Skip to main content
The agent watches keywords from two sources:

​
1. Movers (algorithmic)

Pulled from /v1/keywords/movers per run. Filtered to ±5 rank delta to cut noise — small daily wiggles don’t hit the briefing. What surfaces:
  • Gainers — keywords whose rank improved by 5+ in the last 7 days
  • Losers — keywords whose rank dropped by 5+
  • New entries — keywords that started ranking
  • Dropped out — keywords that fell off the chart
Each entry includes the rank delta, current rank, and search volume.

​
2. Tracked keywords (user-curated)

The keywords you explicitly added to your tracking list in the web app, fetched from user_tracked_keywords (web Supabase). For each one the agent enriches with:
  • Volume score (0-100) from app_store_keyword_stats
  • Difficulty score (0-100)
  • Current rank in the country
  • Last 7 days of rank history (deduped to one rank per day, best position kept)
Surfaced like:
Up arrows mean rank improved over the 7-day window, down arrows mean it worsened.

​
How the LLM uses it

  • Movers are wired into the keywordMovers field of the report
  • Tracked keywords drive prioritizedActions like “Defend rank #19 for ‘screenshot search’” or “Push ‘photo stitching’ from #42 → top 30”
  • Combined with the ASO opportunity list, they feed metadata change suggestions