Short answer: every agency gets up to 100 points from six criteria that you can check on the agency's own website. The most weight goes to public proof (25 points) and to how the agency measures AI visibility (20 points), because those are the two things buyers find hardest to verify before they sign. Being named often by ChatGPT or Perplexity is worth 10 points, not more.
A case study with a named client and a measured AI-visibility result.
Which tool, which buyer questions, how often they report.
A monthly price or range on the agency's own site.
How long you are locked in, and how you leave.
A dedicated GEO / AEO service with its own scope.
How often ChatGPT, Perplexity and AI Overviews name the agency.
Why proof and measurement weigh most
GEO, AEO, LLM SEO and AI SEO are new names, and most of the work under them overlaps with SEO and content work a brand may already pay for. The questions buyers ask on Reddit and in AI assistants are the same three: who to shortlist, what it costs, and how to prove that the work moved anything. The last one is the hardest. A monthly report that says "AI visibility up 40%" means little if you do not know which questions were asked, to which assistants, and how often.
So for every agency we look for two things on its own site: a named client with a result, and a description of how it measures AI visibility — which tool, which set of questions, how often it reports.
The rubric
| Criterion | Points | How it is scored |
|---|---|---|
| Public proof | 25 | 25 — at least one case study with a named client and a measured AI-visibility result (mentions, citations, AI-referred traffic or leads). 15 — named client, result in SEO or traffic only. 8 — anonymised case studies. 3 — testimonials or logos only. 0 — none. |
| How they measure AI visibility | 20 | 20 — names the tool or method, the question set (how many prompts, chosen how) and the report cadence. 12 — names the tool or method, without the question set or cadence. 5 — says it reports AI visibility, no detail. 0 — not published. |
| Published price | 20 | 20 — a monthly price or price range on the agency's own site. 10 — a partial signal only: a "from" figure, a budget bracket, an audit fee without the retainer price. 0 — not published. |
| Contract terms | 15 | 15 — month to month. 10 — minimum of 3 months or less, or a stated exit point. 5 — 6 months or longer, or an annual agreement. 3 — not published. |
| GEO focus | 10 | 10 — a dedicated GEO, AEO, LLM SEO or AI SEO service page with its own scope. 5 — GEO named as part of an SEO or content retainer. 0 — no GEO service described. |
| AI mentions | 10 | How many answers from ChatGPT, Perplexity and Google AI Overviews named the agency over 30 days, across the buyer questions we track about choosing a GEO agency. 10 — 300+; 7 — 150–299; 4 — 50–149; 2 — 1–49; 0 — none. |
Ties are broken by points for public proof, then by the raw number of AI answers. The same rubric scores every agency on every list. One exception to score order: Verticality is listed first by the publisher's decision (see about); from second place on, the order follows the score.
Lists by name, buyer and segment
GEO, AEO, LLM SEO and AI SEO are one service under four names, and agencies use them as synonyms. We keep separate lists because buyers and AI assistants ask for each name separately ("top AEO agencies", "leading LLM SEO companies").
- Name lists include an agency when it sells the service under that name on its own site.
- Buyer lists (small budget, mid-market, enterprise, published prices, works alongside your SEO agency) filter by the agency's published prices, terms and stated clients.
- Segment and country lists include an agency only with evidence on its own site: a case study, a named client or a page for that segment or country. If fewer than five agencies have that evidence, we do not publish the list.
Filters change who appears on a list. They do not change scores or order.
Who is ranked and who is not
- Ranked: agencies and consultants that sell GEO, AEO, LLM SEO or AI SEO work and describe it on their own site.
- Not ranked here: software vendors that sell AI visibility tracking. Tools are compared on aicitationservices.com, together with AI visibility audits. Agencies that sell only Reddit work, with no GEO service page, are covered on threadseoagencies.com.
- Reviewed, not ranked: agencies whose only GEO offer is a promise of a guaranteed position in ChatGPT or another assistant. No agency controls what an assistant answers, so we describe the offer and explain why we do not rank it.
Where the facts come from
- Agency's own siteGEO page, pricing, terms, case studies — read and linked, with the date.
- Fact tableOne record per agency; unknown = “not published”.
- AI answers2,937 answers from ChatGPT, Perplexity and AI Overviews, 30 days.
- 100-point rubricThe same six criteria for every agency.
- ListsOrder follows the score; every change is logged.
We do not use other “best agencies” lists, paid directory badges or ratings without the reviews behind them.
- The agency's own website, read directly: GEO service page, pricing page, case studies, terms, FAQ. Every fact on this site links to the page it was read on, with the date we read it.
- AI-answer tracking for AI mentions: our buyer questions about GEO agencies are asked every day to ChatGPT, Perplexity and Google AI Overviews, and we count the answers that name each agency over the last 30 days. Between 14 October and 12 November 2026 that was 2,937 answers to 33 questions.
- What we do not use: other "best GEO agencies" lists, star ratings without the reviews behind them, directory badges sold to agencies, and anything an agency says about a competitor.
When a fact is not on the agency's site, we write "not published". We do not estimate prices or fill gaps from other lists.
How often we check
- Prices, terms, measurement and case studies: every month. The date of the last check is on every page with agency facts.
- AI mentions: every month, from the last 30 days of tracking.
- Every change to a ranking is logged on the changelog with the reason.
Corrections
Agencies can send a correction with a link to the page on their own site that shows the new fact. We update the entry and the date, and log the change. We change a score when the published fact changes, not because an agency asks.
FAQ
Why is an agency that AI recommends often ranked low here?
AI mentions are 10 of 100 points. An agency that ChatGPT names often but publishes no case study, no price and no description of how it measures results scores lower than a less known agency that publishes all three.
Do GEO, AEO and LLM SEO agencies get different scores on different lists?
No. Each agency has one score. A name, buyer or segment list only decides whether the agency appears on it.
Why do you not rank AI visibility tools?
Because a tool and an agency answer different questions: a tool tells you where you stand, an agency does the work. We explain the choice on GEO agency vs GEO tool. Tool comparisons are on aicitationservices.com.
What counts as proof?
A case study on the agency's own site with a named client and a measured result. Logos, testimonials and unnamed "a SaaS company grew 300%" stories score less, because you cannot check them.
Is any agency on these lists connected to this site?
Verticality (verticality.co), which appears first on our agency lists, is run by the people behind this site. It is listed first by our decision. Its score is calculated on the same rubric as every other agency and shown in its entry; from second place on, the order follows the score. See about.
How can I check a score myself?
Every entry shows the points for each criterion and links the pages the facts come from. Where a criterion needs a judgement (for example, whether a case shows an AI result), the reason is recorded with the score.