Almost everything you install, you abandon
Industry retention benchmarks are bleak and remarkably consistent. Average Day 30 retention sits somewhere around 4 to 8 percent depending on whose data you use and how they count. US users uninstall roughly 48 percent of the apps they install within a month, and a good portion of the survivors are never opened again.
More than nine in ten installs are effectively wasted. Some of that is the app being bad. A lot of it is a mismatch that was visible before installing, if anyone had looked.
Below is what to look at. Seven questions, in descending order of how much they predict.
1. How does it make money?
The highest-signal question available, and most people never ask it.
The funding model determines which experiments a company will run for the next five years. It predicts what data the app wants, whether it will interrupt you, and whether it treats you finishing quickly as success or failure.
| Model | What it optimises for | What to expect |
|---|---|---|
| Advertising | Time on screen, ad impressions | Feeds, streaks, engagement notifications, broad data collection |
| Subscription | Renewal, perceived ongoing value | Feature growth to justify the price, retention pressure at cancellation |
| One-time purchase | Being worth buying once | Little pressure to engage you, but weak funding for long-term maintenance |
| Free with paid upgrade | Conversion to paid | The free tier gets deliberately limited over time |
| Data resale | Volume and richness of data | Permissions unrelated to features, vague privacy policy |
| Genuinely unfunded | Nothing yet | Either abandonment or a monetisation change later |
None of these is disqualifying, including advertising. What matters is that the model is legible and that it does not conflict with how you want to use the thing.
The specific case to avoid is a utility funded by attention. A tool exists to be finished with. Advertising pays for time spent. An app in that position has to work against its own purpose, and it will resolve the tension in favour of revenue, because it has to. We go into the mechanics in how free apps actually make money.
The warning sign is not free. It is free with no visible answer to how the lights stay on.
2. Does its scope match your problem?
Most abandonment is a scope mismatch, in either direction.
Too big. You wanted to track one thing. The app offers a workspace, collaboration, templates, and an assistant. Setup takes twenty minutes and you abandon it during setup, which is where a large share of installs die.
Too small. It does exactly one thing beautifully, but your actual problem had three parts, so you keep it plus two others and eventually consolidate.
Judge scope against your problem rather than in the abstract. A wide app is right for a wide problem. The failure is a wide app for a narrow one, because you pay the complexity cost every time you open it and only use a slice.
The related question is whether the scope will stay where it is. Most successful narrow apps widen, because widening is the standard way to raise a revenue ceiling. Whether that is a problem depends on whether the widening happens in your way. We covered the shape of this in what is a single-purpose app.
3. What does it ask for?
The permission list is the business model made visible, and it is available before you install: both stores show the permissions and the data-safety declaration on the listing.
The test is alignment. Every request should map to a feature you can name.
- A packing checklist asking for location: questionable, unless there is a weather feature.
- A packing checklist asking for contacts: no plausible mapping.
- A pet tracker asking for location while recording a walk: coherent.
- A pet tracker asking for location continuously: not coherent.
Mismatches usually mean an advertising or analytics library bundled into the app collecting on its own behalf, which is the most common source of surprising requests in otherwise ordinary apps. Our full breakdown of what each one exposes is in app permissions explained.
4. Will it still be here?
Store listings show version history, and it is the most useful thing on the page.
- Regular updates over 12 or more months. Someone is maintaining this. Strongest available signal.
- A gap of six months or more. Either finished or abandoned. Check whether recent reviews mention bugs that never get fixed.
- Constant updates with no visible change. Often A/B testing or SDK churn rather than improvement.
- Version 1.0, released last week. Might be excellent. Do not build a two-year record in it yet.
Also check whether the developer has other apps. A studio with three maintained products is a safer bet than a single app from an account with nothing else, though it is not decisive either way.
5. How hard is it to leave?
Exit cost is the question people ask last and regret asking last.
If you use something daily for two years, it accumulates your data. The question is whether you can take that with you.
- Low exit cost: exports to CSV, JSON, PDF, or an open standard. You can leave in ten minutes.
- Medium: export exists but is awkward or lossy.
- High: no export, or export in a proprietary format nothing else reads.
High exit cost means that in two years you will keep using it because leaving is expensive, not because it is good. That is captivity, and it is worth noticing that it is a design choice: export is not technically hard, and its absence is usually deliberate.
