Mobile app growth, from the App Store search index to the subscription renewal curve

Why mobile app growth is its own discipline

A web product's distribution surface is the open web. Google Search, direct navigation, paid media landing anywhere on the internet, social referral, and a browser that renders any URL. The operator of a web product owns the domain, the page, the checkout, the analytics, and the customer relationship end to end. Every optimization decision (SEO, CRO, paid, email, lifecycle) runs inside surfaces the operator controls.

A mobile app's distribution surface is the App Store and the Play Store. Every install is either a browse install on the store, a search install on the store, or a paid install that lands on a store product page and completes through a store native install button. The store owns the product page layout, the search ranking algorithm, the review surface, the pricing display, the subscription billing rails, and the analytics that report on all of the above. The operator of a mobile app does not own the top of the install funnel. The operator rents it, one product page at a time, from Apple and Google.

This is not a marginal difference. It is the difference that determines what a mobile growth team spends its time on. A web growth team ships SEO changes, tests landing pages, runs CRO on checkout, and moves on. A mobile growth team ships ASO changes, tests screenshots on Custom Product Pages, configures SKAN postback tiers, manages an MMP dashboard, negotiates review moderation with Apple, submits localized metadata in twenty markets, and does the equivalent of web work only after all of that store native work is in place. The operator who treats a mobile app like a website with a download button consistently underinvests in ASO and store native mechanics, and consistently loses to competitors who treat the store as the primary growth surface it actually is.

Three consequences follow. First, ASO is not a subset of SEO. It is a separate discipline with its own ranking factors, its own metrics, its own testing surface, and its own operating cadence. Second, paid UA on mobile is not a subset of paid social. It sits inside an attribution model (SKAdNetwork, SKAN 4, ATT consent, Google's Privacy Sandbox for Android, MMP normalization) that has no web equivalent, and it converts through store postbacks that arrive on delays with limited signal. Third, subscription monetization on mobile runs through StoreKit and Play Billing, which handle price localization, tax, chargebacks, and renewals differently than any web billing stack, and which take a platform fee that is a first order input to LTV modeling. Growth teams that ignore any of the three build a growth motion that looks reasonable in a dashboard and produces flat installs in reality.

ASO fundamentals: what the store search index actually rewards

App Store Optimization is the set of practices that get an app to rank in store search, that convert store impressions into installs, and that hold those installs into activated users. It has six primary levers on iOS and a parallel six on Android, and the levers move different amounts in different categories.

Keyword strategy is the ASO foundation

App Store search runs on a different index than Google search. Its query set is dominated by short intent queries (the app category name, competitor brand names, feature terms, use case terms) rather than long tail informational queries. Its ranking signals are the title, the subtitle, the 100 character keyword field on iOS, and the long description and short description on Android, weighted by download velocity, ratings, and behavioral signals from the specific query. Its user is already in the store with install intent, not browsing to learn.

The keyword research surface uses tools like AppTweak, Sensor Tower, data.ai (formerly App Annie), and MobileAction. The workflow is: build a candidate keyword list by category, use the tools to filter by search volume and chance to rank, cluster by intent, and prioritize the terms where the app has a legitimate product claim. Chasing high volume terms with no product claim wastes rank because ratings and behavioral signals downgrade the app on those terms over time. Chasing zero volume terms wastes metadata slots. The middle band (moderate volume, moderate difficulty, credible claim) is where ASO earns compounding gains.

Title and subtitle discipline on iOS

The title field is 30 characters and it is the single most weighted ranking signal on the App Store. It should include the brand name and one high value descriptive keyword. Every character not spent on brand or keyword is wasted. The subtitle field is another 30 characters and it should extend the keyword coverage with terms that were not in the title. The 100 character keyword field is hidden from users but visible to the App Store search index. Comma separated, no spaces, no repetition of terms already in title or subtitle (Apple deduplicates), no plurals of words already present (Apple stems), and no filler words. Localized metadata multiplies the keyword surface: each localization has its own title, subtitle, and keyword field indexed separately, and a well localized app effectively runs 30 to 40 parallel keyword slots across markets.

Play Store metadata is a different game

On Play Store the title is 30 characters and the short description is 80 characters, both visible to users. The long description up to 4000 characters is both visible and indexed with density signals similar to web SEO. Play Store rewards keyword density in the long description in a way App Store does not, but it penalizes obvious keyword stuffing that reads badly to a human. The Play Store's algorithm is closer to web SEO than App Store's algorithm is. The operating consequence: the long description has to be written for both a human reader and an algorithmic index, with the top of the copy carrying conversion weight and the body carrying keyword coverage.

App icon, screenshots, and video preview are conversion, not ranking

The icon, screenshots, and preview video do not move ranking directly. They move conversion rate from store impression to install, which is a downstream ranking signal because the store rewards apps with high conversion rate on the queries they surface on. The right way to think about creative on the store page is that it multiplies the traffic ASO metadata earned. A store page with strong metadata and weak creative gets a lot of impressions and a low install rate. A store page with weak metadata and strong creative gets few impressions and converts most of them. A store page with both wins the category.

Screenshot design has its own conventions. The first three screenshots are what most users see without swiping and they carry disproportionate weight. Each screenshot should communicate one benefit or feature legibly at thumbnail size, with a headline and a visual, not a densely annotated app screen. Portrait orientation is standard for most consumer apps; landscape is category native for games. Video preview autoplays muted in the store and needs to communicate value in the first three seconds without audio. Icon design is more constrained than most operators realize: it renders at a small size in the store search results and it has to be legible and category recognizable at that size, not clever at a large size.

Category selection and secondary category

Primary category is the category the app ranks in on Top Chart and Category browse. Secondary category on iOS gets a small additional visibility boost in that category's browse. The choice is more strategic than it looks. A productivity app that categorizes as Business might rank top 20 in Business but never appear in Productivity's more browsed chart. A fitness app that categorizes as Health and Fitness competes with the whole category; the same app in Lifestyle might rank higher in a less competitive chart. The right category is the one where the app can hold a defensible chart position, not the one that sounds most correct in the abstract.

Ratings and reviews as ranking signal

Store ranking is heavily influenced by rating (both the aggregate score and the recent trend) and by review volume. A 4.6 average with 50,000 recent reviews outranks a 4.7 average with 500 reviews on most competitive queries because volume signals that the app is functioning at scale. The operational job is to route rating prompts to satisfied users at the right moment (usually after a value moment, not at first launch), to catch dissatisfied users into a support flow before they leave a public review, and to respond to negative reviews publicly with a fix or a workaround.

The rate limiting rules matter. iOS SKStoreReviewController allows a maximum of three prompts per user per year. Play Store In App Review has a similar quota. Firing the prompt at the wrong moment burns one of the limited chances and leaves the user with a bad taste. Firing it after a first successful use, a completed workout, a successful export, a satisfying interaction, dramatically outperforms firing it at app launch or after an error.

App Store Connect and Play Console: the metrics that predict growth

Both stores expose native analytics that most growth teams underuse. App Store Connect surfaces impressions, product page views, conversion rate, redownloads, subscription events, and retention curves. Play Console surfaces the same categories with different naming and different aggregation windows. Both are the ground truth for what the store index and store product page are actually doing. Third party tools infer, model, and reformat. The store consoles report.

