117 GEO providers: content is common. Crawler governance is not.

Citable screened 150 provider domains and verified 117 active businesses selling GEO, AEO, LLM optimization, or closely equivalent AI-search services. The clearest pattern was not that every agency now sells the same thing. It was that the market has converged on the most visible layer—content—while disclosing much less about crawler governance, precise metrics, and price.

Among sampled providers with known evidence, 99 of 107 (92.5%) sell content or answer-formatting work. Only 27 of 85 (31.8%) explicitly address AI crawler access, crawler governance, or llms.txt. That 60.7-point disclosure gap defines the market’s current maturity problem.

This report explains what providers publicly promise, how they package the work, which AI surfaces they name, and what buyers still cannot verify before a sales call.

GEO is being sold as an SEO expansion, not an SEO replacement

Generative Engine Optimization (GEO) is the practice of improving whether a brand or source is represented, extracted, mentioned, or cited in AI-generated answers. The market could have developed as a separate category with its own agencies, budgets, and operating systems. Public positioning shows a different path.

Among the 109 providers whose relationship to SEO could be classified, 91 (83.5%) either integrated GEO into SEO or offered both integrated and standalone options. Only 18 of 109 (16.5%) positioned the service as separate. Data: Citable GEO Services Market dataset, 13 August 2026.

This is a packaging result, not evidence that GEO and SEO are identical. Many of the underlying capabilities overlap: crawlability, information architecture, structured data, entity clarity, content quality, and authority signals. AI-answer visibility adds a different retrieval surface and a different measurement problem, but providers are usually attaching those additions to an established search foundation.

For buyers, the label matters less than the operating boundary. A provider should be able to show which work improves conventional search, which work targets AI answer systems, and where the same intervention supports both. Without that boundary, “GEO” can become a new name for an unchanged SEO scope.

The dominant GEO product is implementation, not advice alone

The sampled market does not stop at audits and strategy decks. Among providers with known evidence for each engagement type, 99 of 109 (90.8%) offer strategy, 96 of 103 (93.2%) offer implementation, and 88 of 100 (88.0%) offer an audit. Data: Citable GEO Services Market dataset, 13 August 2026.

Retainers appeared in 61 of 85 known records (71.8%). That denominator is smaller because 32 of the 117 included providers did not disclose an engagement structure clearly enough to classify. The common public sequence is therefore: diagnose the baseline, define the plan, implement changes, then monitor the result.

This sequence reflects the mechanics of the work. An AI visibility audit can identify missing entity signals, inaccessible pages, weak answer formats, or absent measurement. None of those findings changes a model’s answer until somebody edits the site, publishes the asset, earns the mention, or changes the technical configuration.

Implementation is also where apparently similar providers diverge. One may edit templates and deploy schema. Another may hand recommendations to the client’s development team. A third may produce content but exclude digital PR, crawler policy, or analytics. The word “implementation” is useful only when the contract names the shipped outputs and the owner of each one.

Content, schema, and measurement form the market’s standard bundle

The most common public workstreams form a recognizable core. Among sampled providers with known evidence for each field:

Disclosed workstream	Providers	Share
Content / answer formatting	99/107	92.5%
Measurement	87/103	84.5%
Schema / structured data	84/104	80.8%
Entity / brand work	83/103	80.6%
Technical SEO	76/98	77.6%
Reporting	70/91	76.9%
Off-site authority / PR	54/92	58.7%
AI crawler access / llms.txt	27/85	31.8%

Data: Citable GEO Services Market dataset, 13 August 2026. Each percentage uses its own known-evidence denominator; unknown records are not silently counted as “no.”

The top of the table shows clear convergence. Providers overwhelmingly talk about creating extractable answers, adding structured data, clarifying entities, and monitoring visibility. Those capabilities are easy for buyers to recognize and easy for agencies to map onto existing content and technical SEO teams.

