Methodology
How signals are collected, how keywords are sized, how decisions are made — and every known limit.
How opportunity scores work
A rule-based engine across five dimensions: demand, competition, intent, timing, execution cost. Rules are versioned; scores are priorities, not probabilities.
Signal sources
X signal
Names used for the first time in 60 days by tracked accounts. Only as good as the list.
GitHub breakout
Repos under 14 days old averaging 200+ stars a day. Public and free.
Sitemap surge
Words shared by several new URLs on watched sites, against each site’s own baseline.
Search Console
Queries your own site gets first impressions for. Yours only.
New sites
Domains registered in the last 90 days already ranking for 20+ keywords. Weekly.
The 14-day review
Every lead re-checked in Google Trends 14 days later: hit rate and days ahead per signal.
The 14-day review
Each lead is re-checked against its own baseline: a hit when its post-report peak clears 2x and +10 points. Hit rates shown on this site accumulate from real runs only.
→ Strata: hit, +3 days (documented sample)Known limitations
Published on purpose. Every conclusion on this site can be traced to a source, a capture date and a rule version.
- Name-judge accuracy measured at 149/170 on a real sample; ~3 true names were missed (synthid detector, veo, v4.1 flash).
- Within five days, most generic terms behind a new name do NOT rise (1 of 8 in the documented sample moved).
- Worldwide Trends series can mask US-only moves; country series requires the optional data source.
- Google Ads volumes run ~2 months behind; Trends is unreliable for terms under a few hundred searches a month.
- Trends’ last two days are unsettled and excluded; a rise must hold two days to count.
- Hit rates per signal are still accumulating — the Strata check (hit, +3 days) is the only documented one so far.