Impressions, product page views, and the two step funnel

Store impressions is the count of times the app appeared in front of a user (search results, browse rows, Today tab features, category charts). Product page views is the count of times a user tapped an impression and landed on the product page. The ratio between the two is a legibility signal: how compelling was the icon, title, and star rating at the impression surface. Conversion rate from product page view to install is the ASO conversion metric, and it is what screenshots, video, and description move.

Reading the two step funnel is diagnostic. High impressions and low tap through means the impression surface listing is not compelling (icon, title, first screenshot). High tap through and low install conversion means the product page itself is not closing (screenshots deeper in the row, description, video). Low impressions across the board means ASO metadata is not ranking. Each of the three has a different fix and conflating them wastes optimization cycles.

Impression source segmentation matters

App Store Connect segments impressions by source: search, browse, referrer (paid campaigns with campaign IDs), and Today tab. The mix tells the operator where growth is actually coming from. A search heavy app is winning ASO. A browse heavy app is winning charts and category rank. A referrer heavy app is dependent on paid UA. A Today tab heavy app got a one time feature and needs to convert that into durable ranking before the feature drops off. Growth teams that report a single install number miss the source shift that predicts the next month's curve.

Redownloads and reinstalls

Redownloads (a user who previously installed, deleted, and installed again) are a signal about brand strength and about the app's memorability. High redownload rate on an app in a discretionary category (fitness, productivity, meditation) usually means the app is being downloaded and abandoned repeatedly, which is a retention problem masquerading as an acquisition win. Redownloads on a utility app (weather, calculator, banking) are neutral. The metric is category dependent and reading it correctly requires understanding what the app is for.

Subscription events

Subscription apps get a dedicated set of metrics: trial starts, trial cancellations, trial to paid conversions, paid subscription starts, active subscribers, subscription cancellations, refunds, and (with recent iOS versions) the categorized cancellation reason. Every one of them is diagnostic. Trial to paid conversion rate is the single most predictive subscription metric. Trial cancellations that happen in the first 24 hours signal a paywall or onboarding mismatch: the user started the trial in a moment of intent and cancelled the moment they saw what the trial actually gated. Trial cancellations that happen near the end of the trial signal a value moment that never landed. Both patterns have different fixes.

Retention curves inside the store consoles

App Store Connect and Play Console expose first year retention curves at the cohort level. Day 1, Day 7, Day 30, Day 90, and beyond. The shapes are diagnostic. A curve that drops steeply from Day 1 to Day 7 and then flattens is an onboarding problem: users are trying the app once and not finding a reason to return. A curve that holds through Day 7 and drops between Day 30 and Day 90 is an engagement or value problem: users found initial value but did not build a habit. A curve that holds through Day 90 and drops at Month 3 is a subscription renewal problem, not a retention problem in the general sense. Reading the shape locates the fix.

Custom Product Pages, A/B testing, and Product Page Optimization

Both stores now support testing on the product page. On iOS, Custom Product Pages let the operator create up to 35 alternate product page variants, each with different screenshots, video preview, and promotional text, each with its own URL. Product Page Optimization lets the operator run A/B tests on the default product page's icon, screenshots, and preview video with statistical readouts inside App Store Connect. Play Store has parallel functionality through Store Listing Experiments.

What Custom Product Pages are actually for

The primary use of CPPs is not standalone landing pages. It is matching paid campaign creative to a product page that carries the same message. If the paid ad promises weight loss coaching, the CPP the ad points to should show weight loss coaching screenshots first, not general fitness. If the paid ad promises language learning through short lessons, the CPP screenshots should show short lessons visually, not the app's full feature surface. The install rate lift from CPP to matched creative is typically in the 15 to 30 percent range against the default product page, and it compounds with the ROAS math on paid channels.

What Product Page Optimization is actually for

PPO is for testing icon variants, screenshot variants, and video preview variants on the organic store traffic. A test surfaces the winning variant to a share of users, reports statistical significance based on install conversion inside the console, and lets the operator promote the winner to the default. The tests take real time (typically two to four weeks to reach significance on a mid volume app, longer on a small one) and the traffic split reduces total conversion during the test window, so the operator has to prioritize tests that are worth the tax.

Sample size, significance, and the pitfalls

Statistical significance on a store A/B test needs enough impressions and installs to move above noise. Most ASO tools recommend a minimum of 1000 to 2000 impressions per variant per day and running the test until 95 percent confidence is achieved on the primary metric (install rate). Below those thresholds the tests report noise as signal and the operator makes changes based on random variation. Small apps often cannot reach significance on the store's native test infrastructure and have to run sequential testing (change, wait, measure, revert if worse) instead of parallel testing.

The most common pitfall is testing too many variables at once. A test that changes the icon, the first screenshot, and the video preview simultaneously reports an install rate delta that cannot be attributed to any one change. The disciplined operator tests one variable at a time and sequences the tests. Another pitfall is running tests during known seasonality (Black Friday, New Year for fitness apps, back to school for edtech) where the underlying traffic mix distorts the readout. Tests should run during representative traffic conditions or they should be repeated at multiple seasonal windows.

Paid user acquisition on mobile: channel by channel

Mobile paid UA is a stack, not a channel. Different channels do different things and a durable growth motion runs them in complement rather than substitution. The order below is roughly the priority most subscription apps assign, with variation by category.

Apple Search Ads

Apple Search Ads is the paid placement above organic results in App Store search. It is the highest intent channel in mobile UA because the user is already in the store searching for something and the tap goes to the product page immediately. It has three campaign types: Search Match (Apple's own targeting on discovery keywords), Keyword based (operator selected keywords with bids), and Discovery campaigns for exploration.

Bidding is per tap (cost per tap, or CPT) not per install. The install conversion happens on the product page after the tap. The metric that matters is cost per install (CPI) and, one level deeper, cost per activated user or cost per trial start on a subscription app. Bidding strategy depends on category: brand keyword defense (bidding on your own app name to keep competitors from placing above you) is table stakes and usually the cheapest CPI in the account. Competitor keywords are more expensive but produce net new users, and the value depends on how well your product page converts users who searched for the competitor. Category keywords (the top intent terms for the category) are the most expensive because every serious app in the category is bidding on them.

ASA's data advantage is LATAT (Limited Ad Tracking and Attribution Tracking) which is Apple's own deterministic attribution inside its own ad system. Because Apple runs both the ad and the store, ASA attribution does not depend on SKAN or ATT consent. Every tap and every install is measurable at the campaign, ad group, and keyword level. That is a significant advantage over channels that depend on SKAN postbacks for iOS install attribution.

Google App Campaigns

Google App Campaigns runs installs and in app events across Google Search, YouTube, Play Store, Discover, and the Google Display Network. The bidding model is Target CPI (bidding to a cost per install target), Target CPA on in app events (bidding to a cost per action target like a trial start), or Target ROAS on in app purchase apps (bidding to a return on ad spend target).

The mechanics of AC campaigns require creative asset diversity. Google's machine learning explores which creative combinations work on which placements, and campaigns fed with a wide asset library (multiple text lines, image assets in multiple aspect ratios, video assets of multiple lengths) outperform campaigns with a small library. The operational job is producing enough creative volume to feed the algorithm without letting the creative go stale. Most durable AC accounts refresh a portion of the creative library every two to four weeks.

