Why ratings and replies are conversion work
The rating appears where the decision is made. In store search results, a person compares the icon, the name, the first screenshots and the star rating of several apps at once, and a weaker rating loses some of them before the listing has been opened. On the listing, reviews are Social ProofEvidence that other people have chosen, used or approved something, used to reduce a buyer's doubt.: the closest thing to a friend's opinion that a stranger has. Both affect the listing's conversion rate (CVRShare of visits or clicks that convert.), and a better listing conversion lowers the cost of every paid install that lands there, which is why the app stores page for app installs puts ratings timed to success among the store tactics.
The levers are narrower than on the web. The ASO guide explains that both stores provide a system rating prompt that can be shown inside the app, that Apple's review guidelines direct developers to use it rather than a custom one, and that both stores prohibit incentivising ratings or filtering who is asked by how happy they look. A pre-question that sends satisfied users to the store and unhappy ones to a feedback form is exactly that filter. The legitimate levers are timing, quality and replies: ask at the right moment, fix what reviews complain about, and answer in public.
| Store behaviour | As the ASO guide states it | What the routine does with it |
|---|---|---|
| System rating prompt | Both stores provide one; Apple's guidelines direct developers to it | Wire the prompt to a chosen moment; build no custom rating dialog |
| No incentives, no filtering by mood | Both stores prohibit them | Ask everyone who reaches the moment; offer nothing for a rating |
| Public replies | Both consoles let the developer reply publicly | A reply owner, a triage rhythm and reply standards |
| Summary rating reset | Apple allows a reset when a new version ships | A deliberate decision after a fixing release, not a habit |
| Recency and country | Play ratings are weighted towards recent reviews and shown per country | Track the rating by version and country, not as one number |
Whether a requested prompt is actually displayed is decided by the operating system, not by the app. Nothing in this tutorial depends on how often a prompt can be shown; check the current App Store and Play documentation for the rules that apply when you build it.
Before you start
Gather four things. First, the app's event data, so you can see which actions retained users complete early; the deep links and app analytics guide describes the event taxonomy, including the activation event that marks a new user's first experience of the app's core value. Second, access to the ratings and reviews sections of App Store Connect and Play Console for everyone who will reply. Third, a place to tag and count reviews, which can be a shared spreadsheet at first. Fourth, a named owner for each part of the routine.
Write the owners into a RACI matrix before the first reply goes out: who is accountable for the prompt's placement, who replies, who decides on escalations, who receives bug reports, and who decides on a reset. Without it, replies stop the week the enthusiastic person goes on holiday.
| Decision or task | Responsible | Accountable | Consulted | Informed |
|---|---|---|---|---|
| Choose and change the prompt moment | Product manager | Head of product | Growth, analytics | Support |
| Read, tag and reply to reviews | Support lead | Head of support | Product, marketing | Growth |
| Route bugs and requests to product | Support lead | Head of product | Engineering | Marketing |
| Track the rating by version and country | Growth analyst | Head of growth | Product | Leadership |
| Decide on a summary rating reset | Head of product | Head of growth | Support, marketing | Leadership |
The steps
Choose the moment of completed value
Pick one point in the app where the person has just got what the app is for: a workout logged, an order delivered, a lesson finished, a goal reached. Prefer a moment that most retained users reach early, which the event data shows; the activation event, which the tutorial on finding the app activation event identifies from retention data, is often the right one or close to it. Rule out first launch, onboarding, anywhere after an error or a failed payment, and the middle of a task. The ASO guide's rule is short: after a moment of completed value, never at launch, never during an error.
Wire the system rating prompt to it
Ask the developers to request the stores' own rating prompt at that moment, not a custom dialog and not a link that opens the store. Remember that the operating system decides whether a request results in a prompt on screen, so the app controls when it asks, not what the person sees. Log each request as an analytics event, so you can see how often the moment is reached and compare ratings before and after a change to it.
Set the rules on who is asked and how often
Write the rules down. Everyone who reaches the moment is eligible; nobody is excluded because they seem unhappy, have contacted support or gave a low score in a survey, because filtering by mood is what the stores prohibit. Nothing is offered in return for a rating. Do not ask again soon after someone has been asked, and leave a long gap after a decline. Set the gap from your own judgement and keep it in the rules document, so a later change is deliberate. The same rules apply when an app business also asks for reviews by email or in person, which the review request launch checklist gates across every channel.
Read every new review and tag it
On a fixed rhythm, daily for an app with a steady flow of reviews and weekly for a small one, read every new review in both consoles and tag it: bug, feature request, praise, question, billing or account, abuse or spam. Record the app version and country each review mentions or the console shows. The tags are what turn reviews into a report; the voice of the customer guide explains how such feedback joins survey and support data into one listening programme.
