Why apps fail after launch is a budget question: the money ran out on the day the build finished. The half of the budget that got spent, the code, was the half that was never in doubt. The half that decides whether anything happens, getting people to install the app, getting them through the first session, measuring what they do and changing what they do not, had no line in the budget and nobody owning it. An app with no plan and no money for the three months after launch has already failed, however good the engineering.
Most briefs that reach us have a figure for the build and nothing after it.
Why apps fail after launch: what does the data say?
They fail because nobody arrives, and most of the people who do arrive leave on the first day. Sensor Tower's State of Mobile 2026 (January 2026) found the top 1 per cent of publishers took 92 per cent of in-app purchase revenue and 81 per cent of downloads in 2025, and said most categories now have "limited room for new entrants" because people already have the apps they need. On the supply side, Apple's 2025 App Store Transparency Report counts 2,172,472 apps on the store, and Appfigures counted 557,000 new ones submitted in 2025, up 24 per cent on 2024. Back in 2023, Appfigures found 80 per cent of apps with any real downloads were getting fewer than 10,000 a month.
Getting through review is the part people worry about, and it is the easy part. Apple rejected 2,093,244 of the 9,100,620 submissions it reviewed in 2025, mostly on performance, legal and design grounds, and approved 387,087 of them after a revision. Review checks whether the app works and follows the rules. Nothing in it checks whether anyone wants it.
How many users does a new app actually keep?
Fewer than one in ten by the end of the first month, on the industry's own benchmarks, and those benchmarks come from apps that had marketing behind them. Adjust's retention figures (April 2024) put Android at 24 per cent of installers returning on day one, 11 per cent on day seven and 6 per cent on day 30; iOS does slightly better at 27, 14 and 8. Europe sits at the Android level. By category, day-one retention runs from 29 per cent for games and social apps down to 18 per cent for shopping and 13 per cent for food and drink.
AppsFlyer's 2025 uninstall report (Android only, 2,200 apps) found installs you pay for are uninstalled at a rate 22 percentage points worse than installs people found for themselves, and that most uninstalls happen on day one. Amplitude's 2025 Product Benchmark Report, drawn from more than 2,600 companies, sets the bar lower still: 7 per cent of users returning on day seven puts a product in the top quarter, and for half of all products more than 98 per cent of new users are gone within two weeks. Amplitude also found 69 per cent of the products in the top band at day seven were still there at three months, which is why the early numbers are the ones to watch.
Every one of those figures comes from a company that paid for analytics or attribution. The median for an app launched with nothing behind it is worse, and nobody publishes it.
What does it cost to get people to install an app?
Between well under a dollar and more than five dollars an install, depending on platform and category, and the industry spent $109 billion on it in 2025. Liftoff's 2024 Mobile Ad Creative Index puts e-commerce installs at around $1.00 to $1.60, finance at just under $1 on Android, and entertainment video ads at about $2 an install on Android against $5.31 on iOS. Adjust's 2025 shopping app report gives a European e-commerce cost per install of $1.66. AppsFlyer measured $109 billion of app marketing spend in 2025, with user acquisition at $78 billion and remarketing, the money spent getting people who already installed to come back, up 37 per cent to $31.3 billion. The fastest-growing line in app marketing is spent after the install.
Run the arithmetic on a shopping app. At $1.66 an install, $10,000 buys about 6,000 installs. At Adjust's 18 per cent day-one figure for shopping, about 1,100 of those people open it the next day. On the European median of 6 per cent at day 30, about 360 are still there after a month, before the paid-install penalty AppsFlyer measured. Ten thousand dollars for 360 regular users is what the first month costs, and it has to be in the plan.
What should the second half of the budget cover?
Five lines, each with an owner and a number against it.
Line | What it buys | Why it needs its own budget |
|---|---|---|
Acquisition | Installs, from ads, search, PR or an existing customer base | Installs cost $1 to $5 each, and paid ones churn 22 points faster than organic (Liftoff 2024, AppsFlyer 2025) |
First session | Onboarding that gets to the point, with the first screen's job done inside a minute | Most of the loss happens on day one (Adjust 2024, AppsFlyer 2025) |
Measurement | Analytics wired in before launch, with day-one, day-seven and day-30 retention as the numbers that matter | Day-seven retention predicted three-month retention 69 per cent of the time (Amplitude 2025) |
Iteration | Versions 1.1 to 1.3 in the first three months, driven by what the data shows | Pixalate found 74 per cent of apps delisted from Google Play in Q2 2024 had not been updated in two years |
Reaching back | Notifications, email and remarketing for people who installed and drifted | Remarketing grew 37 per cent in 2025, faster than acquisition (AppsFlyer) |
The line that gets cut first is iteration, and it is the one that separates a live product from the delisted pile. The line that gets overused is the last one. Ofcom's Online Nation 2025 (December 2025) found 19 per cent of UK adult internet users had deleted apps because they felt they were spending too much time on them and 28 per cent had disabled notifications, and Deloitte's Digital Consumer Trends 2025 (4,150 UK respondents) found half had turned off all notifications for one or more apps. The default response to a new app that asks for attention is to remove it, which is why the behavioural work has to be in the product, and why the first tap has to land before the tenth feature gets built.
How long after launch should you plan for?
Three months, as a minimum, with the money and the decision-maker assigned before launch. No benchmark defines that window; it is ours, and it follows from where the figures fall. The uninstall data runs to 30 days and the retention cohorts to day 30, so the first month tells you whether the first session works. Amplitude's three-month cut is where the day-seven winners were confirmed, so by month three you know whether the app has a future or whether the money should go elsewhere.
Our MVP work includes launch support, analytics wired in from the start and a plan for the version after this one, because a first version that answers the question needs the three months after launch to answer it. If the whole budget only stretches to the build, discovery is where to say so. The answer is usually a smaller build and a funded launch, which is a better bet than a bigger build and silence. Sometimes the answer is that you did not need an app at all.