On iOS, AC installs attribute through SKAN, which means the operator cannot see user level data and cannot bid at the individual level. The SKAN conversion value schema needs to be configured carefully to send back the right postback signal (trial start, first key action, subscription conversion) so that Google's bidding algorithm learns which installs are actually valuable.

Meta app ads

Meta remains one of the largest paid channels for consumer mobile apps despite the ATT era hit to iOS signal. Facebook and Instagram feed placements, Reels placements, and Stories placements all support app install and app event optimization. The creative format that works varies by placement: static creative for feed, short vertical video for Reels and Stories, carousel for feature demonstration. Meta's own machine learning also benefits from creative diversity.

The ATT era changed Meta's mobile attribution meaningfully. iOS installs attributed through Meta now flow through SKAN with the delay and signal limitation SKAN imposes. Aggregated Event Measurement is Meta's SKAN adjacent framework that requires event prioritization: the operator picks up to eight events per app and ranks them by importance, and Meta's bidding uses that priority to allocate. Getting the ranking right (usually: subscription purchase highest, then trial start, then key onboarding action, then generic install) is what makes iOS campaigns bid to value rather than to volume.

TikTok app ads

TikTok is now a large channel for consumer app UA, particularly for categories where the audience is younger and video native (fitness, dating, gaming, social, edtech). TikTok's ad formats mirror the organic feed, which means the creative that works looks native to TikTok rather than looking like a repurposed brand ad. UGC style, creator style, and vertical video with sound on are the formats that convert. TikTok's targeting is more interest and behavior based than demographic, and its bidding supports cost per install, cost per action, and value based optimization.

The creative volume requirement on TikTok is higher than on any other channel. Creative fatigue happens fast (typically within one to two weeks of a hit ad) and refreshing the library requires either an in house creator studio or an ongoing creator partnership program. Apps that ship five to ten fresh creative variants per week on TikTok generally outperform apps that ship one or two, even at the same total spend.

Emerging channels: Reddit, Spotify, Snap, Pinterest

Reddit ads work well for niche category apps where the target audience concentrates in specific subreddits. Spotify audio ads work well for fitness, meditation, and health apps where the audience listens to Spotify during the use moment for the app. Snap works well for the same younger demographic TikTok reaches, with a different creative style and a lower CPM. Pinterest works for category apps where visual discovery is the buying behavior (home, wellness, personal finance). None of the four is usually a foundation channel; each of them can be a meaningful diversification layer at 5 to 15 percent of the total UA budget.

Weighting the channels

A durable growth stack typically runs Apple Search Ads as a foundation (highest intent, deterministic attribution, defends the brand keyword), Google App Campaigns and Meta as scale channels (largest available volume, ATT era attribution challenges), TikTok as a creative volume channel (fast learning, high creative refresh requirement), and one to two emerging channels for diversification and net new discovery. The weighting shifts with category and with the app's stage. Early stage apps often lean heavier on ASA and TikTok because those two produce the fastest signal for the fewest dollars. Scale stage apps have to diversify because ASA volume is bounded and TikTok creative refresh becomes operationally taxing at large budgets.

The attribution stack in the ATT era

Attribution on mobile is the discipline that determines which paid dollar produced which install and which install produced which downstream event. It sat quietly in the background for years, powered by the IDFA (Identifier for Advertisers) on iOS and the GAID (Google Advertising ID) on Android. The IDFA was passed to ad networks with every install, deterministic attribution was possible, and MMPs (Mobile Measurement Partners) like AppsFlyer, Adjust, Kochava, Branch, and Singular normalized the signal across networks. Then iOS 14.5 shipped ATT.

ATT and what it changed

App Tracking Transparency is the iOS system prompt that requires user consent before an app can access the IDFA. Consent rates vary by category but consistently land in the 25 to 40 percent range across published measurements. That means for 60 to 75 percent of iOS users, the IDFA is not available and deterministic third party attribution is not possible.

The consequence is that iOS attribution moved from user level and deterministic to aggregate and probabilistic. SKAN is the mechanism Apple provides to replace it. SKAN sends a postback per install with limited data (source app or ad network ID, a campaign ID with restricted cardinality, and a conversion value payload with 6 or 8 bits depending on SKAN version) on a delay measured in hours to days depending on postback tier. It does not expose user level data and it does not allow real time optimization at the user level.

SKAN configuration and conversion values

The conversion value payload is where operator design matters most. SKAN 3 supports a 6 bit conversion value (64 states). SKAN 4 supports 6 bit coarse values plus source identifiers with more granularity, spread across three postback windows. The design job is to encode into that limited payload the events that predict downstream value: install completed, trial started, key onboarding event, subscription conversion, key retention event. A conversion value schema that only reports install produces a bidding signal that maximizes installs regardless of value. A conversion value schema that reports trial start and subscription conversion produces a bidding signal that maximizes revenue. Most durable subscription apps invest a substantial engineering block into their conversion value design because it is the biggest lever on iOS UA efficiency.

MMP integration

AppsFlyer, Adjust, Kochava, Branch, and Singular are the major MMPs. Their job is to receive postbacks from all ad networks, receive SKAN postbacks from Apple, normalize the signal into a single dashboard, deduplicate installs, and provide attribution reports at the campaign, ad set, and creative level to the extent possible under ATT constraints. The MMP is also the source of truth for iOS SKAN and for Android GAID deterministic attribution.

Choosing an MMP is usually a two year decision because migration is operationally heavy (SDK integration, historical data continuity, campaign UTM parameter conventions). The differences between the top MMPs are marginal at the tier a durable subscription app operates at. The bigger question is how the MMP integrates with the ad platforms the app spends on and how well the operator's team is trained on the MMP's dashboard.

Media mix modeling to fill the gaps

Because SKAN cannot fill the full attribution picture on iOS, cohort level MMM (media mix modeling) has become a required layer for any app spending meaningful iOS dollars. MMM regresses total installs and total revenue against ad spend by channel by week, controlling for seasonality and other factors, to produce an incremental lift estimate per channel that SKAN alone cannot deliver. It is coarser than user level attribution and it is slower to react to changes, but it captures signal that SKAN blocks. The best mobile growth teams run both: SKAN plus MMP for tactical bidding and creative decisions, MMM for strategic allocation and channel weighting.

Android is different

Android still allows GAID access without a consent prompt equivalent to ATT, which means deterministic attribution is still available on Android for most users. Google's Privacy Sandbox for Android is the coming change: it introduces user opt in mechanisms and aggregated reporting similar in shape to ATT and SKAN, and it will roll out over years rather than in a single release. The operational planning position is to assume that Android attribution will eventually look similar to iOS, and to build the SKAN style conversion value discipline into Android measurement now rather than after the Privacy Sandbox rollout forces it.

Subscription mobile app economics: trial to renewal

Subscription apps have become the dominant monetization shape in most consumer mobile categories that are not games. Fitness, health, meditation, productivity, dating, streaming, education, utility. The economics are more forgiving than one time purchase and more punishing than freemium ad supported, and the operating disciplines are specific.

Trial length and its trade offs

Apple and Google both support introductory offers including free trials. The most common trial lengths are 3 day, 7 day, and 14 day. Each has trade offs.

3 day trials produce the fewest trials started because users perceive the shorter trial as less useful, and they convert at higher rates because users who started have real intent. 7 day trials are the most common category default because they align with the way most users evaluate a subscription and because they land the first billing event before the user forgets they started. 14 day trials produce more trials started, particularly in categories where the value moment takes several sessions to arrive (learning apps, habit apps), and they convert at lower rates with more forgetting churn where the user simply forgets to cancel and files a chargeback later.

