Your buyers stopped clicking. When someone evaluating software asks ChatGPT, Perplexity, Claude, or Google's AI Overview "what's the best [category] tool," they get an answer - a short list of named sources - and most of them never see a results page at all. The question is no longer whether you rank. It's whether you get cited.
Citation is not ranking. AI engines don't sort ten blue links; they decide whether your site is a source worth quoting, and that decision turns on signals most SEO has never measured. A company can rank #1 on Google for a query and be completely invisible in the AI answer to the same question.
So we measure it. We are scoring 49 recognizable B2B SaaS companies on how often AI engines actually cite them, using the CITE framework - Technical, Content, Entity, and Trust - and a standardized set of the queries their buyers really ask.
How we scored this. Every company is tested against the same standardized query set for its category, across ChatGPT, Perplexity, Google AI Overview, and Claude. We score four pillars - Technical, Content, Entity, Trust - into one GEO score (0–100), and we measure citation rate: the share of buyer queries where AI engines actually cite the company's own domain, not just mention its name. No company paid for placement; placement cannot be bought. The Index is refreshed quarterly. Read the full methodology →
The Index
First edition coming soon
The first quarterly edition is in measurement right now: 49 recognizable B2B SaaS companies - CRMs, project management, analytics, payments, dev tools - each scored on the four CITE pillars and tested against the same standardized buyer queries across ChatGPT, Perplexity, Google AI Overview, and Claude. The full ranked leaderboard publishes here, on this URL, and refreshes quarterly.
You don't have to wait for the Index to know your own number - AI engines are already deciding whether to cite you.
Methodology
How the Index is measured
AI engines don't rank pages; they decide whether to cite a source. The CITE framework - Technical, Content, Entity, Trust - measures the four things that decision turns on.
What we measure: the CITE framework
Every company is scored on four pillars, each 0–100, combined into one overall GEO score (0–100) with the four pillars weighted equally - the same framework and weighting behind every Osoro scan, so an Index score and a scan score are directly comparable.
Technical
Whether AI crawlers can access, parse, and render the content cleanly
Signals: Crawlability, render-without-JS, clean HTML semantics, llms.txt posture, speed
Content
Whether the content is shaped like an answer to the questions buyers actually ask AI engines
Signals: Question-led headings, extractable answer blocks, depth, structured comparisons
Entity
Whether AI engines can verify who the company is and what it's authoritative on
Signals: Organization schema, author bios, consistent entity references, knowledge bases
Trust
Whether the signals that make an engine willing to cite a source are present
Signals: Citations out, third-party mentions, freshness, factual consistency, reputation
The AI engines we query
Each company is tested against the AI engines B2B SaaS buyers actually use to research software: ChatGPT, Perplexity, Google AI Overview, and Claude. Where an engine returns no answer for a query, that combination is recorded as untested - it never counts against a company's citation rate.
The standardized query set
Every company is tested against the same 10-query structure, with the brackets filled by the company's own category, market leader, and primary competitor:
- best [category]
- best [category] for startups
- [category] alternatives
- top [category] tools 2026
- which [category] tool should I use
- alternatives to [market leader]
- [Company] vs [primary competitor]
- is [Company] worth it
- [Company] reviews
- [Company] pricing
The fairness rule: every company is scored against the same query structure for its category. We do not cherry-pick queries that flatter or punish any company. A company's category is determined by its own primary positioning.
How citation rate is computed
Citation rate = (query × engine combinations where the company's own domain is cited) ÷ (total combinations tested). "Cited" means the domain appears as a named source in the answer - a link, a footnote, or an explicit "according to" reference - not merely that the company name is mentioned in prose. We measure citation, not mention.
Refresh cadence
The Index is refreshed quarterly. Each refresh re-runs the identical query set and engine set against the same company set, so movement is real movement, not a methodology change.
Limitations
- AI engines are non-deterministic: the same query can yield different sources across runs. A quarterly snapshot is a point-in-time measurement, and we report this as a known source of noise.
- Engines change continuously - models and retrieval update between quarters, which is exactly why the Index is refreshed rather than published once.
- Category assignment is a judgment call. We assign each company one primary category based on its own positioning, and disclose it in the table.
This Index is not affiliated with any ranked company. No company paid for placement, and placement cannot be bought.
Frequently asked questions
How do AI engines decide which companies to cite?
AI engines don't rank pages; they decide whether to cite a source. That decision turns on four signal groups - Technical (can crawlers access and parse the site), Content (is it shaped like an answer to buyer questions), Entity (can the engine verify who the company is), and Trust (is the source worth quoting). The CITE framework measures all four.
What is a GEO score?
A GEO score is a 0-100 measure of how likely AI engines are to cite a site, composed of four pillar scores - Technical, Content, Entity, and Trust. It is the same score every Osoro scan produces, so an Index score and a free-scan score are directly comparable.
What is citation rate?
Citation rate is the share of tested buyer queries where AI engines cite a company's own domain as a source - a link, a footnote, or an explicit reference - not merely mention its name. We measure citation, not mention.
How is the AI Citation Index measured?
Every company is tested against the same standardized set of 10 buyer-intent queries for its category, across ChatGPT, Perplexity, Google AI Overview, and Claude, and scored on the four CITE pillars. No company paid for placement, and placement cannot be bought.
How often is the AI Citation Index updated?
Quarterly. Each refresh re-runs the identical query set and engine set against the same company set, so movement between quarters is real movement, not a methodology change.
Your company has a GEO score right now, and AI engines are already deciding whether to cite you or a competitor for the queries your buyers ask.
In 30 seconds you'll see your overall GEO score, your four CITE pillar scores, the competitor AI cites instead of you, and the top fixes that move your citation rate most. No account required.