What people mean by AI SEO, GEO, and AEO
"AI SEO" is a sales label as often as it is a method. Some vendors mean "keep showing up when Google's AI Overviews or AI Mode answer a question." Others mean "get named when someone asks ChatGPT for a plumber near them." Those are not the same job, and treating them as one product is how owners end up buying files, chunking tips, and mention packages that Google itself says to ignore for Google Search.
Google Search Central's May 2026 guide is explicit: from Google's perspective, optimizing for generative AI features in Search is still SEO. The terms "AEO" (answer engine optimization) and "GEO" (generative engine optimization) describe work people want to do for AI answers, not a separate ranking system Google runs alongside Search.
Adoption numbers confuse the pitch further when they are mixed. In the U.S. Chamber of Commerce / Teneo Empowering Small Business report published August 18, 2025, 58% of small businesses self-identified that they use generative AI—up from 40% in 2024 and 23% in 2023. That is a broad, self-reported use question. The SBA Office of Advocacy's September 24, 2025 research spotlight, drawing on Census Bureau Business Trends and Outlook Survey (BTOS) data through August 2025, put AI use in producing goods or services at 8.8% of small firms (fewer than 250 employees). Same country, different question. Neither figure measures whether an answer engine names a local business.
| Name | What people claim | What the evidence supports |
|---|---|---|
| AI SEO | A new discipline that replaces ordinary SEO | For Google AI Overviews and AI Mode: same index, same eligibility rules; foundational SEO still applies (Google Search Central, May 2026) |
| GEO / AEO | Special markup, chunking, or llms.txt to "get into" AI answers | Google says those are unnecessary for its AI features; other engines are not documented as ranking on llms.txt |
| "Get listed on ChatGPT" | A Business Profile–style claim form | OpenAI documents crawler controls (OAI-SearchBot vs GPTBot), not a business listing form |
| Budget split (e.g. 20% of SEO for GEO) | A required spend mix | No primary source in this research set backs a fixed percentage |
What Google changed, and what it told site owners to ignore
Google's generative AI features in Search—AI Overviews and AI Mode—retrieve pages from the normal Search index. The May 2026 optimization guide describes retrieval-augmented generation (grounding answers in ranked pages) and query fan-out (related sub-queries that fetch more supporting results). The companion page AI features and your website states there are no additional technical requirements beyond being indexed and eligible for a snippet. A Search Console setting must also leave the site included in Search generative AI features.
For local and product answers, Google names Business Profile and Merchant Center as inputs. That is ordinary listing hygiene, not a new schema type.
The same May 2026 guide mythbusts tactics that still show up in agency "GEO" posts. For Google Search, including its generative AI capabilities, Google says you do not need llms.txt or other "special" AI markup; Google Search does not use them. Creating such a file neither helps nor hurts Google visibility. Google also says you do not need to "chunk" content into tiny pieces for AI, rewrite pages only for AI systems, chase inauthentic mentions, or invent special schema solely for generative AI. Structured data remains useful for rich-result eligibility overall; it is not a secret AI Overviews key.
That is Google's instruction for Google. It is not a description of ChatGPT, Perplexity, or any other assistant.
Those products crawl and cite under their own rules. Confusing the two is how "Google said ignore llms.txt" turns into "llms.txt does nothing anywhere," which is a stronger claim than the sources support. No engine in the source set for this article documents llms.txt as a ranking file. Optional files for other systems are fine; they are not a ranking strategy for Google.
What is actually different
Customers use AI to shortlist, then check elsewhere
BrightLocal's Local Consumer Review Survey 2026 (1,002 US adults, published February 11, 2026) found that use of ChatGPT and other generative AI tools for local recommendations rose from 6% the prior year to 45%. BrightLocal's consumer search-behavior channels research (2026) found that among consumers already using AI for local searches, only 18% felt ready to contact the business the AI named without further checking. Of consumers who started on an AI tool, 2% used it as their only channel, 43% went on to check Google, and 39% looked at social media.
So the profile, the reviews, and the website still close many of the jobs. AI names a shortlist; people verify. That is why local SEO in 2026 is not "replace Google with ChatGPT"—it is keep the verification surfaces accurate while the shortlist moves. (See local SEO in 2026 and AI Overviews.)
Engines do not cite the same sources
BrightLocal's 2026 AI directory-sources analysis found Google Business Profile (google.com) accounted for 28.63% of citations across the AI Overviews / AI Mode / ChatGPT mix it studied, and that Yelp appeared as a source in 80% of ChatGPT's local answers in that work. About 93% of cited domains were individual business websites—platforms dominate the top ranks, but the long tail is still company sites.
Steady Demand's Citation Ledger (Ben Fisher; Gemini API with Google Search grounding; 1,500 queries across 50 US metros; September 2026 update dated September 9, 2026) found Gemini's largest citation category was the business's own website at 49.6% (down from 58.8% in July). A re-run of the same design through ChatGPT cited the business's own site 10.5% of the time and general business directories 46.0% (versus 17.7% for Gemini). On identical questions, the two engines' cited domains overlapped 8.3% of the time, and they named the same top business 4.9% of the time.