The right answer is category specific and it should be tested. Directional bands from published measurements: healthy trial to paid in a subscription fitness app runs 15 to 25 percent on a 7 day trial. Productivity apps often run 20 to 40 percent on a 7 day trial. Meditation and mental wellness apps run 20 to 35 percent. Dating apps where subscription is optional on top of a free tier often run 10 to 20 percent. Language learning and edtech vary widely by product design. The operator's job is to test trial length paired with paywall design and to instrument the reason for cancellation so the shape of the churn is legible.

Paywall design: hard versus soft, and the placement question

A hard paywall is placed before the user gets to use the app: an install lands, the user completes an onboarding flow, and the paywall gates the first real use. Duolingo Super Duolingo variants, Calm, Cash App investing tiers, and many streaming apps use hard paywalls. Hard paywalls produce the highest trial to paid conversion on the users who convert, and they filter out users who would have installed but not paid.

A soft paywall is placed after the user has experienced some value in the app: several free actions, a free tier that gates premium features, or a free session that shows the app's capability before asking for a subscription. Most fitness apps, most productivity apps, and most dating apps use soft paywalls. Soft paywalls produce lower trial to paid conversion in aggregate but higher install to activated user rates.

The right answer is category specific and product specific. The universal discipline is to test the paywall placement, the paywall design (copy, feature list, price display, plan comparison, testimonial density), and the paywall trigger (which event surfaces the paywall) with the same rigor as any other high value conversion surface. Paywall variant testing lifts trial to paid conversion in the 5 to 20 percent range for a typical subscription app, which compounds directly to LTV.

Price testing and price localization

Price is a first order lever on subscription revenue and the store rails make price testing more constrained than web billing. Apple and Google both support price tiers (currently 90 plus tiers on Apple, similar structure on Google) and both allow price by market. Changing a price tier applies to new subscribers only; existing subscribers stay at their grandfathered price unless the operator opts them into the new price.

Price testing on the App Store previously required creating parallel StoreKit products with different price points and randomly assigning users. Recent Apple changes have made A/B testing on prices somewhat easier but the test still runs on new subscribers and takes longer to reach significance than a paywall design test does. The operating discipline is to test price at meaningful intervals (usually annually or semi annually), to test in isolated segments (by geography, by acquisition source, by acquisition cost tier) so that the test does not distort the aggregate revenue readout, and to consider the LTV impact rather than the trial start impact alone. A higher price that lowers trial starts by 20 percent but raises revenue per trial started by 40 percent is a win.

Price localization is the discipline of setting the right price in each market. A US price of $9.99 monthly is not the same value in Brazil or India, where local purchasing power supports a fraction of that price. Most successful global subscription apps run localized price ladders (typically 3 to 6 pricing tiers globally) that respect purchasing power in each market. The revenue lift from correct localization versus a flat global price is often in the 15 to 30 percent range on international revenue.

LTV modeling on a subscription app

Subscription LTV is the projected total revenue from a subscriber over their expected lifetime, discounted for time value, minus the platform fee, minus the payment processing already included in the platform fee. The category typical subscriber lifecycle is 6 to 18 months for most consumer apps, with longer tails on utility and highly retentive categories (banking, health tracking, family planning) and shorter tails on transactional or intent driven categories (dating during a search period, fitness during a New Year push, edtech during a defined course).

The right way to model LTV is cohort based, not blended. Take the cohort of users acquired in a specific month, track their subscription revenue by month, and extrapolate the curve to a defined horizon (12 months, 24 months, lifetime). The blended LTV number that most dashboards report averages long term customers and new customers together and hides the trend. Cohort by cohort LTV curves show whether the current cohorts are trending better or worse than prior cohorts, which is the leading indicator of whether the business is compounding or eroding.

Apple's platform fee is 30 percent for the first year of a subscription and 15 percent for subscribers who renew past year one. Small Business Program applicants pay 15 percent from the start on revenue up to $1M annually. Google's fee is 15 percent for the first year of a subscription and remains 15 percent thereafter for most subscription apps. The platform fee is a first order input to LTV and it makes retention past year one meaningfully more valuable per subscriber than year one revenue would suggest.

Retention curve mechanics for mobile subscription

Retention is where mobile subscription apps live or die past the acquisition phase. The metric matters at multiple time horizons and each horizon has its own diagnostic value. The apps that win their categories over five years have retention curves that shift up cohort by cohort. The apps that die have retention curves that quietly shift down while paid UA hides the erosion.

Install retention: Day 1, Day 7, Day 30, Day 90

Install retention is the percentage of users who installed the app and opened it again on the specified day. It is a measure of onboarding quality and initial value delivery. Directional bands for a healthy consumer subscription app:

• Day 1 install retention: 40 to 60 percent. Below 40 percent typically means the onboarding is confusing, the first session did not deliver value, or the app is broken for a segment of the acquired traffic.
• Day 7 install retention: 20 to 30 percent. This is the point where habit or intent has to be forming. Below the band signals a retention problem that will show up as thin subscription conversion.
• Day 30 install retention: 10 to 15 percent. At this depth the users who remain are the users who found ongoing value.
• Day 90 install retention: 5 to 12 percent. This is roughly the population that becomes long term subscribers or long term active users.

Category matters. Gaming has different curves than health. Utility apps have higher long term retention than entertainment apps because the user opens them for a specific task. Social apps have higher retention when the network effect is real and lower retention when it is not. The operator's job is to know the benchmark for the category and to track the cohort by cohort trend rather than a single top line number.

Subscriber retention: Month 1, Month 3, Month 6, Month 12

Subscriber retention is the percentage of users who converted to paid subscription and are still paying on the specified month. It is a different metric than install retention and it lives on a longer time horizon. Directional bands for a healthy consumer subscription app:

• Month 1 subscriber retention: 90 to 95 percent. Almost everyone who paid the first bill pays the second bill if the price is not shockingly high and the app is not broken.
• Month 3 subscriber retention: 70 to 80 percent. This is the first meaningful decision point where users evaluate whether the app is worth the price.
• Month 6 subscriber retention: 55 to 70 percent.
• Month 12 subscriber retention: 40 to 55 percent.

Above the band on year one retention is a sign of exceptional product value or of pricing that is below the willingness to pay. Below the band is a sign of a churn problem that has to be diagnosed by cancellation reason, cohort, and acquisition source.

The churn diagnostic when trial to paid is fine but renewal is bad

A common pattern that trips up growth teams: trial to paid conversion is healthy (users are willing to pay), but subscriber retention past month 3 or month 6 is weak. The diagnosis is usually one of three things. First, the paywall converts users who are not the target user for the ongoing product experience, and they churn once the initial motivator fades. Second, the app has a strong first month value proposition and a weak ongoing value proposition, which means the operator has to invest in ongoing value delivery (new content, new features, community, coaching). Third, the price is above the ongoing willingness to pay: users paid once because they were motivated, then decided the ongoing value is not worth the price and cancelled at the first opportunity.

