Full description
Expert: with sources and numbers. Plain: the same, without them, for beginners.
4,000 indexed characters, and the field where Google states most directly that keyword stuffing is against policy.
The main indexed body text on Google Play, and the field where the difference from iOS is largest. Apple does not name the description as a search input. Google does.
The rules
A The full description is limited to 4,000 characters. (Create and set up your app, verified 2026-09-18)
A It is part of the metadata Google uses to identify what an app is. (App Discovery and Ranking)
A Google’s guidance: “Make sure your app’s description accurately describes its functionality and content” and “Keep your app’s description succinct and straightforward.” (Best practices, verified 2026-09-18)
Google tells you not to keyword-stuff, in its own words
This is unusually direct for platform documentation and worth having verbatim.
A Google advises avoiding “excessive length, detail, improper formatting, and repetition of words”, and gives an example of what not to write:
“Car racing, car driving, race cars, car races, race track, driving, drive, race, cars, vehicles, automobiles, trucks”
Instead: “Provide a well-written, succinct description of your app and what makes it special.” (Best practices, verified 2026-09-18)
A The Metadata policy makes it enforceable, prohibiting “misleading, improperly formatted, non-descriptive, irrelevant, excessive, or inappropriate metadata” and “repetitive or unrelated keywords or references”. (Metadata policy, verified 2026-09-18)
A The same policy prohibits unattributed or anonymous user testimonials in the description, along with ranking claims, price and promotional text, and Google Play program badges. Same source.
So the common advice to repeat a target term a fixed number of times across the description is advice to do the thing Google names as a policy violation, using an example Google itself supplies.
The density question
B That Google weighs term frequency in the description to some degree is widely and consistently reported by independent practitioners, and is consistent with Google naming the description as a metadata input. It is not documented, and no reliable figure for an optimal count exists.
C Specific density targets (“repeat your main keyword five times”, “aim for 2–3% density”) are claims in circulation. Single-source or unevidenced, they vary wildly between sources, and following them produces exactly the text Google’s own example warns against.
The workable position: use your important terms naturally where they belong, in text a person would want to read. If a term appears often, it should be because the app genuinely is about that thing. Writing to a density number optimises for a mechanism nobody has demonstrated, against guidance Google has published.
How to work with it
Front-load. Only the first lines show before “Read more”. Whatever earns the expand goes there.
Short paragraphs and restraint. Basic formatting is supported, but Google’s guidance is against improper formatting and excessive detail. Walls of text and bullet lists forty items long fail on both counts.
Structure it around use cases. Natural coverage of the way people describe what they want, which also happens to be how they search.
Say what is free and what is paid. Removes a large class of one-star reviews.
Localize it. It is 4,000 characters of indexed text; a machine translation is both a ranking and a credibility problem, 30-05.
What goes wrong
Keyword lists at the bottom. The single most common Play metadata violation, and Google’s documentation contains the counter-example.
Importing the iOS “first 245 characters” rule. That claim is wrong on iOS (CONFLICT-002) and only loosely descriptive on Play, where what matters is that the visible portion earns the expand, not a specific character count.
Unattributed testimonials. “Users love it!” with no attribution is explicitly prohibited.
Copying the iOS description across. It was written for a field that is not indexed. See 20-08.
Never revisiting it. It is not release-bound. There is no structural reason for it to be three years old.
In short
Every fact on this page in one line, with its evidence tier, source and verification date, 4 A · 1 B · 1 C. Generated from the claim register, so it cannot drift from the page above.
The main body of writing the desk actually reads on this side, and the place where the difference from the other centre is largest. On that side, the desk doesn't name the leaflet as something it reads. On this side, it plainly does.
The rules
It's long, a generous, fixed amount of room.
It's part of the writing the centre uses to work out what your shop actually is.
The centre's own advice: make sure it accurately describes what the shop does, and keep it succinct and straightforward.
The centre tells you, in its own words, not to stuff it with words
This is unusually direct for a centre's own guidance, and worth reading closely.
The centre advises against excessive length, excessive detail, poor formatting, and repeating words. And gives its own example of exactly what not to write: a string of near-identical phrases about the same thing, one after another, with nothing but the wording changed. Its own advice instead: write a well-made, honest, succinct account of the shop and what makes it worth a visit.
Its own rulebook backs this up directly, banning writing that's misleading, badly formatted, non-descriptive, irrelevant, excessive, or inappropriate. And specifically banning "repetitive or unrelated words or references." The same rulebook also bans unattributed customer quotes, ranking claims, prices, and any suggestion the centre itself has blessed you.
So the common advice to repeat a target word a fixed number of times through the leaflet is advice to do the exact thing the centre names as against the rules, using an example the centre itself supplies.
The repetition question
That the desk weighs how often a word appears here, to some degree, is widely and consistently reported by experienced shopkeepers, and it fits with the centre naming this leaflet as something it reads. It isn't documented, and no reliable number for how often is "right" actually exists.
You'll hear specific numbers, repeat your main word a set number of times, aim for a certain density. These are unevidenced or single-source, they disagree wildly with each other, and following them produces exactly the text the centre's own example warns against.
The workable position: use your genuinely important words naturally, exactly where they belong, in writing a person would actually want to read. If a word shows up often, it should be because the shop is honestly, deeply about that thing. Writing to a target number optimises for a mechanism nobody has actually demonstrated, against guidance the centre has published in plain words.
How to work with it
Front-load it. Only the first few lines show before a visitor has to expand it. Whatever earns that click goes there.
Short paragraphs and restraint. Basic formatting works here, but the centre's own guidance is against poor formatting and excessive detail. A wall of text and a forty-item bullet list both fail on the same count.
Structure it around what people actually want to do. Natural coverage of the way people describe their own need, which also happens to be how they ask the desk for it.
Say plainly what's free and what costs extra. Removes a whole category of angry write-ups afterward.
Translate it properly. It's a long piece of indexed writing; a rough, machine-made translation is both a filing problem and a trust problem.
What tends to go wrong
A list of loose words at the bottom. The single most common house-rule violation on this side, and the centre's own guidance contains the exact counter-example.
Importing a rule from the other centre about a magic opening length. Wrong on that side to begin with, and only loosely true here, where what actually matters is that the visible portion earns the click. Not any specific number of characters.
Unattributed customer quotes. "People love it!" with nobody's name attached is explicitly banned.
Copying the other centre's leaflet straight across. It was written for a leaflet the desk there doesn't read at all.
Never revisiting it. It isn't locked to a rebuild. There's no structural reason for it to still say what it said years ago.