Mobile App Retention Benchmarks 2026: Day 1, 7 & 30 by Category

Acquisition costs have climbed for years, and in 2026 the metric that decides whether an app can grow is no longer only what an install costs — it is what happens in the days after. Retention sets your growth ceiling: it shapes LTV, which shapes what you can afford to pay per user, which shapes which channels are open at all. Two apps with identical $3 CPIs and different D30 retention are not running the same business.

Yet "is our retention good?" still gets answered with a single global average that blends games with banking apps. This guide gives working category ranges as Exp(G) operating ranges (context, not commandments), the definition hygiene that makes comparisons honest, the first-hour diagnosis that usually explains day thirty, and the fixes that bend the curve. Prefer interactive comparison? Use our benchmarks tool.

How to read retention numbers before you compare any

Definitions matter more than people admit. Day-N retention usually means the share of users who open the app on day N after install. Calendar day versus 24-hour windows, all installs versus attributed installs, and timezone handling can swing figures by several points — one reason published benchmarks disagree. Keep your analytics definition constant and compare trends, not absolutes.

Separate open retention from revenue retention. An app can keep non-payers or lose users while whales remain. Subscription businesses should track both; ad-led games should watch retention beside ARPDAU. Public references such as Business of Apps retention data and partner docs from AppsFlyer and Adjust are useful sanity checks, not board-pack targets.

The 2026 working ranges by category

Averages are brutal: often only around a quarter of users return the day after install, and by day 30 many apps keep low single digits. Within that, patterns hold. These Exp(G) operating ranges are for planning conversations:

1. 01

   Casual & hyper-casual games

   D1 roughly 25–35%, D7 8–15%, D30 2–5%. High churn is structural; the business runs on ad monetisation of short lifecycles.

2. 02

   Subscription lifestyle (meditation, sleep, language, journaling)

   D1 25–35%, D7 12–20%, D30 6–12%. The trial window makes week one existential.

3. 03

   Health & fitness

   D1 25–30%, D7 10–15%, D30 5–8%, with seasonal cohorts that can flatter annual averages.

4. 04

   Social & communication

   D1 35–45%, D7 20–30%, D30 15–25% when network effects are real.

5. 05

   Fintech & banking

   D1 30–40%, D7 20–30%, D30 15–25%+. Utility and stored value keep users; the fight is activation.

6. 06

   E-commerce & marketplaces

   D1 20–30%, D7 10–18%, D30 8–15%, heavily purchase-cycle dependent.

If you sit at the top of your category range, acquisition maths gets easier — you can often outbid competitors and still pay back faster, the compounding behind paid vs organic. Translate retention into LTV and break-even CPI with the ROAS calculator and cash timing in the Payback Engine.

The first hour decides day thirty

The strongest pattern across accounts we run: D30 problems are usually D0 problems. Cohorts that reach a genuine value moment in their first session retain at multiples of those that do not. Before rewriting push copy, instrument the first hour:

• What is the activation event after which retention bends upward? Find it by correlating first-session behaviours with D7 retention.
• What percentage of installs reach it, and how fast?
• What blocks it — signup walls, permission storms, empty states, a paywall before any value?

Moving activation rate meaningfully typically does more for D30 than another re-engagement campaign. Store listing honesty matters too: ads and Custom Product Pages that oversell create D1 churn that looks like a retention bug and is actually a promise bug. Apple's product page guidance is worth reading beside your onboarding screens.

Three fixes that move the curve fastest

1. 01

   Cut time-to-value ruthlessly

   Defer signup when you can, defer permissions, pre-fill defaults, show the product doing its job inside the first minute.

2. 02

   Match the promise to the product

   Retention problems are often acquisition problems in disguise: ads that oversell attract users the product cannot satisfy.

3. 03

   Earn the notification

   Generic streak reminders train mute behaviour. Tie pushes to stored user value so messaging compounds.