The cancellation reason survey (Apple lets subscribers indicate a reason when they cancel, and the operator can also collect this in a save flow inside the app) is the diagnostic tool. Reasons cluster into: too expensive, not using it enough, missing feature, technical issue, found alternative. Each cluster has a different fix and treating them as one homogeneous churn number wastes optimization effort.

The save flow

Most subscription apps invest in a save flow: a set of screens presented when the user initiates cancellation that offer a discount, a pause, a plan downgrade, or a feature reminder before the cancellation completes. Save flows recover 10 to 25 percent of would be cancellations for most consumer subscription apps. The design constraint is that Apple and Google both restrict what can be shown at the cancellation surface (the operator cannot block cancellation, cannot bury the cancel option, and cannot make the flow deceptive). The save flow that works respects the constraint and offers a real alternative rather than trying to prevent the cancellation through friction.

Lifecycle marketing inside mobile apps

Lifecycle marketing on a mobile app is the set of communications and in app experiences that guide the user from install to activation to habit to renewal. It runs across push notifications, in app messaging, email, and in some categories SMS. It is the difference between an app that acquires users at a healthy CAC and churns them at a matching rate, and an app that acquires the same users and compounds them into long term subscribers.

Push notification strategy under permission constraints

iOS requires explicit user permission to send push notifications. Android used to grant them by default; Android 13 and later now require permission similar to iOS. Opt in rates vary by category but consistently land in the 40 to 70 percent range on iOS with a well timed prompt. The timing of the prompt is what determines the opt in rate. Prompting at first launch produces low opt in rates because the user does not yet know what the app is for. Prompting after a value moment (a completed workout, a successful matching, a saved item, a first content session) produces materially higher opt in rates. Prompting through an explanatory prime screen before the OS prompt (educating the user on what push will be used for) further improves opt in rates.

Once opted in, the strategy is category specific but the discipline is universal: push notifications are a scarce channel and each send has to earn its position. Notifications tied to user activity (a workout streak, a match request, a message received, a scheduled reminder the user set) consistently outperform generic broadcast notifications. Notifications sent at high frequency drive opt out and app deletion. Most durable apps land in the 2 to 5 notifications per week range for engaged users, with the frequency dropping for less engaged users and rising for highly engaged ones.

In app messaging

In app messages appear inside the app during a session, either as full screen takeovers, modals, tooltips, or persistent banners. They are the highest engagement lifecycle channel because they reach users who are already in the app and open. The tools that ship in app messaging include Braze, Iterable, Airship, OneSignal, and Clevertap. The category native use cases are onboarding walkthroughs, feature discovery for existing users, subscription upgrade prompts, save flow variants, and educational content tied to user state.

The pitfall is over messaging inside the app. A user who opens the app to complete a task and gets three modals blocking the task deletes the app. The disciplined approach is to treat in app messages as targeted rather than broadcast: user segment plus behavior trigger plus rate limit, so that any given user sees at most one or two in app messages per session and each message is directly relevant to what the user is doing.

Email lifecycle for subscription retention

Email is often underused on mobile apps because the growth team thinks in app first. Email is the channel that reaches users who have not opened the app in a while, users who did not opt in to push, and users approaching subscription renewal on a plan they might not remember starting. The retention lift from a well designed email lifecycle for a subscription app is often in the 5 to 15 percent range on month 3 and month 12 retention.

The email lifecycle for a subscription app typically includes: onboarding sequence (first three to five sends after install to reinforce value), engagement sequence (weekly or biweekly emails during the active subscription period), win back sequence (users whose engagement has dropped), pre renewal sequence (reminders and value reinforcement before annual renewal for annual plans), and reactivation sequence (users who have cancelled). Each sequence has different design constraints and the operator has to run them as separate programs.

Subscription management stack

RevenueCat is the dominant subscription management infrastructure for mobile apps, providing a normalization layer across StoreKit and Play Billing, offering a receipt validation service, exposing subscription events to downstream tools, and providing an admin dashboard for subscription state. Chargebee, Adapty, and Purchasely provide similar functionality with different feature emphases. The subscription management layer is where most lifecycle tooling integrates (Braze, Iterable, Amplitude, Mixpanel) because it is the source of truth for who is subscribed and what state they are in.

Building subscription management in house is possible and it is what larger apps at scale often do, but for most apps the build cost and the ongoing maintenance cost of tracking Apple and Google subscription state changes across every edge case (renewals, refunds, grace periods, billing retry, plan changes, family sharing) is materially more expensive than the platform fee a subscription management tool charges. The build versus buy decision usually tilts to buy until the app is at scale where the platform fee itself justifies internalizing.

Creator and influencer marketing for mobile apps

Creator and influencer marketing has become one of the most important paid channels for mobile app growth, particularly in categories with a native video audience. It runs across TikTok, Instagram Reels, YouTube Shorts, and (for gaming) Twitch and YouTube long form.

Why creator content outperforms brand content

The creative that works on TikTok, Reels, and Shorts is native to the platform: vertical, sound on, first person voice, casual production, and a specific point of view. Brand produced content that mimics the format usually reads as an ad and underperforms. Content produced by creators who already speak in the format's native voice outperforms because it reads as recommendation from a peer rather than as advertisement from a brand.

The mechanics are: identify creators in the app's category, brief them lightly (positioning, key messages, hard constraints), let them produce in their own voice, and use the resulting content both as organic posts on the creator's channel and as whitelisted content for paid amplification. The best programs invest as much in the paid amplification of creator content as they do in the creator relationship itself, because the paid amplification is what turns a single organic post reaching 100,000 native audience members into a paid asset reaching millions of targeted lookalikes.

The three creator commercial models

Flat fee: the creator is paid a fixed amount for one or more pieces of content. Predictable cost, predictable deliverable, no downside if the content flops. This is the most common model at the small and mid tier creator level.

Performance based (affiliate): the creator receives a revenue share on installs or subscriptions attributed to their content, often through a unique promo code or a tracked link. Aligns incentives with performance, requires an attribution mechanism the creator trusts, and generally attracts creators with confidence that they can move volume.

Hybrid: a modest flat fee plus a performance component. Reduces creator downside while preserving performance upside. This is increasingly the standard model for creator programs at scale.

Whitelisting and Spark Ads

Whitelisting is the arrangement where the creator grants the brand permission to run paid ads from the creator's account, so that the amplified content appears as a paid ad from the creator's handle rather than from the brand's handle. TikTok's Spark Ads is the platform native mechanism for this; Meta has a parallel feature called Partnership Ads. Whitelisted creator content consistently outperforms brand handle content on the same creative asset, often by 30 to 60 percent on install rate, because the audience trusts the creator's handle more than a brand handle they have never followed.

Program operations

Running a creator program at scale requires operational infrastructure: a creator sourcing pipeline (through platforms like Whalar, Grin, Aspire, or through direct outreach), a briefing and contracting workflow, a content review process that respects creator voice, an attribution setup (promo codes, tracked links, whitelisting), a paid amplification budget separate from the creator fee budget, and a post campaign performance review that informs which creators to work with again. Programs that treat creators as one off transactions produce inconsistent results. Programs that build ongoing relationships with a roster of tested creators produce compounding results because the creator's audience gets repeatedly exposed to the brand and the creator's understanding of the product deepens over time.

Common failure modes and the fix

Every failure mode below has killed real mobile app growth curves at real companies. Naming them here so operators know what to avoid.

