Reading time: 11 minutes
To sell GEO services, the agency starts from the question the client already asks, “What about AI?”, and answers with the client’s own data. The GEO pitch has three parts: what changed, what the agency measured and a closed-scope pilot. That way, you can sell GEO to clients as an extension of the SEO contract.
In short
- Clients already ask “What about AI?”. In a 2026 AgencyAnalytics survey, 66% of agency professionals reported increased client demand for AI search services.
- To sell GEO services, start with the client’s own data. The GEO pitch persuades more with the client’s brand than with the concept.
- Google treats optimization for AI as part of SEO. So you can sell SEO and GEO together, which may reduce resistance.
- Offer a closed-scope pilot and don’t promise citations, because even OpenAI doesn’t guarantee placement in ChatGPT.
Why will SEO clients ask for GEO services?
Because buyers already use AI to research, and agency clients are starting to ask for it. AgencyAnalytics surveyed 494 marketing agency professionals between February and April 2026. Of them, 66% reported increased client demand for AI search services, and the release does not state the base. The company sells reporting software for agencies, and 64% of respondents are in the United States.
In B2B, G2 surveyed 1,076 B2B decision makers in March 2026, across North America, EMEA and APAC. Among B2B software buyers, 51% say they start research with an AI chatbot more often than with Google. Also, 54% name chatbots among the sources that influenced the vendor shortlist, the top-ranked one. G2 sells software review products, and the page does not state the base for these two figures.
Even on Google, the AI summary changes behavior. The Pew Research Center tracked 900 US adults in March 2025. Users who saw an AI summary clicked a traditional result in 8% of visits, against 15% for those who didn’t. But this is a comparison, not proof of cause.
Therefore, an agency that decides to sell GEO services to clients answers a question that already exists. To understand the concept behind these numbers, see the guide on what is GEO.

How do you explain GEO services to clients without jargon?
To sell GEO services without jargon, explain AI search optimization (GEO) as the work of managing what AI answers say about the client’s category. So leave acronyms like AEO, LLM and fan-out out of the meeting. In client language, the goal is for the brand to be remembered and cited in the answers, building on the SEO the client already pays for.
Google helps remove the fear of the new term, so anyone who wants to sell SEO and GEO to clients has an official argument. Its documentation says that “optimizing for generative AI search is optimizing for the search experience, and thus still SEO.”
| Instead of saying | Say |
|---|---|
| “We will optimize for LLMs” | “We will measure what ChatGPT and Gemini answer about your category” |
| “You need GEO and AEO” | “We will look after your brand in AI answers, besides the Google list” |
| “AI share of voice” | “In how many answers your brand appears, compared with your competitors” |
A 30-second GEO pitch fits in three sentences:
- What changed: “Your customers ask AI assistants who to hire, and the AI answers with a short list of brands.”
- What we saw: “We ran 10 questions from your market, and your brand appeared in this many answers, against this many for your competitors.”
- What we propose: “A two-month pilot, with closed scope, to measure and fix what AI says about you.”
This GEO pitch works because the second sentence uses the number from the test in the next section.
What data should you show in the first meeting to sell GEO?
Show three data points from the client itself: AI impressions in Search Console, visits from the AI Assistant channel in GA4 and a short question test. Market data, however, enters the GEO pitch only as context. After all, anyone who wants to sell GEO to clients must show their brand, not the industry average.
- AI impressions: the generative AI performance report in Search Console shows impressions in AI Overviews and AI Mode. Google rolled the report out to all sites on August 31, 2026, and it groups data by page, country, date and device.
- Assistant visits: GA4 has the default AI Assistant channel, for visits from ChatGPT, Gemini, Copilot and others. However, the GA4 documentation says the channel excludes AI Overviews and AI Mode. The steps are in the guide to AI traffic in GA4.
- Question test: it is the data closest to the client, because it shows the client’s brand in the answers.
For the test, choose 10 questions that a buyer would ask. Then run each one 3 times in the same engine, such as ChatGPT, and note whether the brand and two competitors appear. With 30 answers, the margin of error runs from ± 18 to ± 31 points. The table in the guide on how to measure brand visibility shows this. So present the result as an indication, not as a measurement.
How do you run the first meeting?
Run the meeting in five short steps and leave the proposal for the end. The suggested order puts the client’s data before the price, so the conversation starts from what the client already sees. Also, a lean agenda shows that the GEO pitch doesn’t depend on promises.
- Opening: ask what the client has heard about AI and what they fear losing.
- Data: show Search Console, GA4 and the question test, in that order.
- Explanation: use the “Say” table and the three-sentence pitch.
- Pilot: present scope, price and exit criteria.
- Next step: agree, for example, on who grants access and the date of the follow-up.
What objections come up when you sell GEO, and how do you answer them?
Five objections are predictable: “GEO is a fad”, “we already pay for SEO”, “AI sends little traffic”, “you can’t measure it” and “let’s wait”. When you sell GEO services, agree first with the true part of the objection and only then answer. That way, each answer gets a source or an admitted limit.
