evaluating and trusting AEO and GEO platforms, agencies, and citation-tracking tools — dataset

About this freeze

Frozen measurement export for evaluating and trusting AEO and GEO platforms, agencies, and citation-tracking tools (measured 27 July 2026). Primary evidence — re-derivable from the published methodology. Negatives (cited: false) stay in the freeze and count in rates. The HTML table is a short preview; the full freeze is in the CSV/JSON downloads.

Each row is one direct answer-engine check. We asked the category question verbatim through the engine's live API and recorded whether the answer cited any sources — cited: true means at least one source URL; cited: false is a measured negative, not missing data. Measurement date: 27 July 2026.

Engines and sampling rules (including any variance / test-retest panel) are disclosed on the linked methodology page for “evaluating and trusting AEO and GEO platforms, agencies, and citation-tracking tools.” This dataset page freezes the rows; it does not recompute the study narrative.

Full methodology

Findings from this freeze

Declarative facts computed from this freeze only — not a substitute for the study article.

Overall freeze

• In this freeze, 77 of 92 checks (83.7%) recorded at least one cited source.
• 15 of 92 frozen checks recorded no citations (negatives kept).
• This freeze covers 24 distinct queries.

Citation rates by engine

• chatgpt: 9 of 24 runs (37.5%) recorded ≥1 cited source.
• gemini: 24 of 24 runs (100%) recorded ≥1 cited source.
• perplexity: 44 of 44 runs (100%) recorded ≥1 cited source.

Summary statistics

Computed from the frozen rows and rollup sample counts. No figures are invented at render time.

• Rows92
• Distinct queries24
• Enginesperplexity, chatgpt, gemini
• Checks with ≥1 citation77 of 92 (83.7%)
• Checks with no citations15
• Observation dates2026-07-27
• Queries in design24
• Checks recorded92
• Checks planned92
• Variance panel groups10

Query coverage

Distinct category questions present in this freeze.

Questions in this freeze

• best tools to track ChatGPT citations
• best tools to track Perplexity citations
• best GEO tools for marketers
• AEO platforms for marketing agencies
• what to look for in an AEO platform
• is AEO a scam
• primary source AEO content
• how to vet an AEO or GEO vendor
• questions to ask before buying an AI visibility tool
• red flags when evaluating an AEO agency
• what does an AEO platform actually measure
• how agencies price AEO retainers
• difference between AEO tools and traditional SEO tools
• how to tell if an AEO tool's citation data is real
• case studies proving AEO actually works
• how white label AEO reporting works for agencies
• how to choose between AEO platform vendors
• what makes AEO research trustworthy
• how to verify AI citation claims from a vendor
• common AEO sales pitches marketers should question
• what credentials matter for an AEO consultant
• how to build a business case for an AEO platform
• how enterprise teams evaluate AEO tooling procurement
• what a good AEO onboarding process looks like

Column dictionary

Field-by-field map of the freeze. The cited column is the direct citation detector: true means at least one source URL; false means none were cited.

Core detectors

cited: true / false

Whether this check recorded at least one cited source URL (true) or none (false). False rows are negatives and stay in the freeze.

variance panel

A fixed subset of questions re-asked the same day (k repeats) so citation flip rates are measured, not guessed.

check_group_id

Groups k-repeat draws of the same query on the same engine for test-retest comparison.

run_seq

Repeat-draw index within a same-day variance panel (1 = first draw).

All columns

cited

Whether this check recorded at least one cited source URL (true) or none (false) — negatives are data.

query

The category question put to the answer engine.

engine

Which answer engine produced this observation (e.g. perplexity, chatgpt, gemini).

run_at

ISO timestamp when this engine check was recorded.

run_seq

Repeat-draw index within a same-day variance panel (1 = first draw).

source_urls

Pipe-separated full source URLs cited in the answer, when any.

check_group_id

Groups k-repeat draws of the same query on the same engine for test-retest.

source_domains

Pipe-separated domains cited in the answer, when any.

How to use this data

1. Download the CSV or JSON for independent re-analysis — every row is frozen at measurement time.
2. Read the methodology page for sampling rules, engines, variance panels, and limitations.
3. Cite the study article for the narrative findings; cite this dataset when quoting row-level figures.
4. Treat cited: false (or empty source columns) as measured negatives, not missing data.

Measurement preview

Showing 8 of 92 rows with the scannable columns only (source URL lists stay in the download). Download CSV or JSON for the complete freeze.

Full freeze: CSV · JSON