1. Chasing installs without measuring subscription conversion

Symptom: install volume is growing, cost per install is stable or improving, subscription revenue is flat. The paid team is optimizing to a metric (install) that is not the value metric (subscription). Fix: instrument the SKAN conversion value schema, the MMP event tracking, and the ad platform bidding to trial start and subscription conversion, not to install. Move the paid team's target metric from CPI to cost per trial start or cost per subscription. Report and manage growth to the LTV to CAC ratio, not the install cost alone.

2. Over indexing on Apple Search Ads without diversifying

Symptom: ASA is producing 60 to 80 percent of installs, ASA CPI is creeping up, and the paid mix has no scale channel to absorb the incremental budget when ASA saturates. Fix: build a durable stack that includes Google App Campaigns, Meta, and TikTok as scale channels alongside ASA. Diversification is not just risk management; it is capacity building for when ASA hits its bounded volume ceiling.

3. Ignoring ASO because the paid team owns growth

Symptom: paid UA is spending against a store page that has not been ASO optimized in months or years. The paid campaigns are effectively paying to send traffic to a store page that converts at half the rate a well optimized page would. Fix: staff ASO explicitly, either as a dedicated role or as a cross functional responsibility between the paid team and the growth team. Test screenshots, video, metadata, and keywords on a regular cadence. Treat the store page as the highest leverage conversion surface it actually is.

4. Running trials too short for the product's actual value moment

Symptom: trial starts are healthy, trial to paid conversion is weak, and the cancellation reason data shows users saying they did not use the app enough during the trial to know if they wanted to keep it. Fix: match trial length to the product's value moment. If the value moment lands after five sessions, a 3 day trial is too short. If the value moment lands after the first session, a 14 day trial is longer than needed and produces forgetting churn. Test trial length against actual usage patterns.

5. Price testing without segment isolation

Symptom: a price test is running across all traffic, and the readout is confounded by acquisition source mix (high intent ASA users convert differently than lower intent Meta users), by geography, and by seasonality. The test either reports a false positive or a false negative and the operator makes a bad pricing decision. Fix: isolate price tests by segment. Test a new price in one market first, or on one acquisition source, or on one cohort, before rolling to the full population. Run the test long enough to hit statistical significance in the isolated segment.

6. Treating iOS and Android as one channel

Symptom: campaigns, creative, and conversion value schemas are shared across iOS and Android. The Android side underperforms because the SKAN driven conversion value design does not apply to Android, and the iOS side underperforms because Android's more permissive attribution is not being exploited. Fix: run iOS and Android as separate campaigns with separate creative, separate optimization targets, and separate MMP event schemas. Share what can be shared (creative assets, high level strategy) and separate what has to be separated (attribution setup, bidding targets, campaign structure).

7. Ignoring SKAN postback configuration and losing signal

Symptom: the SKAN conversion value schema was set up at launch and never revisited, or it is misconfigured such that valuable events are not being encoded into the postback. Ad platform bidding is optimizing to the wrong signal because SKAN is telling it the wrong thing. Fix: audit the SKAN configuration quarterly. Update the schema as the product evolves. Coordinate the SKAN schema with the ad platform's bidding requirements (Meta's AEM ranking, Google's SKAN event allowlist) so the signal that reaches bidding is the signal the operator wants bidding to optimize against.

8. Treating cancellation as a compliance detail rather than a retention lever

Symptom: the cancellation flow is a single button that ends the subscription and no save flow is offered. Cancellations that could have been recovered walk away. Fix: build a save flow inside the app or as an in app message triggered by cancellation intent. Offer a real alternative (discount, pause, plan downgrade, feature reminder) that respects the user's choice while giving them a reason to stay. Track the save rate as a KPI and iterate on the flow like any other high value conversion surface.

9. Shipping a paywall that converts on cohort A but has never been tested on cohort B or C

Symptom: the paywall was designed and validated on the traffic mix that existed when it was launched. The traffic mix has since shifted (new markets, new acquisition sources, new creator content) and the paywall no longer matches what the new traffic converts on. Trial to paid conversion is quietly eroding. Fix: revisit the paywall on a regular cadence, test variants against the current traffic mix, and consider running per cohort paywalls (different paywalls for different acquisition sources) if the diversity of traffic warrants it.

10. Underinvesting in creative volume on the video native channels

Symptom: TikTok and Reels campaigns are running a small set of creative that is being repeatedly served to the same audience. Creative fatigue is degrading CPI. The paid team asks for more creative and the brand team ships one asset per month. Fix: build a creator content pipeline that produces five to ten fresh assets per week. Use whitelisted creator content as the primary source. Treat creative volume as a first order growth input rather than a downstream deliverable from a slow brand process.

11. Reading LTV blended rather than by cohort

Symptom: the LTV number in the dashboard averages tenured subscribers with new subscribers, and it looks stable while newer cohorts are actually retaining worse than older cohorts. The business is quietly eroding underneath a flat top line number. Fix: report LTV by cohort. Track the trend cohort by cohort. Any cohort that retains meaningfully worse than the prior cohort is a signal that something in acquisition mix, product quality, or pricing has shifted.

12. Ignoring localization revenue

Symptom: the app runs a single English metadata, a single global price, and a US only paywall. International revenue is flat because international users cannot find the app in store search and cannot afford the US price at their local purchasing power. Fix: localize metadata (title, subtitle, keyword field, screenshots, description) into the top 10 to 20 markets by revenue potential. Localize price to purchasing power bands (usually 3 to 6 pricing tiers globally). Localize creative for the paid campaigns in each market rather than translating US creative literally.

Category application: how the pattern lands in each category

The general playbook applies to every consumer mobile app category, but the specifics differ meaningfully. A brief read across the categories where mobile subscription and mobile freemium patterns are most active today.

Gaming (free to play and premium)

Free to play gaming monetizes through in app purchase (consumable items, unlocks, cosmetic goods, battle passes) rather than subscription in most cases. LTV is dominated by a small percentage of high spending users often called whales, and the growth model is to acquire large volumes cheaply and monetize the small percentage that converts to paying. Paid UA in gaming is more volume oriented than in subscription: ROAS on installs is the primary metric, and the bidding operates at very short measurement windows (Day 3 ROAS is a common target). Creative is heavily video and often uses playable ads. Premium paid games (a one time price) are a smaller share of the App Store and Play Store revenue but they still exist, particularly in the higher production value tier. Their growth model relies on strong ASO, editorial features, and word of mouth because paid UA rarely pencils on a single one time transaction.

Health and fitness

Consumer subscription is the dominant monetization shape. Trial to paid conversion typically runs 15 to 25 percent on a 7 day trial for fitness apps. Retention is highly seasonal (New Year, spring, summer) with churn patterns that follow the same cycle. Push notifications tied to scheduled workouts and streaks are a heavy driver of retention. Creator UGC on TikTok and Reels is one of the most productive paid UA channels because the fitness category has a large native creator ecosystem. Price sensitivity is meaningful and localization of pricing has substantial upside internationally.

Productivity

Consumer and prosumer subscription. Trial to paid conversion is often higher than in fitness (20 to 40 percent) because the user typically installed with a specific task in mind and the paywall gates the feature that solves the task. Retention is often more durable than fitness because productivity apps become habitual tools rather than motivation driven habits. Paid UA is dominated by Google App Campaigns and Meta with meaningful ASA. TikTok is a growing channel for younger productivity apps (note taking, task management with a novel angle) but a smaller channel for legacy productivity categories.

