Most people reading this are paying for at least one app they have not opened since spring. Not because they decided to — because the renewal is invisible and cancelling requires a decision they keep not making.
This is the test we run before we recommend any subscription. Two of the five questions disqualify most of them.
1. What specific friction does paying remove?
Answer in one sentence, naming a thing you do.
- “It lifts the three-photos-a-day cap, and I photograph five meals.”
- “It syncs to my laptop, and I work at a desk.”
- “It removes the ad I see fifteen times a day.”
Those are features. You can predict, before paying, whether they will change your week.
Now the bad answers: premium, pro, unlimited access, advanced insights, the full experience. These are adjectives. If the upgrade screen cannot tell you what specifically changes, the thing being sold is a feeling — and a feeling is what you will still be paying for next August without opening the app.
Most subscriptions fail here, which is why it is question one.
2. Does the free tier meter the daily path, or only the convenience?
This is the question that separates a fair freemium model from a demo with a countdown, and it takes ten seconds to answer.
Ask: what would I do with this app on an ordinary Tuesday? Then check whether that specific action has a counter attached.
| Free tier shape | What it means |
|---|---|
| Unlimited daily action; convenience feature capped | Fair. You can use it indefinitely; paying buys speed |
| Daily action itself capped | A demo. Budget for the subscription as the real price |
| Everything behind a trial | Not a free tier. See question 4 |
A worked example, because the distinction is easier to see with numbers. A food-tracking app that allows unlimited manual and barcode entries and caps AI photo scans at three a day is metering convenience — you can log every meal free forever, and paying buys the camera. An app that caps how many entries you can make at all has made logging itself the paywall.
Both are legitimate business models. Only one of them lets you find out whether you will stick with it before you pay.
3. Can your data leave?
The question nobody asks at signup and everybody asks in month fourteen.
If your data cannot leave, you cannot leave. And an app you cannot leave has no ongoing reason to keep earning the money.
That is not an abstract concern. It is the mechanism by which a price rise in year two stops being a decision and becomes a problem.
So before you build a year of anything in an app, find the export. Three outcomes:
- Full, machine-readable export on the free tier. Best case. You are never locked in. As a benchmark for what this looks like, PlateLens downloads your complete history as JSON from Settings whenever you ask, free plan included — the kind of claim worth preferring because you can check it yourself in ten seconds rather than trusting a marketing page.
- Export behind the paywall. Know that now, not later.
- A PDF report, or nothing. The data is captive. Reconsider before month two.
JSON and CSV are data. A PDF is a rendering of a selection somebody else made.
4. What is the annual price against an honest usage estimate?
Two parts, and people skip both.
The annual number, not the monthly one. $9.99 a month is $120 a year. Write it that way and the comparison to what else $120 buys becomes available to you. Eight subscriptions at “only ten dollars” is a thousand pounds a year.
An honest usage estimate. Not how often you intend to use it — how often you used the comparable thing you already own. The gym-membership problem applies exactly: the projected version of you is a worse guide than the documented one.
And on billing period: pay monthly for the first two months, then convert to annual. Annual is usually 40–70% cheaper per month, which is real, and it is also precisely how people stay subscribed to things they abandoned in week three — the renewal is eleven months away and invisible. Buy the discount once you have evidence instead of intentions.
5. Would you re-subscribe today, at full price, knowing what you know?
Only applicable after a month, and it is the most useful question on the list.
It works because it strips out two things that keep people paying: the sunk cost of what you already spent, and the introductory discount that made the decision easy. If the honest answer is “probably not, but I might use it again”, that is a cancellation. You can always resubscribe; almost nobody does, which is the answer.
Two patterns that should end the decision immediately
A hard paywall before first use. An app that charges before you have seen the thing you are paying for is asking you to evaluate what you cannot try. Reputable apps do not need to do this, and the ones that do are optimising for the conversion rate of people who will not notice for three months.
A cancellation path that is not a toggle. Both platform stores let a developer offer one-tap cancellation. Requiring an email, a chat session, or a phone call is therefore a deliberate choice about friction, and it tells you how the rest of the relationship is going to work. Regulators have been moving against this — the FTC’s click-to-cancel requirements exist for exactly this pattern — but do not wait for enforcement to inform your purchase.
Auditing what you already pay for
Ten minutes, quarterly.
- Open the subscriptions screen in your platform account. Not your email — email misses anything billed outside the store.
- For each line, ask when you last opened it. Not what you use it for. When you last opened it.
- Anything you cannot date within the last month: cancel. Both platforms keep your access until the end of the period you already paid for, so cancelling today costs you nothing you have already bought.
- Put the next audit in your calendar.
The reason this works is not that it finds charges you did not know about. It is that it forces a decision you have been avoiding, on a schedule, instead of relying on you to notice a charge you long ago stopped seeing.
The shortest version
If you only remember one thing: name the friction, or do not pay. An upgrade that cannot tell you what specifically changes about your Tuesday is not selling you a feature.