One playbook does not cover both. Directory consistency matters more for ChatGPT-shaped answers; the service pages on your own site matter more for Gemini-shaped answers. (See how to get your business on ChatGPT and local citations for AI search.)
The answer moves
Complete SEO's September 2026 study of 44,988 ChatGPT local answers across 35 home-service trades and 24 metros found that asking the same question twice, minutes apart, returned the same list of named businesses only 3.6% of the time (188,520 pairs in a three-day drift arm). Steady Demand reported that asking Gemini the same question twice, the cited sources matched about 41% of the time.
A single ChatGPT or AI Overview screenshot is not a measurement. It is one draw from a moving distribution.
Analytics can show referrals and Google AI impressions; they cannot turn one lucky mention into a ranking claim. (See tracking AI traffic in GA4 and Search Console.)
AI can act, not only answer
Google's help documentation for AI Mode local availability and pricing describes researching service costs and open slots, completing forms, and calling businesses when information is not available online. Users can type "don't call" to keep the search online-only. Google Business Profile Help documents automated calls on behalf of customers for booking appointments, checking restaurant wait times, and confirming price or availability; calls are recorded; the feature is unavailable in Indiana, Louisiana, Minnesota, Montana, and Nebraska; businesses can opt out; Google never asks for payment.
If hours, services, and prices are missing from the profile and the site, the customer may still get an answer—after a robot calls your front desk, or after it skips you. (See why Google may be calling your business.)
A sequence for a small local business
No primary source in this research set supports a fixed "GEO budget percentage." Scope and fee belong in a written proposal with the work named—not in a blog post inventing a 15% or 50% split. What you can do without that fiction is a sequence: fix the facts customers and engines check, then measure.
Make the Business Profile complete and current. Hours, categories, services, photos, and attributes are the inputs Google names for local AI answers. (See Google Business Profile checklist for AI search.)
Put the same facts on the site, Yelp, Apple Business, and Bing Places, plus the directories your category's answers actually cite. Consistency beats a blast of low-quality listings. Do not solicit Yelp reviews. (See local citations for AI search.)
Let the crawlers in. OpenAI separates OAI-SearchBot (search surfacing in ChatGPT) from GPTBot (training). Allowing search is not the same decision as allowing training. Robots.txt changes for search can take about 24 hours. (See how to get your business on ChatGPT.)
Answer cost, timing, and "what's included" on the service page, not only in a blog post. Customers verify; agents call when the page is empty. (See local SEO in 2026 and AI Overviews.)
Answer the phone and the form quickly. If AI shortlists you and a person (or agent) reaches out, speed is operating work, not a keyword. (See AI receptionist for small business and speed-to-lead statistics for local service businesses.)
Reply to reviews with a human on the send button. AI can draft; the FTC Consumer Reviews and Testimonials Rule (effective October 21, 2024) bans fake reviews, including AI-generated fake reviews. (See using AI to reply to Google reviews.)
Read GA4, Search Console, and Business Profile performance monthly, and keep a repeated prompt log if you care about being named. Month one can show technical readiness and baseline impressions; it cannot prove a causal line from one AI mention to booked revenue. (See tracking AI traffic in GA4 and Search Console.)
What to skip
- Guaranteed placement in ChatGPT, AI Overviews, or any assistant. No listing form, no purchase path, and answers move run to run.
- Fake or paid reviews conditioned on sentiment. The FTC rule effective October 21, 2024, bans fake reviews (including AI-generated ones) and knowing violations can bring civil penalties.
- Mass directory blasts. Citation studies show category-specific sources and major platforms; they do not show that flooding hundreds of low-quality directories moves answer-engine naming.
- llms.txt as a ranking tactic. Google says it is unnecessary for Google's AI features and that Google Search ignores it for ranking. No engine in this source set documents it as a ranking file.
- Budget percentages copied from an agency post. Invented 15%, 20%, or 50% GEO splits are marketing copy, not measured guidance.
- A single ChatGPT screenshot used as a report. Complete SEO's identical-list rate of 3.6% and Steady Demand's engine disagreement figures exist so owners stop treating one draw as proof.
How to tell whether any of it is working
Leading indicators you can actually observe: Business Profile calls, messages, and direction requests; GA4 traffic from AI assistants where the product tags referrals (OpenAI documents utm_source=chatgpt.com on ChatGPT search referrals); Search Console generative AI impressions for Google's AI features (impressions, not a full click-and-query story); and a prompt log—same questions, multiple runs, logged dates, engines, and whether you were named. Report a rate with a band, not a screenshot.
Revenue has other causes—season, pricing, capacity, reviews, ads. Do not claim a straight line from an AI mention to a booked job.
Cleland & Co.'s free technical-readiness scan is a checklist: it inspects the submitted page plus root robots, sitemap, and llms files. It does not observe whether an answer engine names the business, and it does not run PageSpeed Insights or Lighthouse lab tests. Use it to see whether the site is fetchable and structured enough to inspect—not as a visibility score.
