Mobile App Statistics 2026: Downloads, Revenue, Usage and Retention

Mobile app statistics are the most recycled numbers in tech, and the least read. In 2025 people downloaded apps about 149 billion times and spent $167 billion inside them, according to Sensor Tower’s State of Mobile 2026 report. Both numbers are real. Neither says a single word about whether anyone will download, open, understand or pay for your app. A statistic without a source, a date, a geography, a category and a denominator is not evidence. It is decoration, often recycled crap, and founders pay for it twice: once in the pitch deck, and again when the roadmap is built on it.

Our experience designing and fixing mobile products keeps showing us the same costly pattern. A slide says the market is worth hundreds of billions. The team builds for eighteen months. The app launches into a store with more than two million other apps, the paid campaign brings people in, and most of them never reach the moment where the product is worth keeping. The market was real. The users were paid for. The value never arrived.

So this is not another list of forty disconnected facts. Every number below is dated and sourced, complete 2025 totals are kept apart from partial 2026 signals, and each cluster ends with what it means for your product. Numbers we could not trace to a dated primary source are not here.

How to read a mobile app statistic before it misleads you

There is an old line about lies, damned lies and statistics. The statistics are rarely the liars. The people repeating them are, usually by accident.

App statistics travel like a rumour in a school playground. Say that at nine o’clock a report says “3% of new apps in one category reached a revenue milestone in their first year”. By lunchtime the blog summary says “3% of apps succeed”. By the end of the day a pitch deck says “97% of apps fail, so ours needs a big budget”. Nobody lied on purpose. Every retelling dropped a qualifier.

Before any number goes into a deck, a roadmap or a board update, it has to answer five questions:

The denominator is the one teams skip, and it does the most damage. A school can say 100% of its pupils passed the advanced maths exam. True. Two pupils sat it. Nobody lied, and nobody learned anything about the school.

A second trap is the difference between an estimate and a receipt. Market-intelligence firms such as Sensor Tower model downloads and spend from panels and store data; their methodology notes say downloads are counted once per account, Android figures cover Google Play only, China is iOS only, and revenue is gross, before the store’s commission, with advertising excluded. That makes the figures excellent for direction and comparison, and useless as your revenue forecast. It is an estate agent’s guess at what houses on the street sell for, not the price anyone will pay for yours.

Mobile app market statistics: the complete 2025 picture

The latest complete year is 2025, and the reference numbers come from Sensor Tower’s State of Mobile 2026 report, published on 21 January 2026. Across the App Store and Google Play in 2025:

• Downloads: about 149 billion, up 0.8% year over year. Sensor Tower’s summary rounds it to “nearly 150 billion”, a new high above 2023.
• In-app purchase revenue: $167 billion, up 10.6%. That covers paid apps, in-app purchases and subscriptions, before the stores take their cut.
• Time spent: 5.3 trillion hours, up 3.8%, or roughly 3.6 hours per day per mobile user.
• Largest market by revenue: the United States, where consumers spent nearly $60 billion. Western European growth was led by the United Kingdom, Germany and France.

Read the growth rates side by side and the story is obvious. Downloads barely moved. Time spent crept up. Revenue jumped. Sensor Tower calls it a market shifting from adoption to monetization, and for a startup that is not an abstract trend. It means the easy users were acquired years ago. Every new customer now has to be taken from an app they already use, and the incumbents are getting better at charging the customers they have.

Then there is the other famous number. On 4 June 2026 Apple published a study by economists at Analysis Group saying the App Store ecosystem facilitated over $1.4 trillion in developer billings and sales in 2025. It sounds like the same market, more than eight times bigger. It is not the same thing at all. Of that total, $1.1 trillion was physical goods and services (groceries, food delivery, retail and travel bought through apps), $149 billion was digital goods and services, and $151 billion was in-app advertising revenue. Apple also says developers paid no commission on more than 90% of it.

So “the app market” is $167 billion or $1.4 trillion depending on whether you count what people pay inside apps across both stores, or everything that flows through apps on one platform, including the pizza. Both are correct. Put the wrong one in your deck and a sharp investor will ask which pizza you are selling.

