Keyword research for apps: a practical workflow

Keyword research for apps has a reputation for requiring an expensive tool and a spreadsheet with forty columns. It does not. The paid tools estimate search volume and difficulty using models that are, at best, directionally correct — and you can get most of the way there with free signals and a bit of discipline. Here is the workflow I actually use.

Step 1: build a seed list from three sources

Start wide. You are collecting candidate phrases, not judging them yet. Three sources cover almost everything:

Store autocomplete

Type your core term into the App Store and Google Play search bars and write down every suggestion. Autocomplete is ranked by real query popularity, so it is the closest thing to free volume data you have. Do it on both stores — their suggestions differ.

Your own reviews

This is the source people skip, and it is the best one. Read your reviews and note the exact words users use to describe what your app does for them. Real users rarely use your marketing vocabulary. If they keep saying “meal planner” and your listing says “nutrition organizer”, you have found a gap. AppBoard surfaces review themes automatically, but reading fifty reviews by hand works fine too.

Competitors

Look at the titles and subtitles of the apps that rank for your core term. They have already done research; their word choices are data. You are not copying — you are noting which phrases the market treats as important.

Step 2: score each phrase on relevance and difficulty

Now narrow. For each candidate, judge two things honestly: how relevant it is to what your app actually does, and how hard it would be to rank. Relevance is non-negotiable — ranking for a term that brings the wrong users just tanks your conversion and, over time, your ranking.

As a new or small app, live in the “winnable” and “long tail” rows. You will not outrank an incumbent for “fitness” this quarter, but you can absolutely own “strength training log for lifters” — and those users convert better anyway because the match is exact.

Judging difficulty without a paid metric

Search the term and look at the top results. Are they big, polished apps with tens of thousands of ratings, or a thin field of smaller apps? Is the exact phrase in their titles, or are they ranking loosely? A term where the top apps do not even target it directly is a term you can take.

Step 3: place keywords by store

The two stores want the same research placed differently, because their indexing differs.

• App Store: highest-value phrases in the title and subtitle; everything else in the 100-character keywords field, comma-separated with no spaces, never repeating title words.
• Google Play: no keywords field, so weave your phrases into the title, the 80-character short description, and naturally into the full description — two or three mentions, not twenty.

A single phrase might live in your App Store subtitle and your Google Play short description — same research, different slot. Keeping that mapping straight across languages is the tedious part; a listing editor with per-store counters helps, but a table works too.

Step 4: measure, then iterate

Research is a hypothesis. Rankings are the result. Record where you rank for your target phrases before a change, ship the change, and check again in a week or two — store indexes take days to update, so do not panic on day one.

Track the top 50 positions for your target set and watch the trend, not the daily noise. Positions bounce; what matters is whether a phrase is drifting up or down over weeks. AppBoard tracks keyword positions in the top 50 for exactly this loop, but the discipline is what counts: change one thing, wait, measure, keep what worked.

Change one variable at a time

The temptation on release day is to rewrite the title, swap the keywords, and redo the screenshots all at once. Then something moves and you have no idea which change did it. If you can, isolate metadata changes from big visual overhauls so each experiment tells you something.

What you can safely ignore

You do not need precise search-volume numbers to start. You do not need to target hundreds of keywords — a focused set of ten to fifteen you can realistically win beats a scattergun of a hundred you cannot. And you do not need to re-research from scratch every month; revisit when you add a major feature, enter a new market, or see a phrase in your reviews you had not considered.

The whole workflow fits on an index card: gather candidates from autocomplete, reviews, and competitors; keep the relevant, winnable ones; place them per store; measure and iterate. Do that consistently and you will out-rank apps spending far more, because most of them never do the boring part.