Depth beyond those three still matters — habit loops, social graph, personalisation — but teams rarely fail because they lacked a fifth lifecycle email. They fail because nobody owned activation.

Cohort cuts that change the story

Blended retention hides the truth. Cut D1/D7/D30 by paid versus organic, channel, geo tier, OS, and creative family when volume allows. Creative-level retention is especially underused: two ads can share a CPI while selecting different users. Feed that into the D7 ROAS kill line so cheap non-retainers never look like wins.

Trial and paywall timing also distort cross-app comparisons. When you borrow a competitor's boasted D30, ask what event they counted. Pair retention reading with CPI benchmarks so budget debates stay two-sided.

Retention × acquisition is one system

Winning teams do not run retention and UA as separate kingdoms. Retention data feeds targeting (optimise toward activation, not raw installs), creative strategy, and payback maths. That loop is what we build inside Growth Engine. If your retention curve has been flat while CPI creeps up, book a call — the diagnosis is usually visible in a week of data.

Lifecycle after day one without spamming uninstalls

Once activation is healthy, lifecycle can compound. Prioritise messages tied to user-owned state: unfinished content, real progress, accurate plan reminders, and social loops when a second user creates value. Avoid daily generic pushes that train mute behaviour. Win-back should be segmented by how far users got — never activated versus activated-and-lapsed are different products. Measure win-back on incremental retained users, not open rates.

Coordinate email, push and in-app so you do not stack three nag events in twelve hours. When leadership asks for a retention "campaign", translate the ask into an activation metric and a D7 target by cohort. Campaign theatre without a north-star event fills calendars while curves stay flat. Pair lifecycle experiments with UA quality reviews so product is not asked to rescue systematically bad traffic.

Instrumentation checklist for the first session

• Time to first value moment (median and p90)
• Activation event completion rate
• Drop-off by onboarding step
• Permission deny rates that block core value
• Percent of users who see empty states
• Match between ad/CPP promise and first screen copy

Wire these to the same warehouse or MMP enrichment you trust for revenue. A pretty product analytics board that disagrees with finance cohorts creates political retention debates. Align IDs early. Keep store-side promise match in the same sprint as onboarding work — retention teams that ignore store creative leave a hole at the top of the funnel.

From benchmarks to budget decisions

Retention ranges are for diagnosis and target-setting; budgets should still clear contribution payback. If your D30 is top-quartile for category but payback is broken, you may have a monetisation or CPI problem, not a retention emergency. If D30 is weak and you are scaling spend, you are financing a leak. Lock operating targets: activation rate, D1/D7/D30 by channel, and the max CPI implied by those curves. That is how retention stops being a slide and becomes a healthy spending constraint.

A final operating note: publish retention targets the same way you publish CPI ceilings. When UA, product and finance share one activation definition and one D7/D30 band by channel, debates get shorter and experiments get cleaner. Revisit the bands quarterly or after major onboarding changes — not every time a weekly chart wobbles. And when a cohort looks "bad", ask whether creative mix, geo mix or onboarding shipped that week before you rewrite the entire lifecycle calendar.

If you want a practical starting pack: (1) lock definitions, (2) plot eight weeks of D1/D7/D30 by channel, (3) pick one activation event and put its completion rate on the same dashboard, (4) assign one owner for time-to-value fixes, (5) refuse scale budgets that ignore the Payback Engine output implied by your current curve. That sequence alone usually surfaces more truth than another benchmark PDF.

Teams that treat retention as a monthly report will always be late. Teams that treat it as a weekly operating constraint — beside CPI and the kill line — compound. That is the standard we hold clients to, and it is why retention work inside Growth Engine is inseparable from acquisition work rather than a separate "CRM workstream" that starts after the damage is done. When creative mismatch is the suspect, score top ads in the hook analyzer against the first-session promise.

Sources & further reading

• Business of Apps — app retention data
• AppsFlyer
• Adjust
• Apple — product page
• Google Play Console help
• Exp(G) benchmarks tool