What the 2025 totals mean for a product decision:

• Flat downloads mean acquisition is a fight, not a flood. Plan for switching costs: why would someone leave the app they already use?
• Rising revenue means people pay for clear value. They do not pay for features; they pay for a problem gone.
• The US is the largest revenue market, but a UK or European launch is not a consolation prize. Sensor Tower names the UK among the markets leading Western Europe’s growth.

Live 2026 signals: what the first half shows

Full-year 2026 numbers will not exist until early 2027. What we have are quarterly readings, and a quarter is a weather report, not the climate. Useful for direction; dangerous to annualise.

Sensor Tower’s Q2 2026 Digital Market Index, published on 11 August 2026, reports:

• Global in-app purchase revenue: $43.6 billion in Q2 2026, up 5.3% year over year, with downloads flat.
• Non-game apps: a record $24.4 billion, up 14.6%, while game revenue fell 4.5%.
• Stores: Google Play revenue grew faster than iOS, up 10% against 3%.
• Markets: US revenue fell 3% year over year, while the UK, Germany and France together added $500 million across both stores.
• Generative AI: consumer spending on AI apps more than doubled, up 108% year over year.

Two things stand out. Growth slowed from 2025’s double digits to single digits, so a five-year forecast at “10% a year” is already out of date. And the growth that remains sits in non-game apps and AI, exactly where every second founder is pitching right now.

Treat these as signals, not foundations. One breakout app, a currency swing or a price change at a giant can move a quarter. If your business case only works when Q2 repeats itself for five years, you do not have a business case. You have a holiday snap, not a climate.

Mobile app download statistics: volume is flat, supply is not

Sensor Tower’s report puts 2025 at about 284,000 app downloads per minute, and says that across the 80 markets it tracks for usage (China excluded), the average user opened 34 apps a month and about 10 different apps a day. Those ten are the real market. Everything else is fighting for a place in the queue.

Meanwhile the number of competitors is exploding. Apple’s 2025 App Store Transparency Report counts 2,172,472 apps on the App Store. Apple’s reviewers looked at 9,100,620 submissions in 2025 and rejected 2,093,244 of them. The biggest rejection reason, by a distance, was performance: 1,354,418 rejections, against 415,532 for design and 495,673 for legal. More than a million times last year, a submission was sent back for not working properly before a single user saw it.

Subscription apps show the same flood from the inside. RevenueCat’s State of Subscription Apps 2026, built on more than 115,000 apps and over $16 billion in revenue, counted about 2,000 new subscription apps launching per month in January 2022 and more than 14,700 per month by January 2026. AI coding tools made building an app cheap. They did nothing to make people want one.

Picture a careers fair where fourteen thousand new candidates walk in every month. Your CV might be brilliant. It is also in a pile, next to three candidates with the same idea in a different font. Being in the store is not distribution. It is attendance.

What the download and supply numbers mean for your product:

• Launch is not acquisition. Budget for how people will find you, and for the moment they arrive and decide.
• Performance is the first filter. If Apple sends back more submissions for performance than for anything else, users will leave for the same reason, with less patience and no appeal process.
• More supply means pricier users. RevenueCat expects the flood to bring more competition, higher acquisition costs and higher churn. Every paid install that never activates now costs more than it did two years ago.

Revenue by category: apps overtook games, and the pie is not shared

The headline category shift of 2025: for the first time, people spent more on non-game apps than on games. Sensor Tower’s report puts app revenue at $85.6 billion, up 21%, and game revenue at $81.8 billion, up 1%. Non-game spending is nearly three times what it was five years earlier.

The biggest non-game earners are still social media, streaming video and dating. Generative AI is the fastest riser: AI apps were downloaded 3.8 billion times in 2025, double the year before, and earned more than $5 billion in in-app purchases, nearly triple. Sensor Tower’s chief executive noted that ChatGPT was the third-highest-grossing app of 2025, behind TikTok and Google One.

Now the part the roundups leave out. The same report says the top 1% of publishers took 92% of in-app purchase revenue and 81% of downloads in 2025: about $154 billion for the top 1%, and $13.1 billion shared among the other 99% of publishers with any revenue at all. Sensor Tower adds that revenue among the bottom 99% is growing faster than at the top, which is good news, from a very small base.