The lower half exposes two different gaps. Off-site authority or digital PR appears in 54 of 92 known records (58.7%), even though third-party corroboration can affect whether a source is trusted. Crawler access appears in just 27 of 85 (31.8%), despite being a prerequisite for any system that needs to retrieve the page directly.

The finding does not prove that the remaining providers fail to perform crawler work. It proves that a buyer cannot verify that capability from the checked public evidence. This report measures disclosure, not hidden delivery.

The crawler-governance gap is the market’s clearest technical weakness

AI crawler governance is the explicit control of which automated agents may access a site, which paths they may retrieve, and how those rules align with the brand’s distribution and licensing choices. It includes robots directives, CDN or firewall rules, server responses, bot identification, and—in some implementations—llms.txt as a machine-readable discovery aid.

Only 27 of 85 providers with known evidence (31.8%) explicitly mentioned this layer. By comparison, technical SEO appeared in 76 of 98 known records (77.6%), a 45.8-point difference between the two disclosed rates. Data: Citable GEO Services Market dataset, 13 August 2026.

That difference matters because a technically crawlable site is not automatically governed for every AI agent. A provider can repair canonical tags, sitemaps, and JavaScript rendering while never testing whether GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, or Google-Extended receives the intended response at the edge.

llms.txt alone does not close the gap. It is a voluntary convention, not an access-control mechanism. A credible crawler workstream should start with observed server behavior and explicit policy, then use supplementary discovery files only where they support that policy. See our field note on why crawler access is GEO infrastructure and our tested assessment of whether llms.txt matters in 2026.

Measurement is common, but the scoreboard is often undefined

Measurement appeared in 87 of 103 known records (84.5%). Explicit named metrics appeared in only 70 of 114 records (61.4%). The 23.1-point gap indicates that many provider pages promise monitoring without telling a buyer exactly what the dashboard will measure. Data: Citable GEO Services Market dataset, 13 August 2026.

Where providers did name the scoreboard, common metrics included citation share, mention share, prompt visibility, competitive share of voice, AI referral traffic, and conversion or pipeline attribution. These measures answer different questions. A citation metric tracks attributed sourcing. A mention metric tracks brand inclusion whether or not a link appears. Referral traffic captures only the subset of AI interactions that produce a click.

Share of Answer measures how often a brand appears across a defined prompt set, engine set, geography, and observation window. Its value depends on the denominator. A report that says “visibility increased” without preserving the prompts, engines, dates, and scoring rules cannot distinguish a real change from a changed test.

Buyers should therefore request a measurement contract before an execution contract: the prompt universe, engines, locales, observation frequency, citation-versus-mention rules, competitor set, and treatment of unstable outputs. Our Share of Answer measurement guide explains the minimum reproducible structure.

ChatGPT leads provider positioning, while Copilot trails

Provider pages do not distribute attention evenly across AI surfaces. Among sampled providers with known evidence for each engine, ChatGPT was named by 99 of 111 (89.2%), Perplexity by 83 of 101 (82.2%), Gemini by 76 of 103 (73.8%), Google AI Overviews by 70 of 96 (72.9%), and Copilot by 43 of 92 (46.7%). Data: Citable GEO Services Market dataset, 13 August 2026.

These figures measure explicit naming, not actual delivery coverage, query volume, or platform importance. A provider may test an engine without listing it. Another may name five platforms in marketing copy while measuring only two consistently.

The ordering still reveals the category vocabulary buyers are most likely to encounter. ChatGPT and Perplexity anchor the service pitch. Gemini and AI Overviews form a second group. Copilot appears much less often on the evidence pages we could classify.

That unevenness is one reason a single visibility score is insufficient. Different systems retrieve different sources, refresh at different speeds, expose citations differently, and respond differently to the same prompt. The practical unit of analysis is not “AI visibility” in the abstract; it is a versioned result for a defined engine, prompt, market, and date.

