Methodology · FAQ

Every number traces back.
Every gap stays visible.

Gulliver uses free public signals to surface AI-site opportunities. Unavailable data stays `missing`; proxy signals never masquerade as traffic, revenue, or causality.

01

Discovery: read real charts in a browser

The discovery layer reads the Toolify monthly trending chart and There’s An AI For That Trending. Every seed keeps its source URL, chart position, and read time. Chart fields find candidates; they do not prove actual traffic.

02

Proxy score: worth checking, not confidence

Six inputs contribute to the score: discovery chart up to 18 points, domain recency up to 35, Tranco band up to 14, sitemap scale up to 20, pricing entry 8, and reachable homepage 5. A missing input scores zero; each detail page also shows coverage across all six inputs.

03

Deep review: return to the source page

Coarse screening reads robots, homepage, RDAP, Tranco, sitemap, and pricing. Deep review adds Wayback CDX, OpenPageRank, and page structure. Available sources remain clickable on site and report pages.

04

Monetization: pricing is not revenue

A pricing page proves only that a monetization entry exists. Without a verifiable public disclosure, revenue stays `missing`. Any future estimate must show a range, assumptions, and confidence.

05

Reports: specific URLs and the next check

Reports combine real archive dates, sitemap clusters, sample URLs, pricing evidence, and five concrete actions. If Wayback, Similarweb, or OpenPageRank is unavailable, no stale value or guess is substituted.

Frequently asked questions

Is the data accurate?

Toolify growth, Tranco, and OpenPageRank are different proxy signals, not visits. RDAP records registry events; stack and page structure are heuristic observations. Use them to find directions and frame verification questions, not as financial facts.

How can a high proxy score coexist with missing Tranco?

The score orders verification work across multiple signals. A new domain with a large sitemap, pricing entry, and reachable homepage can score well without appearing in the current Tranco top million. Always read input coverage beside the total.

Why are so many fields missing?

A source may require a key, be blocked by robots or Cloudflare, fail on the network, or simply not expose that attribute. Missing means we did not obtain it in this run; it does not mean the value is zero or absent.

When do rising and dropped stories appear?

They require at least two snapshots captured under the same method. T0 is a baseline, not a period-over-period story.