Guide · Ranking factors

Retention and reviews: the ranking factors you cannot write into a keyword field

Almost every ASO guide stops at the fields you can type into: title, subtitle, keyword field, screenshots. Those are the parts a tool can change on a Tuesday afternoon. The inputs that decide whether the change sticks are the ones you cannot edit at all, and they are about what happens after the install.

The loop nobody draws

Visibility gets people to your page. Your page gets them to install. What happens next decides whether the store keeps sending you more of them. An app that people open twice and delete is, from the store's point of view, a bad answer to the query that surfaced it, no matter how well the keyword field was written.

This is worth being precise about, because the topic attracts confident claims. Apple has never published its ranking factors, so any specific weighting you read is inference from correlation, and the guides quoting percentages are guessing with conviction. The safe reading is that retention is an input you cannot edit, not a lever with a published exchange rate. It is also the rare ASO topic where you do not need the ranking question settled to act, because the same number decides whether your acquisition maths works at all.

The windows that get watched

Retention is usually measured at day 1, day 7, and day 28, sometimes day 14. Those windows matter to you regardless of what Apple does with them, because they are where a subscription business is won or lost. A keyword that delivers installs which all churn inside a week is not a cheap keyword, it is an expensive one that has not finished charging you yet.

Uninstalls inside the first 30 days are the harshest version of the same signal. If a segment of your traffic uninstalls at a noticeably higher rate than the rest, the interesting question is which keywords sent it. Broad high-volume terms tend to produce exactly this: plenty of installs from people who were looking for something adjacent to what you built.

Review recency beats review history

A rating average is a lagging number, and Apple appears to care more about the recent slice of it than the lifetime figure. Five stars from three years ago do not offset a month of one-star reviews about a crash on launch. In practice this means a bad release can cost you visibility within days, which is faster than any metadata change has ever helped anyone.

The response that works is unglamorous: ship the fix quickly, then reply to the people who reported it. Replies are public, they are read by the next person deciding whether to install, and a visible pattern of responses changes how a run of bad reviews reads. A short honest reply from the developer beats a polished one from anybody else, so the only real problem is writing them at all: if typing is what stops you, a local dictation tool like JustGhostWriterlets you speak a reply and paste it, without the audio leaving your Mac.

Conversion quality, not just conversion rate

Custom product pages started appearing in organic search results for specific keywords in July 2025, and the active limit rose to 70 per app in early 2026. That is usually framed as a conversion story, and it is, but the more useful framing is fit. A page built for the query a person actually typed sets an accurate expectation of what they are about to install, and expectations set accurately are the cheapest retention work available.

The inverse is worth stating plainly, because it is the trap: a screenshot set that oversells will lift install rate and depress everything after it. Conversion rate is easy to measure and easy to game. What it costs you shows up four weeks later in a number most people never join back to the keyword that caused it. Our guide onwhich keywords deserve a custom product pagegoes through the selection side.

Where paid fits

Apple Search Ads and organic performance are related, though not in the tidy way the "ads boost your ranking" claim suggests. Nobody outside Apple can prove a direct causal link, and you should be suspicious of anyone who sells you one. What is defensible is smaller and more useful: ads buy installs and conversions you would not otherwise have had, and those are the same behaviours organic visibility responds to.

You do not need to settle that debate to act. The answerable question is whether a given keyword returns more than it costs, which needs your rank, your spend, and your revenue joined on the same keyword and the same window. That join is whatturns a keyword list into a decision.

Give changes time to mean anything

Three to four weeks. The App Store re-indexes the listing, then observes how people respond to it, and both halves take time. A change judged after three days is measuring noise, and a listing rewritten weekly can never be attributed to anything, because there is no stable period to compare against.

This is where release markers earn their keep. When a rank move lines up with the day you shipped a version, the version is usually the explanation, not the keyword work you did the same week. Separating the two is most of the analysis.

The one shortcut that ends the account

Do not buy reviews. Apple runs detection on rating patterns, and the penalty is not a ranking adjustment you can recover from with better work. Everything else on this page is slow and compounding. That one is fast and terminal.

What to actually do this month

• Find which keywords send the traffic that leaves. Start with your broadest, highest-volume terms.
• Read the last 30 days of reviews before you read the lifetime average.
• Reply to the negative ones, in the reviewer's language, and ship the fix first.
• Mark your release dates on any rank chart you look at, so version effects stop being read as keyword effects.
• Change metadata once, then leave it alone for a month.

Where Veldo fits, and where it does not

Veldo tracks organic rank across 60+ storefronts, joins it to your Apple Ads spend and your RevenueCat revenue per keyword, marks your releases on the charts, and notifies you when new reviews land. That covers the detection half of this page: which keywords moved, when, around which release, and whether the money followed.

It does not measure retention. Day 7 and day 28 curves live in App Store Connect and in your subscription analytics, and Veldo does not pull them today. Anyone telling you their ASO tool measures your retention is either reading a different number or reading yours from somewhere you already have. Being clear about that boundary is more useful than pretending the line is somewhere else.

Common questions

Does retention affect App Store ranking?

Apple has never published its ranking factors, so nobody outside Cupertino can prove a weighting, and you should distrust any guide that quotes one. What is widely observed is that apps which hold their users tend to hold their positions. Treat retention as an input you cannot edit rather than a lever with a known exchange rate, and note that it decides your economics either way.

Can a bad update hurt my search visibility?

It can, and faster than a metadata change helps. A release that breaks something produces a run of low ratings, and recent ratings carry more weight than old ones, so a five-star history from three years ago will not protect you. The fastest recovery is usually a fix shipped quickly, not a metadata rewrite.

How long should I wait before judging a metadata change?

Three to four weeks. The App Store needs to re-index the listing and then observe how people respond to it, so a change judged after three days is mostly noise. Rewriting weekly guarantees you never learn which change did anything.

Do Apple Search Ads help organic ranking?

Not directly, and anyone selling you a direct causal link is guessing. The honest version is that ads buy you installs and conversions you would not otherwise have had, and those are the same behaviours organic ranking responds to. The measurable question is whether a keyword returns more than it costs, which you can answer without settling the ranking debate.