RevenueCat’s subscription data says the same thing from the other end:

• Only 4.6% of newly launched subscription apps reach $10,000 in monthly revenue within two years.
• For the apps that do reach it, the median time to $10,000 a month is 109 days; business apps take 113 days just to reach $1,000.
• Apps launched before 2020 still earn 69% of all subscription revenue. Apps launched in 2025 or later earn 3%.
• Median year-over-year growth in monthly recurring revenue is 5.3%; the top 10% grew by more than 306%, and shrinking by more than a third puts an app in the bottom quartile.

A rising tide lifts all boats, the saying goes. In the app stores it mostly lifts the yachts. Premier League television money is enormous, and it does not buy the kit for a Sunday league team. Quoting category revenue to justify your app is quoting the Premier League’s broadcast deal to justify your pub team’s budget.

What this means for a startup decision:

• Category growth is a reason to look, not a reason to build. Growth in AI or finance tells you where demand is moving. It does not say which problem inside it is unsolved.
• Your realistic benchmark is the median, not the leader. Model your plan on median conversion and median growth, and treat anything above it as upside you have to earn.
• Speed to value decides survival. With most revenue held by apps that are years old, a new product wins only by solving a specific job more clearly or faster than the incumbent.

App usage statistics: time spent is not engagement with your product

Sensor Tower’s numbers on attention are enormous. People spent 5.3 trillion hours in apps in 2025, about 3.6 hours a day per user, more than 600 hours for every person on Earth. Social media alone took nearly 2.5 trillion hours, up 5%, which is more than 90 minutes a day for the average mobile user. AI assistants entered the top ten subgenres by time spent, with time up 426%, and generative AI apps as a whole took 48 billion hours, about 3.6 times the 2024 figure.

Here is the uncomfortable arithmetic. Those 3.6 hours are not waiting for you. An hour and a half goes to social media before your app gets a look in. The rest is shared with messaging, video, maps, banking and whatever the user’s employer makes them open. Your product does not compete with the other apps in its category. It competes with everything else on the phone for the same thumb.

Footfall on a high street tells a shop owner that people walk past. It does not say they will come in, and it certainly does not say they will buy. “People spend 3.6 hours a day on their phones” is footfall. What you need to know is whether the person who opened your app got what they came for in the first session.

Usage statistics are useful for three decisions:

• Session design. If your app gets minutes, not hours, the first useful result has to arrive in those minutes. A long setup before any value is a luxury only apps with habits can afford.
• Notification and re-engagement strategy. You are one interruption among hundreds. An alert that is not personally useful trains people to switch you off.
• Category choice. Time-hungry categories (social, video, games) are the hardest to enter. Many strong businesses need only a few focused minutes a week, and they win on how reliably those minutes deliver.

What usage statistics cannot tell you is whether your users are activated. For that you need your own cohort data, which is where most of the money actually leaks.

Retention and churn statistics: the numbers that decide whether the rest matter

Market size tells you how big the room is. Retention tells you whether anyone stays in it. And the most useful retention data available in 2026 says the decision is made faster than most teams design for.

RevenueCat’s State of Subscription Apps 2026, published in March 2026, reports:

• 55.4% of all cancellations of three-day trials happen on the day the trial starts, and 84% happen by the end of the next day.
• About half of all paid conversions (50.6%) also happen on day 0.
• Among annual subscribers, about 72% cancelled during year one in the 2026 data, up from about 56% a year earlier. Cancellations spike in the first month, go quiet, then spike again just before renewal.
• AI apps earn more per payer, with a 41% higher first-year realised lifetime value ($30.16 against $21.37 at the median), but their monthly plans retain 36% worse over 12 months than non-AI apps.
• On Google Play, 31% of subscription cancellations are involuntary billing failures, against 14% on the App Store.

A three-day trial is effectively a one-hour trial. It works like a house viewing. Buyers decide in the hallway. If the hallway smells of damp, nobody goes upstairs to admire the loft conversion. If the first session of your app is a registration form, a permissions wall and an empty dashboard, the loft conversion you spent a year building might as well not exist.

Choose a benchmark that matches the product you are improving. RevenueCat measures paid subscriptions; AppsFlyer’s European finance report compares a different set of users and tasks. Use category and channel splits to find the first-session problem your team can fix, then measure whether more people reach value and return.