Objections about the value of GEO
- “GEO is a fad.” The name is new, but the buyer’s behavior is not. Cite G2’s 51%, noting that they are B2B software buyers, and say that, for Google, the work is still SEO. That way, the client uses what they already pay for.
- “We already pay for SEO.” Ranking on Google doesn’t guarantee showing up in the AI answer. So the GEO pitch should sell SEO and GEO as work that complements each other. The guide to GEO for agencies shows the low overlap between the links that assistants cite and Google’s top 10.
- “AI sends little traffic.” On average, yes. SE Ranking measured AI platforms at 0.32% of the traffic of 101,574 sites worldwide in January to April 2026, but it sells AI monitoring tools. So sell presence in the decision, not visits.
Objections about proof
- “You can’t measure it.” You can, with limits. SparkToro ran a study with Gumshoe.ai, which sells AI tracking. In it, the chance of ChatGPT or Google’s AI repeating the same list of brands was below 1 in 100. So measure the presence rate across many repeated prompts, not the position.
- “Let’s wait.” Waiting has a cost, and the pilot keeps it small. In a Scribewise and Scrunch survey of 602 US marketing and PR professionals, 62% felt their organization was already behind competitors on AI search visibility. However, only 45% were testing brand visibility across AI platforms. Scribewise is a B2B marketing agency and Scrunch sells an AI search platform, and the release gives no field dates. So the client may not be in a hurry, but most marketers surveyed say they are behind.
How do you propose a low-risk pilot to sell GEO?
The pilot is the safest way to sell GEO services to clients who don’t know the result yet. Propose two months for one site, with scope, price and exit criteria defined beforehand. The first month measures the baseline and fixes what blocks; the second repeats the measurement. If the client doesn’t want to continue, they keep the diagnosis and the report.
What goes into the pilot:
- Entry diagnosis, charged as a closed project, with the same prompts, engines and competitors in both months.
- Countable fixes, for example crawlers allowed, a sitemap in Bing and revised pages, which the client sees published.
- Repeated measurement, with the same frozen prompt list, for example 50 prompts, 5 runs and 1 engine, to compare the second month with the baseline. Those 250 answers per month have a margin of ± 6 to ± 14 points, according to the measurement guide.
- Closing report, with what changed, the margin of error and the proposal for the monthly scope.
Read the result with care. In two months, treat any change as an indication, because only a change above the margin and confirmed the following month counts as real. So agree on this rule with the client in the proposal. The GEO pitch ends at this point, and the structure of the diagnosis and the monthly scope is in the guide to GEO for agencies. To write the pilot proposal, use the GEO proposal template. To set the price of the pilot and the monthly scope, see how much to charge for GEO as a retainer, project or package.
What should you promise when you sell GEO services?
Promise deliverables and method, not citations. According to OpenAI, placement in ChatGPT search results isn’t guaranteed. Likewise, Google reminds site owners that using a service or tool doesn’t guarantee ranking success. So the GEO pitch and the contract describe what the agency does, not what the AI answers.
| Can promise | Don’t promise |
|---|---|
| Diagnosis with prompts and margin of error | Citation or position in ChatGPT |
| Technical fixes and revised pages | A deadline for the brand to appear |
| Monthly report with the same method | Sales growth from the AI channel |
After all, an informed client will compare proposals. The guide on how to choose a GEO agency lists the promises that the client learns to treat as red flags. So your proposal must pass that test. Also, if you sell GEO to clients with countable deliverables, you stay protected when the AI answer changes, which varies from one run to another.
Frequently asked questions
Yes, but the pilot starts with the basics. Google says that, to appear in generative AI features in Search, a page must be indexed and eligible to be shown in Search with a snippet. So the diagnosis checks crawler access and indexing before it measures the answers.
No. In a small pilot for one client, manual prompt collection, GA4 and Search Console are enough, and they cost nothing. With a few hundred answers per month, collection moves to a tool or a script, and the comparison of GEO tools helps you choose. NotHuman, which publishes this blog, sells AI tracking, so weigh our view with that in mind.
SEO is sold by position and traffic, while GEO is sold by presence in AI answers, measured over many repeats. For Google, both are the same work in Search. So the agency can sell SEO and GEO in the same contract, with GEO as an add-on.
Main sources
- AgencyAnalytics, marketing agency benchmarks report (2026).
- G2, survey on AI search and software buying (2026).
- Google Search Central, guide to optimizing for generative AI features (2026).
- OpenAI, how ChatGPT web search works (2026).
- SparkToro, study on the inconsistency of AI brand recommendations (2026).
- Scribewise and Scrunch, “Moving fast, flying blind” AI search survey of US marketers, via press release (2026).
- Pew Research Center, clicks on results with an AI summary (2025).
- SE Ranking, study of the traffic that comes from AI search engines (2026).



