NeverRanked · The State of AEO · 2026-05-10
From The Citation Tape, NeverRanked's standing AI-citation measurement system.
Generated 2026-05-10. Window: 2026-04-14 to 2026-05-10.
Data integrity notice
This week's data is partial. 3 of 3 tracked clients fell below 80% keyword completion due to a known infrastructure issue (filed at
content/handoff-questions/citation-cron-not-firing.md). Numbers below should be read as a lower bound on what AI engines actually retrieve until the fix lands.Affected this week: - and-scene: 1 of 5 keywords queried (20%) - hawaii-theatre: 2 of 14 keywords queried (14%) - neverranked: 6 of 15 keywords queried (40%)
The full per-client completion table appears in the methodology section at the end of this report.
A standing snapshot of what AI engines actually cite when answering questions about NeverRanked's tracked client universe. Pulled from live 437 citation runs across 5 engines (bing, gemini, google_ai_overview, openai, perplexity), 17 tracked keywords, and 3 clients spanning 3 verticals.
Generated weekly. Same script, same data sources, no manual curation.
The methodology is the script (scripts/state-of-aeo-generate.mjs) plus
the public scoring engine at check.neverranked.com. Reproducible by
anyone with the same query against the same database.
Across 437 captured AI engine responses, the single most-cited third-party source is vertexaisearch.cloud.google.com (Google AI infrastructure (Gemini grounding)), appearing in 132 citation runs across 1 engine and 5 tracked keywords.
Implication: any tracked client whose category overlaps with this source's coverage area (and whose own brand is not yet present on it) is leaving citation share on the table. The path to closing the gap is source-specific. See the per-client baseline section below.
Distribution of citation runs by source type, excluding client-owned URLs (which represent successful citation, not the moat-building question of "where else are they pulling from"):
| Source type | Runs | Share |
|---|---|---|
| aeo-services-agency | 417 | 29.6% |
| seo-publication | 227 | 16.1% |
| aeo-platform | 203 | 14.4% |
| google-ai-infra | 132 | 9.4% |
| google-maps | 112 | 8.0% |
| youtube | 108 | 7.7% |
| news | 108 | 7.7% |
| wikipedia | 41 | 2.9% |
| 18 | 1.3% | |
| directory | 17 | 1.2% |
| tripadvisor | 11 | 0.8% |
| social | 11 | 0.8% |
The domains that appeared most often as cited references across the dataset. Excludes client-owned domains.
| Domain | Source type | Runs | Engines | Keywords | Clients |
|---|---|---|---|---|---|
| vertexaisearch.cloud.google.com | google-ai-infra | 132 | 1 | 5 | 2 |
| google.com | google-maps | 112 | 1 | 4 | 3 |
| youtube.com | youtube | 108 | 3 | 10 | 1 |
| prnewswire.com | news | 102 | 4 | 1 | 1 |
| hawaiitheatre.com | other | 101 | 3 | 2 | 1 |
| geekpoweredstudios.com | aeo-services-agency | 74 | 4 | 1 | 1 |
| eseospace.com | aeo-services-agency | 58 | 3 | 1 | 1 |
| orbitmedia.com | seo-publication | 55 | 1 | 2 | 1 |
| sparktoro.com | seo-publication | 54 | 1 | 1 | 1 |
| evertune.ai | aeo-platform | 54 | 2 | 2 | 1 |
| greenbananaseo.com | aeo-services-agency | 52 | 2 | 5 | 1 |
| arcintermedia.com | aeo-services-agency | 48 | 2 | 4 | 1 |
| animalz.co | seo-publication | 48 | 1 | 2 | 1 |
| generatemore.ai | aeo-platform | 44 | 2 | 2 | 1 |
| boralagency.com | other | 42 | 2 | 2 | 1 |
Top 5 sources per engine. When two engines have very different top lists, that's a signal about how each one's retrieval differs.
Verticals with at least 3 tracked clients enable category-level insights.
Current vertical distribution:
| Vertical | Clients | Runs |
|---|---|---|
| AEO agency | 1 | 403 |
| corporate training | 1 | 3 |
| performing arts venue | 1 | 31 |
As more clients onboard inside any vertical, this section will populate automatically with category-level patterns.
Each client's current citation footprint as of 2026-05-10. These numbers are the baseline against which any future case study is measured. Citation rate climbing from one of these starting points to a higher one over a defined window is the case-study artifact.
Every number in this report comes from citation_runs in the
NeverRanked production database. Each run is one query against one AI
engine for one tracked keyword, with the engine's response_text and
cited_urls captured as raw evidence.
Engines covered: bing, gemini, google_ai_overview, openai, perplexity.
Sample size at this snapshot:
Per-client share of active tracked keywords with at least one citation run in the window. Below 100% means the producer did not complete the full keyword set. A known infrastructure issue causes partial completions on multi-keyword clients (filed in content/handoff-questions/); numbers will rise as the fix lands.
| Client | Active keywords | Keywords with runs | Completion |
|---|---|---|---|
| and-scene | 5 | 1 | 20% |
| hawaii-theatre | 14 | 2 | 14% |
| neverranked | 15 | 6 | 40% |
Honest limits: this is NeverRanked's tracked subset, not a random sample of the AI search universe. Findings are descriptive of what AI engines say to questions in our clients' categories. Generalizing beyond those categories requires more data.
The script that generated this report is at
scripts/state-of-aeo-generate.mjs and reads no private data.
The schema for citation_runs is public in the migrations folder.