App store keyword research is the first stage of a seven-stage cycle: research, organize, select, apply, monitor, optimize and report. You collect every candidate search term from people and data, clean and group them, select the few worth scarce metadata space, write them into indexed fields, then track ranks and search conversion and loop back to whichever stage is weakest.
Key takeaways
- Keyword research is not a one-off export. It is stage one of a seven-stage loop, and later stages regularly send you back to it.
- Every candidate keyword must come from a keyword data source, not a brainstorm.
- Score candidates on volume, competition, relevance, length and copy capacity, then replace guesses with Apple Ads data once you have it.
- Validate every metadata update with a before-and-after comparison before calling a winner.
- Research the fields around ASO too: psychology, competitors' business models and platform politics.
What is app store keyword research, really?
It is the work of documenting every search term real people might type to find an app like yours, so that you can later pick a finite set that reaches the most users who are most willing to download. A keyword is a search term used by real humans, and its search volume is the size of a market.
That definition rules out two habits I see often. The first is writing metadata straight from a raw keyword list. The second is inventing keywords in a meeting. Both skip the evidence, and both waste characters you only have a few of: on iOS, 160 indexed characters across title (30), subtitle (30) and keyword field (100); on Google Play, a 30-character title, an 80-character short description and a 4,000-character long description.
What are the seven stages of keyword optimization?
The seven stages are research, organize, select, apply, monitor, optimize and report. I call this the Seven-Stage Keyword Optimization Flow, and its defining feature is that any later stage can send you back to an earlier one.
- Research. Gather ideas from people, keyword reports, synonyms, store auto-suggest and keyword tools.
- Organize. Clean the list (data hygiene), categorize it, set priority and hierarchy, and compare terms side by side.
- Select. Pick by past performance, competition weight, combinations and perceived potential.
- Apply. Write the picks into App Store and Google Play text, and into listing experiments where the store allows it.
- Monitor. Track keyword ranks, search impressions, conversion rate from search and visibility growth.
- Optimize. Expand the pool, automate what repeats, and iterate.
- Report. Record what moved and why, then loop back to whichever stage needs work.
At Appkitekt we also use a shorter version of the same idea, the Keyword Optimization Workflow: research and document the full pool, split it into three priority groups, assemble the groups into title, subtitle, short description and keyword field, validate with a pre-post comparison, then iterate on the test method, the assembly, the selection or the research. A pre-post analysis compares equal periods before and after a change to estimate its real effect once it is live.
Where do the keyword ideas come from?
From both people and data, and you need both. Human input tells you how users describe their problem; data tells you how many of them type it.
The idea sources I work through:
- Your own product knowledge and your peers.
- Market and competitive analysis.
- The research functions in a keyword tracker.
- Store auto-suggest and a thesaurus for synonyms.
- Google Keyword Planner and Apple Ads keyword data.
- AI chats, for phrasing you would not think of.
Then every idea goes through a keyword data source. Nothing stays on the list because it "sounds right".
How do you score keywords before you have data?
Score each candidate on five things: volume, competition, relevance, length and copy capacity (whether the term fits naturally into good copy). I call this approach Score Keywords, Then Prove Them, because the score is a placeholder for evidence you collect later.
| Criterion | What you score now | What replaces it once Apple Ads data exists |
|---|---|---|
| Popularity | Volume score | Impressions |
| Relevance | Guessed install rate | Installs per 1,000 impressions (IPM) |
| Long-run value | Not scored | Forecast return on ad spend (fROAS) |
| Competition | Apps ranking, big brands | Unchanged |
| Length | Characters used | Unchanged |
| Copy capacity | Fits good copy | Unchanged |
IPM is installs per 1,000 ad impressions; fROAS is the revenue users from an ad are predicted to bring over time, divided by the ad's cost. Once you have them, stop scoring relevance by guesswork.
Drop any keyword you cannot write into natural copy. A high-volume term that turns your subtitle into word salad costs more in conversion than it brings in rank. For the full selection logic, see the five tests for choosing ASO keywords.
Why research beyond ASO data?
Because the best keyword and copy decisions borrow from fields that ASO tools do not cover. I call this the Multidisciplinary Research System: before any store task, also research psychology, competitors' business models and platform politics, then blend them with ASO know-how.
For keyword and copy work, that means reading on creative writing, persuasion and behavioral economics, not only on character limits. It also means asking what your teammates studied. Someone with a languages or history background will spot phrasings and associations a keyword tool never surfaces. Turn what you find into a visual map or a bulleted framework the team actually uses.
From the field
In my experience, conversational, psychology-based copy beat a professional writer's polished texts almost every time. In one case it lifted conversion rate by 30%. The keywords were not the difference. The research behind how the words were put together was.
How do you know the research paid off?
Measure along the funnel after the update goes live, in this order: did rankings improve, did organic search traffic improve, did search installs increase, did conversion rate hold, did revenue rise. Conversion rate (CVR, the share of people who saw your app and installed it) is itself a ranking factor, so irrelevant keywords can lift traffic and damage CVR at the same time.
Then report. If a stage underperformed, go back to it: refine the research, the selection, the assembly or the test method. On iOS the keyword field is a live change, so a pre-post comparison is your validation; read more in the rules for the iOS keyword field. On Google Play, changes index slowly, which is why keyword density and patience matter there.
Signs your keyword process is broken
- Your keyword list is a raw tool export, never cleaned or grouped.
- Keywords were picked without documented research.
- The same terms are scattered across fields at random.
- Nobody tracks search impressions or conversion rate from search.
- There is no pre-post analysis after a metadata update.
- Ranks drop and the team starts from scratch instead of looping back to one stage.
- The team only ever looks at ASO-only data.
If several of these sound familiar, the fix is not more keywords. It is a process with a way back. For how often to run that loop, see how often to update ASO keywords.
Frequently asked questions
What are the stages of app store keyword research?
I run keyword work in seven stages: research, organize, select, apply, monitor, optimize and report. Research collects every candidate term, organizing cleans and groups them, selection picks the ones worth metadata space, applying writes them into store text, monitoring tracks ranks and search conversion, and optimizing and reporting decide what to change. Any later stage can send you back to an earlier one.
Where do keyword ideas for an app come from?
From people and from data. Start with your own product knowledge and your peers, then market and competitor analysis, the research functions of a keyword tracker, store auto-suggest, synonyms, Google Keyword Planner, Apple Ads keyword data and AI chats. Every candidate should end up checked against a keyword data source, never kept on intuition alone.
How do I know if my keyword research worked?
Measure it along the funnel after the update goes live. Check whether rankings improved, then organic search traffic, then search installs, then whether conversion rate held, then revenue. Compare equal periods before and after the change (a pre-post analysis) before calling a winner, because ranks alone can rise for terms that never bring paying users.
Should I trust keyword volume scores from ASO tools?
Use them to start, not to finish. Score candidates on volume, competition, relevance, length and copy capacity. Once you have enough Apple Ads data on those terms, replace the guesses with real impressions, installs per 1,000 impressions and forecast return on ad spend. Real install data beats a relevance score you estimated in a meeting.
If you want this seven-stage process run on your own app, it is the core of our ASO strategy service.
About the author
Binh Dang founded Appkitekt after leading ASO at Flo, Kovalee and Jellyfish, where he was Senior Director of ASO, and starting out as an ASO consultant at Phiture. He wrote The ASO Black Book.
See how Appkitekt runs ASO