Check for an export function before you commit, not when you want to leave. This matters most for anything holding a long record: health data, financial history, pet medical logs, anything you might need to hand to a professional.
6. What do the bad reviews say?
The average rating is close to worthless. It is dominated by review prompts shown at moments of success, which is why so many apps sit at 4.6.
The useful content is the recent one and two star reviews. Sort by most recent, read twenty, and look for patterns rather than individual complaints.
- Repeated mentions of the same bug over months means nobody is fixing it.
- "Was great until the update that added X" tells you the direction of travel.
- "Cancelled and was still charged" is a serious signal about how the company operates.
- A single furious review about something that does not affect you is noise.
Also worth reading: recent five star reviews that are suspiciously short and generic, appearing in clusters. That pattern is usually purchased.
7. Does it finish?
The last question, and the one that separates a tool from a habit.
Open it. Do the thing you came to do. Then watch what happens.
Does the app let you leave, or does it offer you something else? Is there a next item, a suggestion, a streak to protect, a tab you have not visited?
For entertainment, not finishing is the point and that is fine. For a utility, not finishing is a defect. Something you installed to solve a specific problem should be content when the problem is solved.
A five minute pre-install check
Realistically you will not run seven questions on every app. Here is the compressed version, done on the store listing before installing:
- Scroll to the permissions and data safety section. Anything that does not map to a feature you can name?
- Check the version history. Updated in the last three months?
- Sort reviews by recent, filter to one and two stars. Read ten. Any repeated pattern?
- Find the pricing. If it is free, can you tell how it makes money in under a minute?
- Search the name plus the word export. Can you get your data out?
Five minutes, and it will stop most of the installs you would have abandoned anyway.
Applying it honestly to our own
It would be poor form to write this and not run the test on ourselves, so here is Welltide against the seven questions, including the parts that do not flatter us.
How we make money: currently, we do not. PackPilot and PawDex are free on Google Play with no advertising. By the standard in question 1 that is a legitimate warning sign, and the honest answer is that the plan is a paid tier once each app has enough users to justify it, not advertising. You are entitled to weigh that as an unresolved risk, because it is one.
Scope: deliberately narrow. One job each. If your problem is wider, these are the wrong tools and a suite will serve you better.
Permissions: short lists, and each request maps to a named feature. PawDex asks for location to record a walk route, active during a recording. Neither app requests contacts, microphone, or SMS.
Longevity: both are actively developed, and both are young. That cuts both ways and you should weight it accordingly.
Exit cost: the weakest area, and worth stating plainly rather than burying. PawDex has PDF export of logs on the roadmap for vet visits and it is not shipped yet. Until it is, exit cost is higher than we would like.
Does it finish: yes, by design. No streaks, no feed, and notifications only when a dose is due or a trip is approaching.
Two of those answers are marks against us. Publishing them is the point, because an app that will only tell you its strengths has already failed question 1.
Related Reading
- What Is a Single-Purpose App? covers scope in more depth.
- App Permissions Explained covers what each request exposes.
- How Free Apps Actually Make Money is the full version of question 1.
- Dark Patterns in Apps: A Field Guide covers what a badly aligned app does once you are inside it.
Sources and further reading: App retention rates, Business of Apps, mobile app retention benchmarks, UXCam.
Frequently Asked Questions
Industry benchmarks put average Day 30 retention at roughly 4 to 8 percent, meaning well over 90 percent of installs are effectively abandoned within a month. US users uninstall around 48 percent of the apps they install within 30 days, and many of the rest simply go unopened.
How does this make money. The answer predicts almost everything else: what data it wants, whether it will interrupt you, whether it will add features you did not ask for, and whether it rewards you finishing quickly or staying longer.
Check the update history on the store listing. Steady updates over 12 or more months is the strongest available signal. An app with no update in a year is either finished or abandoned, and the review section usually tells you which.
How hard it is to leave with your data. An app that exports to a standard format has low exit cost. One that traps years of records in a proprietary system has high exit cost, which means you keep using it out of captivity rather than preference.
The average score is close to useless because it is dominated by prompts shown at moments of success. The recent one and two star reviews are far more informative, since they describe the specific ways the app fails.
Paying aligns incentives more often than not, because the app makes money when you find it worth keeping rather than when you look at it for longer. It is not a guarantee, since paid apps can still be abandoned or add advertising later.