subscribers pay for an average of 4.2 months. Net value per subscriber is 4.2 multiplied by £6.99, or £29.36.
- Total net revenue from 1,000 installs is 35 multiplied by £29.36, or £1,027.60.
- Net revenue per install is therefore £1.03.
If you are willing to spend half of net lifetime value to acquire, your allowable CPI is about £0.51, and your subscriber acquisition cost works out at roughly £14.68. Those are your two operating numbers. Anything the ad account does that keeps you under them is working.
Now change one input. Hold everything else and drop trial-to-paid from 35% to 25%. Subscribers fall to 25, net revenue per install falls to £0.73, and allowable CPI falls to about £0.37. A ten point movement in paywall conversion has cut the price you can pay for an install by roughly 29%. Your target subscriber acquisition cost has not moved at all. This is why paywall and onboarding work belongs in the marketing strategy rather than in a separate product backlog: the paywall sets the media budget.
Read that ceiling alongside a real payback window. Day 7 return on ad spend is the earliest honest checkpoint for most weekly and short-trial apps, and it needs reading against SKAdNetwork postback delay on iOS rather than at face value.
Which Meta optimisation event to run
Meta's app event optimisation lets you optimise delivery towards specific in-app events rather than installs, and its documented standard events include Start Trial and Subscribe. Meta describes the mechanism as reaching the people most likely to take the action you specify rather than the people most likely to install. Value optimisation, which optimises towards expected revenue, remains available on a limited basis to advertisers on an allow list rather than to everyone.
The trade-off is signal volume against signal quality. Optimising for purchases aligns delivery with the thing you actually want, but a purchase event that fires a few times a week gives the system very little to learn from, and the campaign will spend most of its life in the learning phase. Trial starts usually fire an order of magnitude more often, and because they land on install day they feed back quickly.
The pragmatic sequence for most subscription apps is to optimise for trial starts while volume is thin, then test moving deeper once the purchase event fires often enough to sustain learning. Avoid the two absolutist positions. Optimising for installs is not always wrong, and optimising for purchases is not always right. What is always wrong is picking one at launch and never revisiting it as spend changes.
Creative is the strategy, not the execution of it
Once the ceiling is set and the optimisation event is chosen, creative is the only lever with meaningful headroom left. It is also the one most frequently under-resourced, because it is treated as production rather than as the learning mechanism.
The distinction that matters is between concepts and assets. Twenty variations of one idea teach you almost nothing beyond which thumbnail performs. Five genuinely distinct psychological angles, each addressing a different reason someone might want the app, teach you which problem your market actually feels most sharply. That finding then improves your onboarding, your paywall copy and your positioning, not just your ad account. We have written about creative diversity against raw volume and about how creative fatigue actually presents in app accounts, and both feed this point.
With Oli Help, the work that moved trials up 1,750% and revenue up 394% was a reframing of who the ideal customer was and what they were actually worried about, expressed as distinct creative territories rather than more executions of an existing one. That is what a creative strategy is for on a subscription app: finding the angle, not filling a content calendar.
Where to intervene when a number slips
Most growth meetings fail because the whole funnel is discussed at once. This is the shortcut. Find the pattern that matches your account, and work on that stage only until it moves.
| What you are seeing | What it usually means | Where to work |
|---|---|---|
| Install cost is high, trial and paid rates are healthy | An acquisition problem. The funnel works, you are paying too much to fill it. | Creative concept diversity and audience expansion before bid or budget changes. |
| Install cost is low, install-to-trial is weak | A message match problem. You are buying attention the product does not pay off. | Align the ad promise with the first screen after install, not with the app's feature list. |
| Install-to-trial is strong, trial-to-paid is weak | A trial or paywall problem. Interest is real, perceived value at the point of payment is not. | Trial length, what the user achieves during it, and the paywall itself. Not the ad account. |
| Paid conversion is strong, subscriber CAC is still too high | A pricing or retention problem masquerading as an ads problem. | Plan mix, price, and renewal. Every extra renewal raises what you can afford to bid. |
| Everything looks fine, revenue is flat | Usually a measurement problem, or a cohort quality decline hidden by blended reporting. | Split new cohorts from renewals before touching the account. |
When you are ready to scale, and when you are not
Scaling a subscription app is not a budget decision. It is a decision about whether the next pound of spend will behave like the last one. Three conditions usually need to hold at once.