Reply to problems with a fix or a next step
Both consoles let the developer reply publicly, and the ASO guide notes that a reply is visible to everyone who reads the review afterwards. Reply first to reviews that describe a problem. A good reply thanks the person, names the problem in their words, says what has been done or what happens next (a fix in a named version, a workaround, a support address), and never argues or reveals account details. The service recovery and complaints guide sets the tone for public complaints, and the local reviews and reputation guide covers replies to unfair or fake reviewsA review not based on a genuine experience, such as one written by the business, paid for without the experience or bought from a service., where a calm factual correction, or the store's own reporting route if the console offers one, does more than a public argument.
Route bugs and requests to product with the app version
Every review tagged as a bug goes to the product team's tracker with the review text, the app version, the platform, the country and how many reviews mention the same thing. Feature requests go to the product backlog with the same counts. Where requests also arrive from sellers, support and business customers, the tutorial on handling feature requests from sales and customers merges them by the job behind each one so that review requests join the same record. When a fix ships, go back to the reviews that reported it and reply again with the version that fixes it; people sometimes update their rating, and every later reader sees that the problem was heard.
Thank and answer the rest briefly
Praise, questions and neutral reviews deserve short, specific replies when time allows: a thank-you that mentions what the person liked, or a direct answer to the question. Skip templates that read the same on every review; prospective users notice. Do not reply to abuse or spam; check the current console documentation for how each store handles reports of it.
Track the rating by version and country
Keep a weekly table of the rating and review count by app version, platform and country, with the tag counts beside it. The ASO guide notes that Play ratings are weighted towards recent reviews and shown per country, so improvement shows faster on Play and differs by market; a single global average hides both. Read small counts with care: a version with few ratings can swing on a handful of reviews.
Decide whether a reset is earned
Apple allows a developer to reset the App Store summary rating when a new version ships. Decide on it only after a release that fixes what the old reviews complained about, and only once the new version's early ratings and tagged reviews show the complaint has gone. The ASO guide warns that a listing reset to no ratings can convert worse than a mediocre score, so a reset after a cosmetic release costs more than it gains. Record the decision and the reason either way.
Review the routine each release
At each release, check the prompt still fires at the chosen moment and nowhere else, that the moment still makes sense if the flow changed, that the reply owners are still the right people, and that bugs reported in reviews were answered when fixed. Add the prompt and reply checks to the release list; the app store listing checklist already asks for a prompt after a moment of success and replies to recent reviews before any listing update. The app marketing operations playbook sets this routine beside the weekly review replies, release notes and deep link tests that every release carries.
A habit tracker's review routine (illustrative)
Tasselwick Habits moves its prompt and answers a sync bug
Tasselwick Habits, an invented habit-tracking app, has asked for a rating on first launch since release. Its App Store rating stands at 3.6 from 2,140 ratings, and the latest reviews are dominated by one complaint: check-ins logged on a watch do not sync to the phone. The team moves the rating request to the moment a user completes a seventh check-in, which its data shows most users who are still active a month later reach in their first fortnight. Support starts reading reviews daily, tags 46 one-star reviews from the past month as the sync bug and replies to each: an apology, the cause in plain words and the version that will fix it. Version 5.3 ships the fix three weeks later, and support replies again to every review that reported it; 11 of those reviewers raise their rating. The new version's rating over its first six weeks is 4.4 from 610 ratings, with sync complaints down to two. Because the summary score has already risen to 3.9 and the older reviews now carry replies showing the fix, the team decides not to reset, records the reason and reviews the decision at the next release.
The way ratings sit beside reviews, testimonials and other proof on a website is covered in the reviews and social proof guide; the store-specific part is that the rating appears where the install decision is made. The app marketing hub shows where this routine sits beside listing and paid work, and the glossary entry on social proof sums up why it moves conversion. Where the app is one of several places customers leave reviews, the customer reviews and reputation audit scorecard rates app replies alongside the web profiles and the business's own pages, so one routine can be judged across all of them.
Common mistakes
- Asking at first launch. The person has not yet had any value from the app. Ask after a moment of completed value.
- A happy-path gate. Asking "do you like the app?" and sending only the yeses to the store is filtering by mood, which the stores prohibit.
- A custom rating dialog. Apple's guidelines direct developers to the system prompt; use the stores' own.
- Offering anything for a rating. Incentives are prohibited by both stores, however small.
- Replying only to praise. Complaints are what prospective users read most closely; answer them first.
- Replies that argue or template. Short, specific, calm replies with a fix or a next step help the next reader.
- Bug reports that stop at support. Without the app version and a route to product, the same review keeps arriving.
- One global rating. Play ratings are recency-weighted and per country; track by version and market.
- Resetting after a cosmetic release. A reset only helps when the cause of the old complaints has gone.
