ASO is often presented as keyword research. Keywords matter, but they only determine whether you are seen. Everything after that determines whether you are installed. This checklist is ordered by impact.
Before anything: know your two numbers
Two metrics explain nearly every ASO decision:
- Impressions — how often your listing was shown. Keywords and category move this.
- Conversion rate — the share of viewers who install. Icon, screenshots and title move this.
If impressions are low, work on discoverability. If impressions are healthy but conversion is poor, more keywords will not help — your listing is failing to persuade. Diagnosing which problem you actually have first can save months of misdirected effort.
Keyword research
- Start from how users describe the problem, not from what you built. People search "split bill", not "expense reconciliation". - Mine competitor listings for terms you have not considered, especially titles and subtitles, where the strongest terms live. - Prefer specific phrases with less competition early on. Ranking third for a modest term beats ranking fortieth for a broad one. - Check the store's own autocomplete. Suggestions reflect real query volume. - Remember the stores differ: Apple has a dedicated keywords field; Play indexes your descriptions instead.
Metadata
- Title under 30 characters, containing your most valuable phrase. - Subtitle (Apple) covering a different benefit, not a restatement. - Short description (Play, 80 characters) written as a complete, persuasive sentence. - Keywords field with no duplicates of title words, commas without spaces, no plurals, no "app". - Long description written twice — keyword-aware for Play, persuasion-first for Apple, which does not index it. - Category chosen for where users browse, not where you feel you belong. A less competitive secondary category can be worth more than a crowded primary one.
Visual assets
This is where most listings quietly lose installs.
- Icon: legible at 48 px, distinct in silhouette, not a shrunken logo with text in it.
- First two screenshots: each carrying one benefit-led headline of seven words or fewer.
- Consistent visual language across all screenshots — one background palette, one type family.
- Feature graphic (Play, mandatory): 1024 x 500, readable at quarter size, centre kept clear for a possible play button.
- Preview video (optional): the first three seconds must show the app doing something, because most viewers stop there.
- Localised assets for your top non-English markets, at least the first two screenshots.
Ratings and reviews
Ratings affect both ranking and conversion, and they compound over time.
- Prompt at a moment of success, not on launch. After a task completes, not while one is starting.
- Use the native review APIs, which do not interrupt the user or push them out of the app.
- Never prompt twice in a session, and respect the platform's own rate limits.
- Reply to negative reviews, especially ones describing a bug you have since fixed. Some users update their rating, and prospective users read your replies as evidence the app is maintained.
- Recover from a bad release quickly. Recent reviews are weighted more heavily, so a fast fix repairs the average faster than you would expect.
Technical factors people forget
- App size. Large downloads reduce install completion on mobile data. - Crash rate. Both stores factor stability into ranking, and it drives one-star reviews directly. - Fast, honest onboarding. Retention feeds back into ranking, and a listing that oversells produces installs that churn within a day. - Accurate permissions. Requesting more than you need visibly reduces installs.
The post-launch loop
ASO is not a launch task. A sustainable cadence:
- Weekly: scan new reviews for recurring words. They are free copy research — the phrases users repeat are the phrases your listing should use.
- Monthly: check keyword rankings and conversion rate. Change one thing.
- Every release: update screenshots if the UI changed. Stale screenshots showing an old interface actively cost installs.
- Quarterly: re-check store requirements. Sizes and rules change, usually with little warning.
Common mistakes worth naming
- Optimising keywords while the real problem is conversion. - Changing five things at once and learning nothing from the result. - Writing for search engines rather than for the person reading. - Ignoring the smaller store because it is smaller. - Treating the first screenshot as an afterthought when it is the highest-leverage asset in the entire listing.