Dating

Subscription is optional on top of a functional free tier in most dating apps (a free tier that provides basic matching with paid tiers unlocking discovery, filters, boosts, and premium features). Trial to paid conversion depends on the paywall design and typically runs 10 to 20 percent on the users who see the paywall. Retention is intent driven: users cancel when they are in a relationship or when they burn out on the search. Paid UA is dominated by Meta and TikTok with meaningful ASA. Creative style is dominated by real world outcomes (dates, relationships) rather than product feature demonstration.

Streaming and subscription content

Music, video, podcast, audiobook, and news subscription apps. Trial to paid conversion is often high (25 to 45 percent) because the user knows what they are getting and the value is legible. Retention is dominated by content depth and content refresh: a user who does not find enough content churns regardless of feature quality. Paid UA is high volume and heavily creator driven for content aligned to specific creator audiences. Family plans and multi user plans are meaningful revenue and retention levers in these categories.

Social apps

Monetization is usually advertising rather than subscription, with subscription tiers as an optional premium (ad free, verification, boost, extended reach). Retention is dominated by network effect: an app where the user's people are already active retains, an app where they are not does not. Paid UA is early stage focused, because at scale organic growth through the network effect and virality dominates paid. The failure mode specific to social is running paid UA that acquires users into a network they have no connection into, and losing them within the first week because there is no one there for them to follow.

Marketplace mobile apps

Marketplace apps (ride share, delivery, gig services, resale, dating in its marketplace shape) run the two sided liquidity dynamics of a marketplace on top of the mobile app dynamics of the store. Supply side acquisition often happens through channels the demand side does not use (driver acquisition through Craigslist and industry networks rather than TikTok, host acquisition through direct outreach rather than paid social). Demand side acquisition uses the standard mobile UA stack. Retention on the demand side is highly correlated to supply density in the user's market; retention on the supply side is highly correlated to demand density and earnings. The playbook for marketplace launch specifically is covered in a separate case study; the mobile specific mechanics on top of the marketplace model are described here.

Edtech

Language learning, professional skill building, kids education, tutoring, test preparation. Subscription is the dominant monetization shape, with trial lengths often longer than the consumer default (14 day trials are common in edtech because the value moment often takes several sessions to arrive). Retention is defined by course completion and by skill acquisition, both of which are longer time horizons than most consumer apps. Push notification strategy in edtech has to balance encouragement with fatigue: a language learning app that pushes daily reminders can build a streak habit or can burn out the user, depending on tone and timing. Duolingo's public playbook on push and streak mechanics is a widely referenced pattern.

Utility apps

Weather, calculator, VPN, translator, scanner, photo editor, and the long tail of function specific apps. Monetization is usually subscription with a hard paywall that gates the utility, or one time purchase with in app upgrades. Retention is very high on the utility apps that become the user's default in the category and very low on the ones that do not (a user only installs and pays for one weather app, one calculator, one VPN). The competitive dynamic is winner take most within a category, which puts a premium on being the app that a user picks first, which puts a premium on ASO and on early ratings and review volume.

Tools around the mobile growth stack

ASO tooling. AppTweak, Sensor Tower, data.ai (formerly App Annie), and MobileAction for keyword research, competitive tracking, and rank monitoring. AppFollow for review moderation and response workflows. Storemaven and SplitMetrics for pre store A/B testing of screenshots and creative concepts.

App Store Connect and Play Console. The primary source of truth for organic install performance, subscription events, and cohort retention. Both consoles have improved substantially in recent years and are now central rather than supplementary in most operator workflows.

MMPs. AppsFlyer, Adjust, Kochava, Branch, and Singular for cross channel attribution, SKAN normalization, deep linking, and cohort reporting. The MMP is usually a two year decision because migration is heavy.

Product analytics. Amplitude, Mixpanel, or first party equivalents for user level event tracking, funnel analysis, and cohort behavior analysis inside the app. The distinction from the MMP is that the MMP tracks acquisition attribution while product analytics tracks in app behavior.

Subscription management. RevenueCat, Adapty, Purchasely, or in house subscription infrastructure. RevenueCat has become the market default for early stage and mid stage subscription apps. Larger operators eventually internalize.

Lifecycle marketing. Braze, Iterable, Airship, OneSignal, and Clevertap for push notification, in app messaging, and email orchestration. Customer.io for smaller stage teams. Klaviyo where email is the primary channel.

Creator program tooling. Whalar, Grin, Aspire, Modash, and Traackr for creator sourcing, briefing, contracting, and reporting. TikTok Creator Marketplace and Meta Creator Marketplace are the platform native equivalents.

Paid UA platforms. Apple Search Ads for iOS store search. Google Ads (App Campaigns) for cross Google inventory. Meta Ads Manager for Facebook and Instagram. TikTok Ads Manager for TikTok. Reddit Ads, Snap Ads, Pinterest Ads, and Spotify Ad Studio for the diversification layer.

SKAN and conversion value tooling. Most MMPs provide SKAN conversion value schema management and optimization suggestions. Singular, AppsFlyer, and Adjust all have dedicated SKAN dashboards. Some larger operators build their own SKAN aggregation and modeling layer on top of the MMP.

Media mix modeling. Recast, Rockerbox, or in house MMM built on Python or R. The MMM layer is what fills the attribution gap SKAN cannot close, and it is what larger operators use to allocate strategic budget across channels beyond what tactical bidding can decide.

KPIs that matter

Store impressions and product page views (per market, per source). The top of the ASO funnel. Segment by search, browse, referrer, and Today tab to know where growth is actually coming from.

Store conversion rate (impression to install). The output of the ASO metadata plus creative. Track by source and by market. The single most direct measure of store page effectiveness.

Cost per install (per channel, per campaign, per creative). The tactical bidding metric. Meaningful only in context of downstream value; a low CPI on installs that do not activate or subscribe is not a win.

Cost per trial start and cost per subscription (per channel). The value metrics that CPI has to be interpreted through. These are the metrics the paid team should be managed against.

Trial to paid conversion rate (per cohort, per acquisition source). The paywall effectiveness metric. Track by cohort to detect drift; segment by source to detect paywall mismatch to acquisition mix.

Install retention curves (Day 1, Day 7, Day 30, Day 90). Onboarding and first value delivery diagnostic. Cohort by cohort trend is more predictive than the current curve.

Subscriber retention curves (Month 1, Month 3, Month 6, Month 12). Ongoing product value diagnostic. Below the healthy band signals a churn problem that needs categorized cancellation reason data to diagnose.

Cohort LTV (12 month, 24 month, lifetime projection). The value the growth team is buying. Report by cohort not blended.

LTV to CAC ratio (by cohort, by acquisition source). The unit economics floor for whether the growth motion is sustainable.

Save rate on cancellation attempt. How much of would be churn is recovered through the save flow. Track as a KPI and iterate.

Push opt in rate (by prompt placement, by cohort). Diagnostic of whether the prompt is well timed. Low opt in rate is fixable through prompt placement and priming.

Rating and review velocity (per market, per version). Store ranking signal. Watch for negative sentiment clusters that need product or messaging response.

Creator content performance (per creator, per amplification tier). Which creators produced content that scaled through paid amplification and which did not. Informs which creators to work with again.