- Your cohorts are still paying back at the previous rate. Not blended revenue. New cohorts, tracked separately, at the same day markers as before.
- Creative supply exceeds fatigue rate. If you are recycling winners because nothing new is ready, more budget accelerates decline rather than growth. Our creative refresh calculator gives a rough read on how much output your spend level needs.
- Your funnel rates hold at higher volume. Broader delivery reaches less qualified people. Install-to-trial slipping as you scale is normal, and it lowers your ceiling. Recalculate rather than assume.
If those conditions do not hold, the honest move is to fix the binding one rather than raise budgets and hope averages carry you. Spending more through a funnel that is deteriorating produces a bigger version of the same problem.
Frequently asked questions
What is a subscription app marketing strategy?
It is the set of decisions that connect paid acquisition to subscription revenue: which users you buy, which event you optimise Meta towards, what your onboarding and paywall convert at, how long subscribers renew for, and therefore the maximum you can afford to pay for an install. A strategy that only covers channels and creative is a media plan, not a strategy. The number that matters is net revenue per install, because that is what sets your allowable cost per install.
What is a good conversion rate for a subscription app in 2026?
RevenueCat's State of Subscription Apps 2026, covering more than 115,000 apps, puts the global median download-to-paid rate at 2.0% within 35 days of install. That rises to 2.9% for health and fitness apps and falls to 1.0% for gaming. Apps using a hard paywall convert at a median 10.7% on the same measure, against 2.1% for freemium apps. Adapty's 2026 report, covering more than 16,000 apps, gives a global average of 10.9% install-to-trial and 25.6% trial-to-paid. These are different measurements, not competing answers to the same question.
Should a subscription app optimise Meta campaigns for installs, trials or purchases?
It depends on how much signal your account produces. Meta's app event optimisation lets you optimise towards standard events including Start Trial and Subscribe. Optimising deeper down the funnel aligns delivery with revenue, but only works when the event fires often enough for the system to learn from it. An app generating a handful of subscriptions a week will usually learn faster on trial starts than on purchases. The right answer changes as spend grows, so treat it as a decision to revisit rather than a setting to fix.
How much does a subscription app need to spend on Meta ads to grow?
There is no universal figure, because the budget follows the funnel rather than the other way round. The useful starting question is how many installs you need to produce enough trials to read a creative test, and what those installs cost. An app with a 10% install-to-trial rate needs roughly ten times the install volume of an app converting at 100% off a hard paywall to reach the same number of trials. Budget from the event volume you need to learn, then check the answer against your runway.
Why is my CPI low but my revenue flat?
Because cost per install measures the cheapest part of the funnel. Creative and targeting that win on install cost often win by attracting curious users rather than users with the problem your app solves, and those users do not start trials or convert on the paywall. If install costs fall while revenue per install falls faster, the acquisition is getting worse, not better. Judge creative on cost per trial or cost per subscriber, and only look at CPI to explain a change you have already seen downstream.
How long should a subscription app wait before judging paid acquisition?
Long enough for the cohort to actually convert, which is longer than most dashboards suggest. Adapty's 2026 data shows 89.4% of trial starts happen on the day of install, so trial volume reads quickly. Paid conversion does not, because it lands at the end of the trial, and Apple's SKAdNetwork postbacks add further delay on iOS. Day 7 return on ad spend is the earliest honest read for a weekly or short-trial app. Anything shorter is measuring the ad, not the business.
Working through this on your own app
Most of what is above you can do without hiring anyone. Calculate your net revenue per install, set a payback window you can actually finance, derive your allowable CPI, then find which stage of the funnel is the binding constraint before touching the ad account. Teams that do only that tend to make better decisions than teams running twice the budget without it.
Where we tend to be useful is the creative half, once the economics are understood and creative supply is the thing capping learning. Our consumer apps page sets out how we approach mobile app acquisition, the case studies show what that has looked like across fitness, parenting, language and music apps, and if your app is already spending and you want to check the commercials, our fees are published. If you are still deciding what kind of partner you need at all, our comparison of Meta ads agencies for mobile apps covers the field, including where we are the wrong choice.