Your café app has a star rating whether you manage it or not. It sits directly under the app name on the App Store and Google Play listing, next to the install button, and it is the last thing a curious local looks at before deciding whether to download. A 4.7 gets the tap. A 3.1 with two one-star reviews from a rocky launch week does not.
The frustrating part is that ratings for a small local app are decided by a tiny number of people. Twelve reviews is a normal total for a single-location café app, which means two annoyed customers can set the number for a year. The good news is the reverse: a modest, deliberate ask can fix it in a month. Here is how to do it without breaking Apple’s or Google’s rules.
Why ratings matter more than review volume
For a national app, ratings feed discovery and category ranking. For a neighbourhood café app, discovery barely matters — nobody browses the App Store for coffee shops. Almost all of your installs come from people who already know you: a QR code on the counter, a link in your Instagram bio, a receipt, a staff member saying “we have an app.”
That changes what the rating is for. It is not a ranking signal you are optimizing, it is a trust checkpoint on the way to install. The person is already sold; the rating just has to not stop them. Which means:
- The star average matters much more than the review count.
- The three or four most recent reviews matter more than the ones from a year ago, because they are what shows on the listing.
- A single unanswered one-star does disproportionate damage, because with twelve total reviews it is both visible and a big share of the average.
If you have not set up your listing itself — name, subtitle, screenshots, keywords — do that before you chase ratings; it is covered in app store optimization for café apps.
The rules: what Apple and Google actually allow
Both stores let apps request a review from inside the app, through a native API that shows a standard system card. Both also put firm limits on how you may use it.
| Rule | iOS (App Store) | Android (Google Play) |
|---|---|---|
| How you ask | System-provided review prompt via StoreKit | In-app review card via the Play In-App Review API |
| Frequency cap | Max 3 prompts per user per 365 days, regardless of how often the app asks | Time-bound quota, not published and changeable without notice |
| Custom pop-up instead? | No — use the provided prompt | No — the card must be shown as-is, unmodified, no overlays |
| A “Rate us” button? | Allowed to deep-link to the store listing | Not allowed to trigger the API from a button; link to the Play Store listing instead |
| Pre-qualifying question (“Do you like the app?”) | Discouraged | Explicitly prohibited before or during the card |
| Gating content on a rating | Prohibited (Guideline 3.2.2) | Prohibited |
| User can disable prompts entirely | Yes, in device settings | Yes, quota and user state may suppress it |
Two practical consequences of that table are worth spelling out, because they are the ones vendors get wrong:
You cannot filter for happy customers. The old growth trick — ask “how are we doing?”, route the happy ones to the store and the unhappy ones to a support form — is explicitly prohibited on Google Play, and Apple’s design guidance points the same way. Do not let anyone build it into your app.
There is no way to know whether the prompt appeared. Neither platform tells your app whether the card was actually shown. So “we’ll just ask again if they didn’t rate” is not a strategy that exists. You get a small number of shots per user per year, silently.
When to trigger the prompt in a café app
The API decides whether to show the card. You only decide when to ask. Ask at a bad moment and you burn one of three annual chances on a distracted, irritated person.
For an order-ahead café app, the moments in rough order of quality:
- Just after a completed pickup. The order was ready, the coffee is in hand, the app just worked. This is the single best trigger and it costs the customer nothing to act on.
- When a loyalty reward unlocks. A free drink just landed. Goodwill is at its annual peak.
- After the second or third successful order, never the first. One order is not enough experience to review, and a first-run prompt reads as desperate.
- After a successful eGift card send. The customer just did something generous and it went smoothly.
And the moments to never ask:
- On first launch or during onboarding
- Mid-order, in checkout, or while payment is processing
- After any error, failed payment, or cancelled order
- Immediately after a push notification opens the app — they came for the offer, not to be interrupted
A sane rule for a café app: eligible after three completed orders, at most one ask per 120 days, only on the pickup-confirmed screen, and never within a week of a failed or refunded order. That respects the three-per-year ceiling and spends the asks on your best moments.
If Tany or another vendor builds your app, that logic sits on their side — ask them directly what the trigger rule is. “After every order” is a red flag; it wastes the annual allowance on a random sample of customers rather than a happy one.
The offline half nobody does
The in-app prompt reaches people who already installed. Your bigger constraint is usually total installs, and there the work is physical and social. This is the same muscle as getting customers to download your restaurant app, pointed one step further down the funnel.
What actually works at counter level:
- Train staff to ask once, specifically. “If the app’s been good to you, a quick review really helps us” — said to a regular collecting a mobile order, not shouted at a queue. Staff scripts beat signage, and it belongs in the same training as the app launch itself.
- Ask your regulars by name. Twenty regulars asked personally will do more than any campaign. These are the people whose reviews will also be the most credible to a stranger, because they mention real drinks and real staff.
- Time it around a good week. After a smooth seasonal launch, not during a POS migration.
- Never script the star count. Asking for “a five-star review” is the line that turns a legitimate ask into an incentivized one. Ask for a review; take what comes.
You may absolutely send a push notification asking for a review — it is not incentivized, not gated, and goes to people who already opted in. Send it once, to customers with several completed orders, and never to someone whose last order had a problem. The general discipline for that channel is in push notifications for cafés.
What to do about a bad review
You will get one. The most common café app one-star reviews are not about the app at all: a late order, a drink made wrong, a store that closed early. The review is about your café; the app is just where they found a text box.
The response playbook:
- Reply publicly, on both stores. Developer responses are supported on the App Store and Google Play, and the reply is visible to everyone who reads the review afterward. That reply is doing work for every future visitor, not just for the person who complained.
- Answer the actual complaint, not the star count. “You’re right that the 4:45 cut-off wasn’t clear — we’ve changed the wording in the app” beats any apology template.
- Fix it and say so. If two reviews mention the same confusing pickup flow, that is product feedback delivered for free.
- Do not ask them to change the rating. It reads badly in public and rarely works.
- Reset the summary only with a real release. Apple lets you reset your rating summary when you publish a new version. That is a legitimate tool after fixing a genuine launch bug, and a bad habit if you use it to dodge feedback.
The tone that works here is the same one that works on Google — measured, specific, non-defensive, and written for the next reader rather than the current one. We covered that at length in responding to a negative Google review, and every word of it transfers.
A realistic target
Do not chase a hundred reviews. For a single-location café app, a reasonable goal is 15-30 genuine ratings with a 4.5+ average, refreshed with a few new ones each quarter so the visible recent reviews stay current. That is achievable in one month with a correct in-app trigger plus twenty personal asks, and it is enough to clear the trust checkpoint for anyone deciding whether to install.
The prerequisite is boring and non-negotiable: the app has to be good. Orders ready when promised, loyalty that tracks correctly, no failed payments. Ratings are a mirror. If yours are bad and honest, the fix is operational, not promotional.
That is the standard Tany builds to — branded iOS and Android ordering apps on your existing Square POS, with order-ahead, self-running loyalty, eGift cards, and push, live in about a day at $99 CAD/month per location. Compliant review prompting is part of shipping an app properly; if you are evaluating vendors, it is a fair thing to ask about.