Public pricing remains the exception, not the rule

Only 18 of all 117 verified providers (15.4%) displayed a confirmed, buyer-visible numeric price for the relevant service. Forty-six of 117 (39.3%) were confirmed as quote-led or had no visible numeric price on the checked service and discoverable pricing pages. The remaining 53 of 117 (45.3%) stayed unknown because absence could not be proved under the study’s stricter rule. Data: Citable GEO Services Market dataset, 13 August 2026.

The published offers spanned low-cost local packages, fixed audits, implementation sprints, monthly retainers, and five-figure enterprise programs. Those products are not comparable enough to support one meaningful market-average price. Currency, geography, site size, language count, deliverables, and service depth all change the scope.

The useful buyer question is therefore not “What is the average GEO retainer?” It is “What does this exact price include?” A public number without prompts, engines, pages, content volume, implementation ownership, and reporting frequency creates false precision.

Citable publishes its current GEO scopes and prices. That commercial choice is not evidence that one model fits every provider. It is evidence that baseline scope can be made visible before a discovery call.

The GEO Services Disclosure Model gives buyers four verification layers

The GEO Services Disclosure Model evaluates whether a provider publicly explains four buyer-relevant layers: workstreams, implementation, measurement, and commercial terms. It does not score delivery quality. It separates what a buyer can verify before a sales process from what remains an assertion.

1. Workstreams: Does the provider name content, schema, entity work, technical SEO, off-site authority, and crawler governance separately?
2. Implementation: Does the provider ship changes, advise the client’s team, or divide ownership—and is that boundary explicit?
3. Measurement: Are the prompts, engines, locales, metrics, and reporting cadence named?
4. Commercial terms: Is there a numeric price, a scope boundary, a minimum term, or a clear statement that pricing is bespoke?

This model turns a vague category comparison into a reproducible buying process. A provider does not need to offer every workstream or publish one universal package. It does need to make exclusions and ownership legible enough for a buyer to compare like with like.

The model also prevents a common error in market reports: treating the longest service page as the strongest provider. Disclosure quality is useful, but it is not client outcome evidence. Contracts, implementation artifacts, baseline data, and post-engagement deltas remain necessary.

This study maps public claims, not the entire global market

The dataset is a purposive convenience sample of publicly discoverable providers, not a census of the global GEO services market. Researchers screened 150 unique candidate domains on 13 August 2026. 117 met the inclusion criteria, 19 remained uncertain, and 14 were excluded. Every included record had a qualifying public page offering GEO, AEO, LLM optimization, AI search optimization, or a closely equivalent service.

The coding used three states where appropriate: yes, no, and unknown. Unknown was preserved when a page was inaccessible, ambiguous, or insufficient to prove absence. That is why most percentages use a known-evidence denominator instead of all 117 providers.

The source material was primarily provider-owned service and pricing pages. That makes it suitable for analyzing public positioning and buyer-visible disclosure. It does not independently verify service quality, client outcomes, team capacity, revenue, or actual platform coverage.

The study therefore does not support claims about global market size, category growth, provider rankings, average retainers, or the share of agencies that deliver a capability in private. It supports a narrower and more useful conclusion: among the sampled providers with readable evidence, content, schema, and measurement are established parts of the pitch; crawler governance, explicit scoreboards, and numeric pricing are much less consistently disclosed.

The next market advantage is operational clarity

The GEO services category no longer lacks vocabulary. Most sampled providers can describe content optimization, schema, entities, audits, and monitoring. The next differentiator is the operating system behind those words.

Buyers should be able to see what gets changed, who changes it, which engines are observed, how a result is scored, what is excluded, and what the work costs. Providers that make those boundaries explicit reduce sales friction and make later performance claims auditable.

The market’s 60.7-point gap between disclosed content work and disclosed crawler access is not a verdict on agencies. It is a map of what the category has explained well—and what it still needs to make legible.

If you want to benchmark a provider or your internal program against those four disclosure layers, start with the CITE Framework or establish a documented baseline with an AI Visibility Audit.