Monetization statistics: what the paywall data actually says

Revenue per user is where market statistics meet your pricing page, and the 2026 data contradicts a lot of received wisdom.

From RevenueCat’s State of Subscription Apps 2026:

• Hard paywall versus freemium. Apps that ask for payment up front convert 10.7% of downloads to paid by day 35. Freemium apps convert 2.1%. That is roughly five times better.
• But retention evens out. After a year, hard-paywall apps keep 27% of annual subscribers and freemium apps keep 28%.
• Revenue per install. By day 60, hard-paywall apps make $3.09 per install against $0.38 for freemium.
• Trial length. Trials of 17 to 32 days convert 42.5% of trial users at the median; trials of four days or less convert 25.5%. Yet the share of apps using trials of four days or less rose from 42.1% to 46.5%.
• Region. The median first-year realised value per payer is $32 for apps by North American developers, $25 for Western Europe and $23 globally. Median day-35 download-to-paid conversion is 2.6% in North America and 2.0% in Western Europe.
• Web purchases are still only 3.2% of subscription revenue globally.

Free samples in a supermarket are the classic comparison. Plenty of people eat the cheese. Very few put a block in the trolley, and the shop pays for every cube. Freemium works when the free users are the product’s marketing: when they invite colleagues, share results or create the network that makes the paid tier valuable. Without that, freemium is a very generous cheese counter funded by your acquisition budget.

It would be easy to read the numbers and slap a hard paywall on everything. Don’t. A hard paywall only converts when the product shows its value before the paywall appears. If your onboarding cannot prove value in the first session, a hard paywall just moves the cancellation from day 3 to minute 3.

Beyond subscriptions, advertising is not a side business. Apple’s Analysis Group study put in-app advertising revenue in the App Store ecosystem at $151 billion in 2025, more than the $149 billion in digital goods and services. Ads, subscriptions, one-off purchases and transaction fees each shape the product differently. The comparison is in our guide to app monetization models.

What monetization statistics mean for your product:

• Pricing is a design decision, not a finance afterthought. The paywall’s position, the trial length and the first session are one flow and should be designed together.
• Android revenue needs billing hygiene. With 31% of Google Play cancellations caused by failed payments, retries, grace periods and a clear “update your card” path are revenue features.
• Model with regional medians. A North American price-and-conversion assumption applied to a European or Asian launch will overstate revenue from day one.

Platform statistics: iOS or Android depends on where your users stand

“Android has 70% of the world, so we build Android first” and “iPhone users pay more, so iOS first” are both half-truths. The honest answer depends on the market, the audience and the business model.

StatCounter’s mobile operating system share for the August 2026 snapshot:

A caveat that matters: StatCounter measures page views from mobile browsers on websites carrying its tracking code. It is a good proxy for which phones people carry, not a count of app installs or spend. It also shows the UK almost evenly split, which is why “our users are on iPhone” is an assumption to test, not a fact to inherit from a US competitor.

Revenue tells a different story from share. Sensor Tower’s Q2 2026 data shows Google Play revenue growing faster than iOS (10% against 3% year over year), and AppsFlyer’s State of Subscriptions for Marketers 2026, published on 31 March 2026 and covering 2,900 subscription apps, found subscription acquisition spending up 24% year over year, with Android growing at four times the rate of iOS; the Indian subcontinent and Latin America drove most of the net Android growth in paid installs.

Choosing a platform by global share is choosing which side of the road to open a shop on by counting every car in the country. You need the cars on your street. For a UK fintech the split is near even; for a US consumer subscription, the 60/40 iOS split and higher North American value per payer usually make iOS the first candidate; for a mass-market product in India, Android is not optional.

What platform statistics mean for your product:

• Decide by your segment’s devices, from your own analytics, research or competitor audience data, not from world share.
• If you must choose one platform first, choose where your earliest paying users are, then plan when the second follows.
• Cross-platform is a delivery choice, not a UX choice. Both platforms still need their own navigation conventions, permissions and billing recovery.

