Keyword research is not finding one magic phrase and pasting it into App Store Connect. It is a weekly system for learning how users search, which competitors Apple already trusts, where your app can realistically rank, and what deserves space in your title, subtitle, and keyword field.
The mistake most founders make is starting with a tool score. The better starting point is the user. What are they trying to do? What words would they type before they know your app exists? Which result would make them stop scrolling?
Build the first seed list
Start with the app’s job, not the app’s features. A habit tracker is not only “habit tracker”. It may be “daily routine”, “goal tracker”, “streaks”, “quit vaping”, “adhd planner”, or “morning routine” depending on the user’s reason for opening the store.
Use four seed sources: app name and subtitle, core benefits, user review language, and competitor titles/subtitles. Do not trust any source alone. Your own words are often too founder-centric. Competitor words may be crowded. User words are messy but usually closer to intent.
A useful seed list has categories, not just keywords. Keep separate buckets for category terms, job-to-be-done terms, pain terms, audience terms, feature terms, competitor-adjacent terms, and country-specific terms. That structure prevents the list from turning into one giant comma-separated mess.
Expand from the store
Autocomplete and related search behavior are useful for expansion because they expose phrases people may type. They are not proof of exact volume. Treat them as candidate generation, then validate each candidate against live search results.
A good expansion pass should produce branded terms, category terms, use-case terms, pain terms, feature terms, and long-tail combinations. The goal is not to track everything forever. The goal is to create enough candidates to identify reachable pockets.
Do not skip long-tail keywords just because they look smaller. Long-tail phrases are where early apps often find their first useful rank movement. A broad word might be the dream, but a narrower phrase can teach you which promise converts.
Score relevance before popularity
Relevance should be the first filter. If the user searching the phrase would not be satisfied by your first screen and your core paywall promise, the keyword does not belong in your title or subtitle.
Popularity comes second. A keyword with meaningful demand but weak relevance creates bad conversion. Bad conversion tells Apple your app is not the right result.
Use a simple relevance label before looking at any score: exact fit, close fit, possible fit, or wrong fit. Exact-fit and close-fit terms can go into metadata tests. Possible-fit terms usually belong in screenshots, Custom Product Pages, Apple Ads discovery, or a watchlist. Wrong-fit terms should be ignored even when they look popular.
Treat popularity as a signal, not a forecast
Apple does not publish exact organic App Store search volume. When a tool shows popularity, volume, traffic, or downloads by keyword, it is usually turning available signals into a model. That can still be useful, but it should not become fake precision.
Bucket demand into low, medium, and high. Then ask the harder question: can this app win enough of that demand? A medium-demand keyword where your app can reach top results is often more useful than a huge keyword where the first screen is full of entrenched apps.
Read difficulty from the top results
Difficulty is not a mystical number. It is a summary of the current result page. Look at the top apps’ rating count, rating quality, review velocity, title match, subtitle match, brand strength, freshness, category fit, and screenshot clarity.
A result page with small apps, partial title matches, stale screenshots, weak ratings, and inconsistent localization is beatable. A result page with exact-match category leaders and massive rating velocity is not a good first target.
This is also where competitor research becomes useful. You are not trying to copy the winner. You are trying to understand why Apple may be ranking that app and whether there is a visible weakness you can beat: clearer metadata, better localized screenshots, stronger ratings, fresher positioning, or a sharper use case.
Check country differences
The same keyword can be crowded in the United States and reachable in Canada, the United Kingdom, Germany, Brazil, or Japan. Do not assume global difficulty from one storefront. Country scans are useful when they explain where to test first.
This matters most for apps with broad international relevance: productivity, health, fitness, education, photo/video, finance, and utility apps.
Country work is not only translation. It changes the result page. A keyword can have different competitors, different buyer expectations, different rating standards, and different screenshot conventions. If you only inspect the US storefront, you miss easier markets.
Place keywords intentionally
The title is for the strongest identity plus the strongest search theme. The subtitle is for the second cluster. The keyword field is for unique words not already covered. Repeating the same word across all fields wastes space.
Metadata should read like a coherent product page. Apple’s ranking systems and users both respond better when the title, subtitle, screenshots, and app experience point in the same direction.
A clean first version is usually better than a clever one. Pick one primary cluster for the title, one supporting cluster for the subtitle, and keep the keyword field for useful unique words. If the page tries to rank for everything, it usually feels vague to everyone.
Use screenshots to support the keyword
Metadata can make you visible, but screenshots help the searcher decide whether the app is the answer. If you target “meal planner”, the first screenshots should make planning meals obvious. If you target “invoice maker”, the first screenshots should show invoices, clients, payments, or export.
This is where many ASO workflows break. They update keywords, but the product page still speaks in generic feature language. The better workflow is to connect each serious keyword cluster to a visible promise in the first few screenshots.
Track after every change
Do not change metadata every day. Batch a focused update, wait, and watch rank history, search impressions, product page views, downloads, and conversion. If rank improves but conversion drops, the keyword may be too broad. If conversion improves but impressions are flat, the listing is clearer but not reaching more searches yet.
The weekly loop is simple: discover, prioritize, update, measure, and repeat. The discipline is what compounds.
A practical weekly workflow
Once a week, pick one small decision instead of reopening the whole strategy. Add a few candidate keywords, check live results in the most important countries, mark which competitors changed, choose one metadata or screenshot hypothesis, and write down what should happen if you are right.
After the update ships, watch movement without overreacting to one day. App Store results move for reasons outside your app too: competitor updates, Apple ranking changes, paid traffic mix, review velocity, and country-level volatility. The point of rank history is to see the pattern, not panic over every scan.