We analysed over 11 million AI citations.

Here’s what determines whether AI recommends your brand over your competitor’s.

Datamapping_Graphic
14 research cosine similarity page title per citation

2,000 hours. 11 million citations.
Here's what we found.

Our data science team cracked it.

In partnership with DataForSEO, we ran the world’s largest study of its kind across ChatGPT and Gemini.

Not what AI says it prefers. What it actually does.

Invisible is the best-case scenario.

The worst case is someone else shaping what AI says about you.

12% overlap between ChatGPT results and the Google SERP. 

Ranking #1 on Google does not mean you exist in AI.

Most brands are only engineering one of two retrieval systems.

chatgpt-google-overlap-12-percent

AI doesn't cite the loudest brand. It cites the most connected one.

AI doesn't cite the loudest brand. It cites the most connected one.

20-research-co-citation-cluster-semantic-similarity

Topic similarity — not keyword matching.

Pages semantically aligned to a query get cited.

Pages optimised for an exact phrase don’t.

We measured cosine similarity across keyword-to-title, keyword-to-content, and keyword-to-topic. The signal is clear.

Co-citation clusters — the authority bubble.

When authoritative sources cite your brand alongside each other, AI interprets you as part of a trusted cluster.

Getting inside one is more valuable than any single backlink.

This is how we engineer

AI citation probability.
01

Audit

Common Crawl audit

We audit your Common Crawl indexing status and map exactly what AI currently knows about your brand.

02

Map

Co-citation network map

We map the co-citation networks your competitors are already inside. Not keywords. Clusters of semantic authority and the harmonic centrality of every node in them.

03

Score

Cosine, not DR

Every content and backlink opportunity is scored by cosine similarity to your target clusters. Not domain rating.

04

Engineer

Build citation presence

We engineer your brand into the citation networks AI uses to recommend sources. Then we track, measure, and compound it.

The research produced the tools.

No other agency ran a study at this scale.
01

Cosine similarity scoring

Opportunity value calculated against your target cluster vectors. Not DR.

02

Harmonic centrality mapping

Clusters mapped by AI retrieval weight. We find the nodes that matter to the model.

03

Hallucination correction

We track what AI says about you against ground truth. Engineer corrections into the corpus.

Your brand is either inside the citation clusters.
Or it isn't.

Your brand is either inside the citation clusters. Or it isn't.

We'll show you which.
  • Your citation presence across your target queries
  • Which authority bubbles you’re absent from, and who’s inside them
  • Your content’s semantic alignment score vs. the sites AI is currently citing
  • Your backlink profile scored by topical similarity, not domain rating
  • A roadmap to begin engineering citation probability from day one
Book a strategy session If we can't show you something your current agency missed, we'll say so.
AI citation probability
dots background