A note on operator judgment

Every metric and mechanic in this playbook is downstream of decisions an operator has to make with imperfect information. Which trial length to test first when the segment is small enough that both would take months to reach significance. Whether to spend the next dollar on Apple Search Ads defense or on TikTok creator content. Whether to prioritize Day 7 retention improvement or Month 3 subscriber retention improvement when the engineering team has bandwidth for one but not both. Whether to raise price and lose trial volume or hold price and lose revenue growth.

The playbook does not answer those questions. It gives the operator the vocabulary, the metric structure, and the sequencing so that the answers get argued against a shared baseline rather than against opinion. The best mobile growth operators are the ones who know the playbook well enough to know when to deviate from it, and who can articulate why the deviation is right for the specific product and stage they are working on.

FAQ

Why is mobile app growth its own discipline separate from web growth?

Because the App Store and Play Store sit in front of the app as a required distribution surface that the operator does not own and cannot bypass. Every install is either a browse install on the store, a search install on the store, or a paid install attributed through a store postback. The store owns the product page, the ranking, the review surface, the search index, the pricing display, and the subscription billing rails. Web growth has none of that. The operator who treats a mobile app like a website with a download button underinvests in the ASO surface and the store native mechanics that decide half of the install curve.

What is ASO and what does the ASO discipline actually cover?

ASO is App Store Optimization: the practice of getting an app to rank in App Store and Play Store search, converting store impressions into installs, and holding those installs into activated users. It covers keyword strategy for the store search index, the title and subtitle and keyword field, the app icon, the screenshots, the video preview, the promotional text, the category selection, the ratings and reviews signal, the localization strategy, and the Custom Product Pages that pair with paid campaigns. ASO is separate from web SEO because store search runs a different index with different ranking signals, and it is separate from paid UA because the conversion happens on a page the operator controls only through the store interface.

How different are App Store and Play Store from each other?

Different enough that they are two ASO programs, not one. App Store ranks off a title field, a subtitle field, and a 100 character keyword field that users never see. Play Store ranks off a title, short description, and long description that users do see, with density signals. App Store screenshots can be portrait or landscape and up to ten. Play Store screenshots are shown differently and paired with a feature graphic. App Store review prompts route through SKStoreReviewController with a rate limit. Play Store uses the In App Review API with a similar limit. Subscription billing runs through StoreKit on iOS and Play Billing on Android with different tax handling and different price localization mechanics. The operator who treats both stores as one loses on both.

What does the ATT era do to mobile attribution?

ATT is App Tracking Transparency, the iOS prompt that requires user consent before an app can access the IDFA. Opt in rates vary by category and typically sit in the 25 to 40 percent band, meaning the majority of iOS users are attributed through SKAdNetwork or SKAN, Apple's privacy preserving attribution framework. SKAN sends a postback per install with a limited conversion value payload, with a 24 to 72 hour delay depending on postback tier, and it does not expose user level data. The consequence is that iOS attribution is now aggregate and delayed rather than deterministic and immediate. The stack that works is SKAN plus cohort level media mix modeling for the gaps SKAN cannot fill, plus MMP integration through AppsFlyer or Adjust or Kochava or Branch or Singular to normalize signal across channels.

What is a healthy trial to paid conversion rate on a subscription app?

It depends on category, trial length, and paywall placement. Directional bands from published benchmarks: subscription fitness apps in the 15 to 25 percent range on a 7 day trial, productivity apps in the 20 to 40 percent range on a 7 day trial, dating apps in the 10 to 20 percent range where subscription is optional on top of a functional free tier, meditation and health apps in the 20 to 35 percent range, and utility apps often higher when the trial gates the actual utility. Shorter trials generally convert at a higher rate but produce fewer trials started. Longer trials produce more trials started but convert at a lower rate and have higher forgetting churn. The operator's job is to test trial length per market and per segment, and to instrument the cancellation reason survey so that the churn diagnostic is legible.

Which retention benchmarks matter on a mobile subscription app?

Install retention (Day 1, Day 7, Day 30, Day 90) is diagnostic of onboarding and first value experience. Subscriber retention (Month 1, Month 3, Month 6, Month 12) is diagnostic of ongoing product value and pricing. The two are separate curves and reporting a blended active user number hides both. Directional bands for a healthy subscription app: Day 1 install retention 40 to 60 percent, Day 7 in the 20 to 30 percent range, Day 30 in the 10 to 15 percent range, Month 3 subscriber retention 70 to 80 percent, Month 12 subscriber retention 40 to 55 percent. Category matters. Gaming has different curves than health, and streaming has different curves than productivity. The operator watches the curve shape and the cohort by cohort trend rather than a single top line number.

Should a new subscription app use a 7 day trial, a 14 day trial, or no trial at all?

Test all three. The 7 day trial is the most common default because it aligns with the App Store's minimum introductory offer period and because most users make a keep or cancel decision within the first week of any subscription. The 14 day trial produces more trials started, especially in categories where the value moment lands after several sessions, but it also produces more forgetting churn where the user simply forgets to cancel and gets a chargeback complaint. No trial with a hard paywall produces the fewest trials started but the highest quality subscribers, and it fits categories where the value is legible without extended use (utility, entertainment content, dating). The right answer is category specific and it is discoverable through Custom Product Page paired paywall variant tests, not through opinion.

How much of mobile UA should sit in Apple Search Ads versus other channels?

Enough to defend the brand keyword and to capture competitor keyword intent, and not so much that the channel becomes the entire growth motion. Apple Search Ads is the highest intent channel in mobile UA because the user is already in the App Store searching for something and the tap goes directly to the product page. It also runs on Apple's LAT and LATAT data, which is deterministic in ways ATT era third party attribution is not. The failure mode is over indexing on ASA and neglecting Google App Campaigns, Meta app ads, TikTok app ads, and creator UGC amplification. A durable growth stack usually has ASA as a foundation, Google App Campaigns and Meta as scale channels, TikTok as a creative volume channel, and a small allocation to Reddit, Spotify, or Snap for diversification and net new discovery.

What are the biggest failure modes in mobile app growth?

Chasing installs without measuring subscription conversion, over indexing on Apple Search Ads without diversifying, ignoring ASO because the paid team owns growth, running trials too short for the product's actual value moment, price testing without segment isolation, treating iOS and Android as one channel, ignoring SKAN postback configuration and losing signal, treating the cancellation flow as a compliance detail rather than a retention lever, and shipping a paywall that converts on cohort A but has never been tested on cohort B or C. Every one of those has killed real subscription growth curves at real mobile app companies.

Does this playbook apply outside subscription apps?

Yes. The ASO fundamentals, the App Store Connect metrics, the Custom Product Page mechanics, the paid UA stack, the attribution model, the retention curve discipline, the lifecycle marketing pattern, and the creator UGC motion all apply to gaming (both free to play with in app purchase and premium paid), health and fitness, productivity, dating, streaming and subscription content, social, marketplace mobile apps, edtech, and utility. What varies by category is the monetization shape, the value moment, the retention curve baseline, the paid channel weighting, and the creative style. The underlying operating pattern is category agnostic.

Related reading

• Two-sided marketplace launch playbook
• Inkgility Design Studio and Design Services playbook
• GoHighLevel CRM playbook
• Brand strategy and identity playbook
• Content marketing operations playbook
• Email lifecycle marketing playbook
• All case studies and playbooks