Do Ratings Change App Store Search Rank?

A meditation app at 3.8 stars is not “dead in search.” It is a product page with a visible objection. Shoppers see the number before they see your breathing timer. Some of them leave. That loss is a conversion problem you can measure. It is not a rank formula Apple has published, and inventing one (“each tenth of a star is worth N places”) is how indie posts go wrong.

Apple’s metric definitions define conversion rate as total downloads and pre-orders divided by unique device impressions. A pre-order counts once, and it is not counted again when the app later downloads. If a low average scares people off the product page, you will see it there. How to ask, and how to answer the people who already wrote, are separate jobs: how to ask for an App Store review and how to reply to App Store reviews.

Separate search rank from the download decision

Search rank is where the app shows up. Conversion is whether the person who saw it installs or pre-orders. Apple indexes name, subtitle, and the keyword field for discovery. Apple does not publish a ranking weight for the star average, the rating count, or the share of 1-star reviews.

The average sits next to the name in search results and on the product page. A 3.8 on a meditation app tells a tired person that other people had a bad time. Some will still tap. Some will not. Those misses shrink downloads relative to unique device impressions, which is Apple’s conversion rate. You do not need a secret rank weight to justify fixing the thing people are rating.

Read the 3.8 as a distribution, not a grade

An average hides the shape. The worked example below is an illustration for this meditation app, not a store export and not a category benchmark. One hundred ratings total 380 stars, so the average is 3.8. The useful question is which bucket the angry people are in, and whether their text names the same bug.

Forty-five 5-star ratings and ten 1-star ratings can average to 3.8. The ten 1-star notes are the work list; the average is only the headline shoppers see.

Check the arithmetic before you trust a screenshot of the page: 225 + 80 + 45 + 20 + 10 = 380, and 380 divided by 100 is 3.8. If your real histogram is a pile of 5s and a pile of 1s, you do not have a “meh” app. You have a loved app with a sharp failure. If the bars are a mound around 4, the 3.8 is a slow leak of small disappointments, and a single hotfix will not move the headline as fast.

What the 1-star cluster usually is

Sort the 1-star and 2-star text, not the stars. On a meditation app the repeating sentences tend to be concrete: the session ended with an ad, the reminder fired at 3 a.m., the annual plan was harder to cancel than the paywall implied, or audio died when the phone locked. Those are product bugs and promise bugs. They are not an ASO tuning problem.

People in pain write first. A bad build can collect a handful of furious notes before the calm majority bothers to rate. Reply with the version that fixes the named bug, and put that fix in What’s New so the next shopper sees you already shipped it.

What not to do when you want the number up

Two shortcuts show up in every indie Slack. Both are a bad trade. Incentivized reviews — a week of premium, a gift card, a shout-out — buy the wrong words and the wrong risk. Review gating is the other one: a custom “Are you enjoying the app?” step that opens the system prompt only for the happy path and hides everyone else behind a feedback form. That filter is a known rejection pattern. It also trains you to ignore the people whose 1-star notes are the actual bug list.

Apple’s Requesting App Store reviews says the system displays the review prompt a maximum of three times within a 365-day period, and that people can turn requests off entirely. You choose a calm moment after a finished job. You do not get to pre-sort who is allowed to see the sheet. The honest version of “get more ratings” is the request post: call the system API after a completed session, not after a smile survey.

Reset the overview rating only when the old number is a lie

Apple lets you reset the overview rating when you release a new version. The reset help is specific. On the version you are submitting, there is a Reset Overview Rating section; you select reset when this version is released. After release, the product page says the rating was recently reset. A new overview rating appears once enough people have rated that version. Apple does not publish how many ratings “enough” is, so do not plan a launch around a guessed count.

The reset is not a delete button for history. Written reviews stay on the product page. You cannot restore the previous rating after the version is out. The reset hits every region at once. For a universal app it applies only to the platform version you reset, and you cannot reset an app bundle. Use it when a fixed bug left a headline that no longer describes the app, not as a ritual every time you ship a minor build. If you reset and the same reminder bug is still in the binary, the new average will find the same hole.

What to change on the page while the average moves

While the 1-star cluster is being fixed, the product page still has to tell the truth. If reviews say the free session is three minutes and the screenshots imply an endless library, the average is doing the job your screenshots refused to do. Match the first screenshot to the session a new person actually gets. Put the subscription boundary in the description, not only in a modal after the second night. A 3.8 with honest framing converts better than a 3.8 that looks like a bait.

Promotional text can change without a new version, and it does not affect ranking. A one-line “reminder bug fixed” note belongs there while you wait on the binary. It will not rewrite the star average, and neither will a keyword edit. Spend the version on the bug, the public reply, and a screenshot that matches the session people actually get.