Spearleaf's generative engine optimization services work to get your business cited and recommended in ChatGPT, Perplexity and Google AI Overviews. GEO is the citation side of AI search optimization: reviews, directories, entity facts and pages AI crawlers can read. We run one engagement that delivers on both fronts, Google and AI answers, for local, mid-market and enterprise brands.
Google's local pack held. ChatGPT's shortlist moved.
Mean overlap (Jaccard) between two checks of the same local buying question. 1.0 means the same list both times.
Spearleaf study · 216 US markets · August 2026Generative engine optimization (GEO) is the work of getting AI engines to cite and recommend your business inside their answers.
Researchers led by Pranjal Aggarwal named it in a 2023 paper on generative engines. Their benchmark tests reported visibility gains of up to 40% in generative engine responses, with results varying by domain.
SEO optimizes to rank. GEO optimizes to be cited. Use the term to mean one thing: being the source an AI answer names.
Every few months a new acronym shows up, but GEO, AEO, AI SEO and LLMO name the same practice. Our SEO vs GEO vs AEO guide sorts out where each label fits.
SEO earns a spot in Google's ranked list. GEO earns a mention inside the answer an AI engine writes. You can be ranking on page 1 and still miss ChatGPT's shortlist for the queries that actually matter.
| SEO | GEO | |
|---|---|---|
| Goal | Rank a page in Google | Get named and cited in the AI answer |
| Surfaces | Google Search and the local pack | ChatGPT, Gemini, Perplexity, Claude, Grok, AI Overviews, AI Mode |
| What it reads | Your pages, links and Google Business Profile | Directories, review profiles, third-party sites and your pages |
| Stability (local study) | Local pack overlap 0.82 between checks, local buying questions, Aug 2026 | ChatGPT local panel overlap 0.36, same questions |
| Measured by | Rankings, clicks, Search Console | Presence rate over repeated asks |
Strong SEO is the floor, and GEO is the second surface. In our ChatGPT study, ChatGPT's picks overlapped Google's local pack on only 33.1% of cards. Read ChatGPT vs Google for why a ranking doesn't carry over.
The same #1 held in just 34.1% of markets (73/214). One check is one roll of the dice.
So no honest GEO agency can promise you a place in that answer, and we don't. We measure how often you appear across repeated checks, then work the sources behind the answer.
The model's taste is fairly steady, but its reading list moves. Read what AI answers reward for the pattern behind the businesses that stayed.
Joshua Albanese ran this study on local buying questions in 216 US markets in August 2026.
Google's local pack held at 0.82 overlap between checks, while ChatGPT's panel held at 0.36.
17.4% of resolved displayed ratings were Yelp's, so both review profiles count.
Businesses that stayed had a median of 226 reviews vs 164 for those that dropped.
Google's AI Overview fired on 25.9% of local buying queries.
Our GEO work covers 6 levers, each tied to what AI engines read before they answer. Start with the levers your AI visibility report flags, because changing what the engine reads is how the answer can move.
In our local study, ChatGPT's reading list changed between asks. We get you listed and accurate on the directories, comparison lists and review sites your vertical's answers cite.
Businesses ChatGPT kept showing had deeper review counts. We build steady review volume on both profiles, because ChatGPT displays ratings from each.
Vercel's crawler analysis found the major AI crawlers don't render JavaScript. We serve key pages as plain HTML with answer-first AEO content.
We keep your name, address, services and hours consistent everywhere, including your ChatGPT Business Hub listing, then add schema markup. Schema is hygiene, not a switch.
AI engines weigh what other sites say about you. We earn mentions on the trade sites, communities and press your buyers' questions pull from.
Mentions, citations, recommendations and inquiries each get their own number. We tie AI referrals to leads in GA4 and track Google clicks in Search Console.
Before work starts, we fix a set of real buyer questions and record who each AI engine names today.
We re-ask the same questions across ChatGPT, Gemini, Perplexity, Claude, Grok and Google's AI, because one ask is noise.
We trace each missed answer to the sources the engine read, then build your presence there.
Mentions, citations, recommendations and inquiries stay separate numbers. You see how often you appear and how that trends over time.
Real ChatGPT answers from one October 2, 2026 check of fixed buyer questions. One check is a snapshot, so results vary by market and by ask.
ChatGPT’s top recommendation for Reiki in Studio City, on every buyer question we tested
Real ChatGPT · 6 buyer questions · Oct 2026Buyer questions where ChatGPT recommends the same client when asked who to hire
Real ChatGPT · Oct 2026European car repair questions in Hermosa Beach where ChatGPT’s map shows our client
Real ChatGPT · Oct 2026Use these 6 checks, which we pulled from 14 buyer threads on Reddit about GEO agencies.
Ask for a locked baseline of real buyer questions before any work starts.
Ask for the method. Repeated checks beat a single screenshot.
Walk away from any promise of a #1 spot in ChatGPT.
Expect mentions, citations and inquiries reported as separate numbers.
Check that the plan goes past blog posts relabeled as GEO.
Ask who does the work, and whether a senior person stays on your account.
Half of it is. SEO is the floor, and Google's guide to AI features says the same basics apply. Yet ChatGPT's picks overlapped Google's local pack on only 33.1% of cards in our study. Treat GEO as the second surface on top of strong SEO.
No. Google's AI Overview fired on only 25.9% of local buying queries in our August 2026 study. Google's local pack was the steadiest surface, at 0.82 overlap between checks. Keep SEO running and add GEO for the AI answers.
No, and we don't promise it either. The same #1 held in only 34.1% of markets when we re-asked ChatGPT days later. Ask any GEO agency for its method and its baseline instead of a promise.
We ask a fixed set of buyer questions on repeat and report how often you appear. Our study ran the same way across 216 markets, with a 50-card hand audit that agreed on 49 of 50. Read our AI visibility guide for checks you can run yourself.
There's no honest fixed timeline, and our study holds no timeline data. We lock a baseline first, then you watch the trend across repeated checks. Ask any agency to show you that baseline before work starts.
Tools report presence. They don't change what ChatGPT reads. The work that can move answers is reviews, directory coverage, both rating profiles and crawlable pages. Use a tool to watch, and hire for the work.
We scope GEO after your AI visibility report, based on the engines, markets and buyer questions you need covered. You get a written quote after a free audit of your market.
Joshua Albanese is the founder of Spearleaf, a national SEO and AI search optimization agency. He designed and ran the 216-market ChatGPT study behind this page, and he works hands-on in every GEO engagement. He has built six service businesses from zero with organic content and SEO.
Your AI visibility report shows where you stand in ChatGPT, Gemini and Google's AI, and what building presence would take.
Get your AI visibility report