GEO Authority Score (GAS): domain authority for generative search
GAS is an open-algorithm, domain-level score measuring how ready a site is to be cited by generative engines, how far it has actually been adopted, and how its authority has grown over time.
GEO Authority Score = readiness × 0.45 + adoption × 0.40 + trajectory × 0.15, capped by adoption tier and domain tenure.
Core capabilities
- GEO Authority Score (GAS): an open-algorithm domain readiness score with a public reference script
- Multi-engine citation monitoring across ChatGPT, Perplexity, Doubao, Kimi and DeepSeek
- Keyword matrices and prompt libraries across brand, category and problem layers
- Long-form pipeline with source provenance, human top-up markers and a pre-delivery review gate
- Index push, index detection and 72-hour / 4-week / 12-week post-publish tracking
- Enterprise API: GEO reports, competitor analysis, white-label client reports and webhooks
Six readiness dimensions
Machine access 18 (explicit robots.txt rules for GPTBot, PerplexityBot, ClaudeBot, Google-Extended, plus llms.txt), structured data 20, entity consistency 17, answer extractability 18, freshness and depth 15, citation signals 12. Site-level checks run once; page-level checks average the homepage plus up to nine pages sampled evenly from the sitemap.
Adoption is measured, never assumed
Adoption combines a deterministic model knowledge probe (temperature 0, raw answers archived), presence in model source recommendations, and real citations observed in the last 90 days. When a probe is unavailable it scores zero — no speculative credit.
Trajectory: authority accrues over time
Tiered by the first archive.org snapshot age, combined with archive continuity, in-platform score growth and citation momentum. A brand-new domain cannot score high in one pass.
What to fix first when the score is low
1) Allow major AI crawlers explicitly in robots.txt and publish llms.txt. 2) Open every page with an extractable answer capsule and a clear H2 structure. 3) Add Organization, BreadcrumbList, FAQPage and Article structured data. 4) Attribute figures to named sources with an as-of date. 5) Keep publishing and earn external citations — this part cannot be bought with markup.
Classic SEO vs Generative Engine Optimization (GEO)
| Dimension | Classic SEO | Generative Engine Optimization (GEO) |
|---|---|---|
| Goal | Rank positions and clicks | Being cited and adopted inside AI answers |
| Retrieval | Keyword matching and link authority | Semantic recall, entities and source trust |
| Content bar | Cover search intent | Extractable answers with verifiable data |
| Technical base | Crawlability, sitemap, structured data | Same base plus llms.txt and AI-crawler access |
| Measurement | Rankings, CTR, organic traffic | Citation rate, share of voice, answer adoption |
Frequently asked questions
Does a high GAS guarantee AI citations?
No. GAS quantifies readiness and adoption; it is not a citation-rate promise. We report descriptive statistics only, with no causal claim.
Can the algorithm change silently?
No. Any weight change bumps the version, is recorded in a public changelog, and is mirrored in the public reference script so third parties can reproduce results.
References
By the 23SEOGEO team · Last updated: