App Store Metadata: More Edits Don't Mean More Traffic
Most App Store metadata edits are unmeasurable: each change buries the last. Edit one indexed field at a time, keep a baseline, and watch retention.
A team sent me screenshots last month and asked why their rankings had not moved after a dozen rounds of App Store metadata edits. New title, new subtitle, rewritten keyword field, three months of it. I went through their change log. The problem was not that they edited too little. It was that they edited too often: every change buried the result of the one before it, so by the end they had no clean read on any of it.
The short version
Metadata is not a dial. It is a budget. The limits are fixed, the fields that reach search are fixed, and what one edit can teach you is fixed. Editing more does not widen your coverage — it dilutes the cause of every edit into the next one. My take: the edits genuinely worth making in a year are usually fewer than most teams currently ship in a quarter.
Why "change it enough times" fails
Three things stack up.
First, only a few fields reach search. On the App Store side, relevance comes mainly from the name, the subtitle, the keyword field and the primary category, plus behavioural signals such as downloads, ratings and reviews. Apple also states plainly that promotional text does not affect search ranking. Stuff keywords into your description and users can read them; the index cannot.
Second, the indexed fields travel with a build. Name, subtitle and keywords are tied to a version submission. You cannot change them today and read the result tomorrow. So every edit rides along with a release, and a release changes other things — crash rate, review flow, returning users. Two variables move at once and you cannot tell which one did the work.
Third, the sample is too small. Impressions and conversion are noisy, and a three-day curve is pattern-hunting, not analysis.
Change one layer per round
Split what you want to touch into three layers and take them in order.
Index layer: name, subtitle, keyword field on the App Store; app name, short description, full description on Google Play. This layer decides whether you show up for a query at all.
Conversion layer: icon, screenshots, preview video, the first paragraph of the description. This decides whether the people who find you install.
Structure layer: primary and secondary category, markets, language coverage. This decides which pool you compete in.
Touch one layer per round, ideally one field per round. Then watch a full cycle before the next move. It is slow, and every round leaves you a conclusion you can reuse. Change four fields at once and what you get is "it seems a bit better", which nobody can hand over to anyone.
One test before you start: can this edit answer a yes-or-no question? If not, it is a rephrasing, not an experiment.
What each App Store field can do
Name and subtitle run to 30 characters each. The keyword field is 100 characters, comma-separated, and you should not put a space after the commas — every space is a character thrown away from the bank. Those three fields are the main battlefield, and the most frequently fidgeted with.
A common waste: repeating in the keyword field words already in the name and subtitle. The system matches across all that text, so duplicates add no coverage, only spend.
Promotional text is 170 characters, the one field you can change any time without resubmitting — and it does not enter search ranking. Its job is conversion around a moment, not keyword storage. The description is 4,000 characters and is not a search door on iOS; it exists to convince someone who already found you. Reverse those two and you get the classic result: six versions of the description, zero movement in rankings. Check the current official guidance before you ship.
Google Play runs a different loop
Play has fewer fields: app name at 30 characters, short description at 80, full description at 4,000. The real difference is the full description. Google's own help says the text in your store listing helps users discover your app, so the description there is both a discovery surface and a persuasion one. It deserves real writing.
There is a risk running the other way. Google's developer policy states that repeated or irrelevant keywords in the app name, short description or full description hurt the user experience and can lead to the app being removed. On Play, editing more is not merely ineffective — it is exposure. Teams that turn the full description into a keyword list lose conversion and compliance in one move.
What both stores share: character limits are hard caps, and filling them is not the same as writing well.
Localisation is where this gets out of hand
Once you carry a handful of languages, batch editing becomes irresistible: swap the keyword field in a dozen locales in one afternoon. It feels efficient. But each market is its own set of fields and its own search behaviour. Change them together and you control nothing cleanly.
My approach is to move two places at a time: your main market and a smaller test bed. Main market gets only the edits you are confident in. The test bed gets the ideas that might fail. The same concept is spelled and searched differently per market, and machine-translated terms usually look right while nobody types them.
One more thing people miss: name and subtitle have hard caps, and German or Japanese versions routinely blow past them. Nothing truncates for you, so someone has to decide which word gives way. A batch script will happily hand you grammatically correct phrases nobody searches for.
Before you edit, check you're not measuring noise
Three preconditions. Miss one and skip the round.
A baseline: two weeks of impressions, product page views and conversion by market, recorded before the edit. Without it you have nothing to compare.
A single variable: one field in one window. Want to test a new title and new screenshots? Two rounds, not one.
A long enough window: keyword changes do not land the same day. Index updates and review both add lag. Measure in weeks, not hours.
When to leave metadata alone
Right after a release. Wait for the data to settle. The build itself moved things, so editing now means subtracting one unknown from another.
While a native experiment is running — App Store product page optimization, Play store listing experiments. Adding manual edits on top makes the test unreadable, because you can no longer attribute the difference to the variant.
When the sample is tiny. A small market with a few dozen impressions a day cannot support a conclusion. You changed the wording, not the outcome.
When rankings dropped but conversion held. That is usually not a metadata problem. Look outside first: a competitor shipped, a campaign paused, the store changed what it surfaces. Field edits do not fix a demand-side shift.
When nobody keeps records. If the change log does not say what changed, why, and how long you watched, then in six months all you own is a pile of contradicting impressions.
How to tell whether an edit worked
Separate two numbers: impressions, which is your ability to be found, and conversion, which is whether being found turns into installs.
For index-layer edits, look for a structural shift in impressions for the target terms, not a rise in total impressions. Totals bend to seasonality, paid spend and competitors.
For conversion-layer edits, look at product page views to installs, split by market. A good global number can be one large market dragging an average upward.
Then check the number everyone skips: retention. Whether the new arrivals are the right people is a question only retention answers. Sharp screenshots can flatter impressions and conversion. Retention will not.
If you want a second pair of eyes
If you are sitting on four versions of edits and cannot say which one did anything, or your keyword bank has a structural problem — misspellings, competitor brand names, terms unrelated to what the app does — that is not a tuning job. The bank needs rebuilding. We (Youou) build two-store metadata and keyword systems, and we often turn "we keep editing" into "each edit counts once". Bring the change log and the screenshots.
What to do next
1. Turn the last six months of edits into a table: field, date, and whether anyone recorded an outcome. If there is no outcome, treat the edit as unproven. 2. Export a two-week baseline by market: impressions, product page views, conversion. You will compare every future round against this. 3. Audit the App Store keyword field for words already in your name and subtitle, and for spaces after commas. On Play, check whether your full description sacrificed readability to fit terms — repeated keyword stuffing carries removal risk. 4. Pick one field for the next round, set the observation window, and read the data when it closes. Do not touch it twice inside the same window.
Want your app to rank for the keywords that matter?
Get a Quote