Global numbers, local customers

Sensor Tower’s report notes that nearly four out of every five downloads and dollars across the two stores in 2025 went to apps developed outside the user’s own country. Mobile is global. The customer is still local.

The differences are large enough to change a plan. Median first-year value per payer ranges from $32 for North American developers to $14 for developers in India and South-East Asia in RevenueCat’s data, more than a twofold spread. US spending fell 3% in Q2 2026 while the UK, Germany and France grew. In European finance, AppsFlyer finds neobanks acquire faster and traditional banks retain better.

A global average works like an average of 1.7 children per family: arithmetically fine, and no help at all when you are buying shoes for the 0.7. Your launch market has a specific currency, payment habit, regulatory setting, dominant platform and incumbent. The global statistic is context; the local one is the plan.

A large market does not validate your product

This is the section most statistics articles avoid, because it makes the other sections less exciting. A growing market is evidence that people spend money on a type of product. It is not evidence that they will spend it on yours.

We see the same slide in almost every early-stage kick-off: a market size with a great many zeros, a growth rate, a chart going up and to the right. Then we ask three questions. Who exactly is the first user? What are they doing today instead? What will they have to stop doing to use your app? The room goes quiet. The market slide cost a few minutes. The silence will cost the next year’s budget if nobody fills it.

The rows are recommendations, not rules, but the pattern holds. Every market statistic answers a question about the market. The questions that decide your product’s fate are about your users, and only your users can answer them. That is what UX research is for: evidence from the people you intend to serve, in the context where they will use the app.

It is also why we treat product discovery as the step between market data and design. Discovery takes the opportunity the statistics suggest and tests the audience, the problem and the priorities before anyone commits a roadmap to them. Skipping it means designing the answer before anyone has checked the question.

The market statistic says people will spend money. Only your users can tell you whether they will spend it on you.

How to turn mobile app statistics into product decisions

Statistics become useful when each one is attached to a decision. This is the order we work in with founders and product teams, and it is a real sequence: each step needs the one before it.

1. Write the decision first. “Should we launch on iOS or Android first?” “Can we charge from day one?” “Is this category worth entering?” A statistic without a decision attached is trivia.
2. Find the primary, dated source. First-party platform data and named industry reports beat blog roundups. Record the date, period, geography, category and denominator.
3. Match the cohort. Choose the benchmark closest to your category, market, business model and platform. If none is close, say so and do not borrow a distant one.
4. Turn it into a hypothesis about your users. “Hard paywalls convert five times better” becomes “our target users will pay after one session if they see their plan first”.
5. Test the hypothesis with your users. Interviews, prototypes and usability tests with the people you plan to acquire, before engineering starts.
6. Scope the smallest product that tests the riskiest assumptions. That is what MVP design is for: a focused first release that answers the questions the statistics raised, not one that tries to match the leader’s feature list.
7. Measure your own cohort from day 0. Activation, first-value time, day-7 retention and payment by channel. From this point your own numbers outrank every report in this article.

The order matters because each shortcut is a screw-up with an invoice attached. Skip step 3 and you plan a European launch on North American conversion. Skip step 5 and you build the prototype in code, which is the most expensive way to find out nobody wanted it. Skip step 7 and you keep buying installs that cancel on day 0, and call it growth.

Why ANODA starts with the first valuable action

A larger feature list cannot rescue a first session that asks for effort before giving value back. Another market slide cannot repair that journey. ANODA connects the audience, core task and payment moment, then designs the states and handoff that make the experience consistent.

In Fusion, our work connected personality and compatibility with signup, tutorial, feed and chat, supported by a component library for the development team. Bring us your audience, business model and acquisition plan; we turn them into a focused mobile experience people have a reason to keep.

Sources and methodology in this article

Every statistic above comes from one of these sources. External numbers change; the dates below are the dates we checked and the periods the sources cover.

The short version

The mobile market in 2026 is mature, huge and brutally concentrated. Downloads are flat, revenue is still growing, non-game apps and AI take the growth, a tiny share of publishers take most of the money, and users decide within the first session whether an app is worth keeping. None of that is bad news for a new product. It is bad news for a new product built on a market slide instead of on its users.

Use current primary data and translate every number into one specific product decision. The market will not tell you what to build. Your users will, if you ask them before the money is spent.