Keyword research

Keyword Research

Produce a keyword plan a developer can act on this week, not a raw dump of scores.

Requirements

GetAppNiche MCP connected: get_keyword_difficulty, batch_keyword_difficulty, get_supported_countries (free). Keyword scoring is the expensive call — 10 credits per keyword — so generate first, score second, and score only the shortlist.

Workflow

1. Collect context. App's core job, target market (default US / en unless told otherwise — verify uncommon markets against get_supported_countries), and any keywords already targeted.
2. Generate candidates by intent bucket, before any scoring:
   • Core: what the app literally is ("habit tracker")
   • Problem: what the user is trying to fix ("stop procrastinating")
   • Feature: specific capabilities ("streak counter", "widget")
   • Adjacent: neighboring queries the audience also types Aim for 15–25 candidates across buckets.
3. Shortlist ~10 by judgment (relevance × plausible traffic) and score them in one batch_keyword_difficulty call. Batch is capped at 10 keywords — prioritize before scoring rather than paging through everything.
4. Read the scores like a strategist. The sweet spot is meaningful popularity with difficulty your app can survive: a new app should mostly avoid head terms and win mid-tail; an established app can contest harder terms. Note which top-ranking apps hold each keyword — outranking a giant is a different project than outranking abandonware.
5. Recommend. Split the final list into: target now, build toward, skip (and why).

Output

A ranked table — Keyword · Popularity · Difficulty · Verdict — then the three-tier recommendation with one sentence of reasoning per tier, and suggested placement (title vs. subtitle vs. keyword field) for the "target now" set.

Guardrails

• Scores are point-in-time snapshots; treat small differences (±5) as noise.
• Never present a keyword list without a recommendation — prioritization is the value.
• State the assumed country/language; rankings differ per storefront.