Methodology & independence
This page is the contract between the lab and its readers. It publishes before any winner is named, and it cannot change after results are known.
The problem, measured
We documented 58 GEO/AEO agencies from primary sources. 35 publish their own “best GEO agencies” listicle; 14 rank themselves #1. Roughly 90% of page-one results for buyer queries in our categories are written by a company that sells the thing being ranked. No review platform maintains a GEO/AEO agency category. That is the vacuum this site fills — with evidence, not another opinion.
Evidence hierarchy
Every fact carries one of six states. Stronger evidence always outranks weaker; a claim without a source is treated as a claim, not a fact.
Lab verified
First-hand test, recorded output, dated capture, or direct observation by this site
Primary documented
The provider’s own pricing page, service page, legal terms, or public directory listing
Independent documented
Regulatory records, reputable publications, customer evidence, research organizations
Provider reported
Submitted by the company, not independently verified — always labeled
Community signal
Attributed discussion and third-party aggregate ratings; informs questions, never establishes facts, never enters scores
Unknown
No current public evidence found — stated plainly, never silently penalized
Agency scoring rubric — 100 points
Published before any category is scored. Weights cannot change after results are known. Rankings go live only after every listed provider receives a factual pre-publication check — a fact check, not an approval request.
| Dimension | Weight | What it measures |
|---|---|---|
| Service capability | 20 | Depth and completeness of relevant delivery |
| Demonstrated outcomes | 20 | Quality, specificity, and relevance of case evidence |
| Method and measurement | 15 | Repeatable process, metrics, reporting, accountability |
| Practitioner credibility | 10 | Named people, relevant experience, visible expertise |
| Commercial transparency | 10 | Scope, minimums, timelines, contract clarity |
| Technical and integration depth | 10 | Ability to work across systems, data, implementation |
| Independent corroboration | 10 | Credible third-party references and customer evidence |
| Buyer-fit clarity | 5 | Clear ideal customer, exclusions, trade-offs |
Rules that never bend
- Inclusion is free and payment never touches rank. Future sponsor placements are labeled, visually identical in density, and structurally separated from scoring. The same eligibility criteria apply to every provider.
- No user reviews, no imported reviews. Third-party ratings may appear only as dated, attributed context facts (community-signal tier), never in scores, and never as structured data.
- AI benchmarks run in browsers, not APIs. Published research shows ~24% brand overlap between API and browser answers; we measure the surface buyers actually use, with prompts, dates, and account states documented.
- Corrections are public. Errors get fixed, dated, and logged — and a provider never needs to pay to correct a fact.
- Unknowns stay unknowns. “Pricing was not publicly available when we checked” is a finding, not a blank — and it never silently lowers a score outside the published transparency dimension.
methodology v1 · published 2026-08-31 · material changes will be logged on this page