AI Search Optimization: How AI Overviews, ChatGPT and Perplexity Choose Businesses

What AI search optimization means: AIO, GEO and AEO

Google's guide spells out the acronyms, AEO for answer engine optimization and GEO for generative engine optimization, and then settles the question: "From Google Search's perspective, optimizing for generative AI search is optimizing for the search experience, and thus still SEO." [1]

The term GEO comes from a research paper first submitted in November 2023 and published at KDD 2024, which proposed "Generative Engine Optimization" as a framework for website owners [6]. Semrush describes GEO as optimizing your presence to appear in responses from AI-powered search systems, and AEO as practices that increase your brand's visibility in AI-generated answers [7][8]. AIO can mean Google's AI Overviews or AI optimization in general, and Microsoft's own guidance shrugs at the naming: "Whether you call it GEO, AIO, or SEO" [5]. Our AEO vs GEO explainer takes each term in turn.

Google's hiring advice now addresses these services directly: "If they have advice on optimizing for AI experiences (also known as "AEO" "GEO" services), is their advice aligned with Google Search's official guidance on optimizing for generative AI features?" [9]

One note on our own wording. "AI suggestions and answers" is our phrase for what AI tools say, not a platform term. Google says AI Overviews, AI Mode and generative AI features; OpenAI says ChatGPT search; Microsoft says Copilot Search.

Where AI answers appear, and how often

Google is candid about frequency: "AI Overviews are only shown when our systems determine that it is additive to classic Search, and as such, often don't trigger." [2] Measured frequency depends on the sample and the date:

Pew's data and our own checks point the same way: AI Overviews appeared far more often on question-style and informational searches than on short or commercial ones. Pew found them on 8% of one- or two-word searches and on 60% of searches that began with a question word [20]. In our Las Vegas check, the map results showed on 11 of the 12 commercial searches and an AI Overview on 2 [23].

How Google's AI Overviews and AI Mode pick sources

Google describes the retrieval step as "relying on our core Search ranking systems to retrieve relevant, up-to-date web pages from our Search index." [1] It defines query fan-out as "A set of concurrent, related queries generated by the model to request more information and fetch additional relevant search results to address the user's query." [1] Fan-out, Google says, allows it "to display a wider and more diverse set of helpful links associated with the response than with a classic web search" [2].

AI Mode goes further. Google's help page says it divides a question into subtopics and searches for each one at the same time, and shows a set of web links when it lacks enough confidence in an answer [24]. Google also says AI Overviews appear when its systems judge generative AI especially helpful, and admits that "AI Overviews can and will make mistakes." [25]

Two practical consequences:

1. The cited pages are not simply the organic top ten. Studies disagree on the overlap: BrightEdge found 16.7% of citations came from the top ten (September 2025), Ahrefs found 76% (July 2025), and Ahrefs' own March 2026 update put it near 38% [26][27][28]. Our article on how to rank in AI Overviews compares the studies.
2. AI Mode and AI Overviews behave like different products. Ahrefs found only 13.7% of citations overlapped between them across 730,000 response pairs [29].

What helps a page get picked, in Google's words: "Our AI systems take a look at a variety of sources, so it can be helpful to have a unique viewpoint that stands out." [1] And: "Focus on making unique, non-commodity content that visitors from Search and your own readers will find helpful and satisfying." [30]

How ChatGPT, Perplexity and Copilot find businesses

ChatGPT. OpenAI's help page says ChatGPT search "sometimes partners with other search providers" and "typically rewrites your query into one or more targeted queries" [31]. It ranks results "using multiple factors intended to help users find relevant, reliable information," and says plainly: "Placement is not guaranteed." [31] To be eligible, OpenAI says to allow OAI-SearchBot to crawl the site, and hosts and CDNs must let OpenAI's published IP addresses through [31]. For local questions, ChatGPT can use an approximate location based on your IP address, and it draws on third-party providers for business listings without naming them [31]. Our ChatGPT SEO article covers it step by step.

Perplexity. Perplexity SEO starts with its crawler. Perplexity says PerplexityBot "is designed to surface and link websites in search results on Perplexity" and is not used to crawl content for AI foundation models [4]. Its user-triggered fetcher, Perplexity-User, "generally ignores robots.txt rules," by Perplexity's own account [4].

Copilot. Microsoft says Copilot Search "cites its sources prominently" [14]. Its guidance says Copilot is "Powered by Bing's search index" and that "Crawlability, metadata, internal linking, and backlinks remain essential," adding that "there's no secret strategy for being selected by AI systems" [5].

Gemini and Google Maps. Google says the Gemini app answers place questions from public Google Maps information, such as addresses, ratings and opening hours [32]. Ask Maps draws on more than 300 million places and reviews from more than 500 million contributors [18].

What Google says you need, and what you don't

Three lines from Google's documentation settle most of the debate:

• "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." [2]
• "To be eligible to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet, fulfilling the Search technical requirements." [2]
• On llms.txt and similar files: "Doing so will neither harm nor help your site's visibility or rankings in Google Search, as Google Search ignores them." [1]

The llms.txt proposal came from Jeremy Howard in September 2024 [35]. Chrome's Lighthouse treats the file as optional in its AI-agent audits, which have nothing to do with Search ranking [36], and when our research checked OpenAI's, Perplexity's and Microsoft's crawler and help pages, none committed to using it. Publishing one does no harm. Expecting it to get you cited is the mistake.

Two newer rules matter. Search Console now has a per-site switch to exclude a site from Search's generative AI features; sites are included by default, and Google says the switch isn't used as a ranking signal for the rest of Search [37]. Google's guide says a site must be included there to be eligible [1]. And Google's spam policies now name "attempting to manipulate generative AI responses in Google Search" [38]; mass-producing pages for each fan-out query to do that counts as scaled content abuse [1].

The foundations: indexable pages, clear answers, accurate profiles

Clear answers, without contortions. Google says "You don't need to write in a specific way just for generative AI search." [1] Microsoft takes a more structural view: assistants break content into smaller pieces, and with question-and-answer pairs they "can often lift these pairs word for word" [5]. One habit satisfies both: put each common question on the page with a plain, complete answer beneath it.

Accurate profiles. Google's guide says: "Using products like Merchant Center (such as Merchant Center feeds) and Google Business Profiles can help your products and services to be visible in both AI responses and other Google Search results." [1] Google's John Mueller adds: "ensure that your Merchant Center and Business Profile information is up-to-date" [30]. Microsoft tells local businesses to register with Bing Places so their address, hours and contact details stay eligible for AI answers [34]. Our Google Business Profile optimization guide covers the profile in full.

Reviews, citations and mentions as AI inputs

Mentions. In Ahrefs' study of 75K brands, web mentions correlated with AI Overview brand visibility at 0.664, against 0.218 for backlinks, and 26% of brands had no mentions in AI Overviews at all [39]. A later Ahrefs study across ChatGPT, AI Mode and AI Overviews found YouTube mentions correlated most strongly, at about 0.737 [40]. These are correlations, not proof of cause. A 2025 research preprint found AI search engines showed "a systematic and overwhelming bias towards Earned media" [41].

Websites and listings. Yext analyzed 6.8 million AI citations for location searches from July to August 2025 and found 86% came from sources brands control, websites (44%) and listings (42%); reviews and social supplied 8%, and Reddit and similar platforms 2% [42]. Each engine leaned differently: Gemini favored websites (52.1%), OpenAI leaned on listings (48.7%), and Perplexity spread across sources such as MapQuest and TripAdvisor [42]. Yext sells listings software, so read it as a vendor study.

Expert opinion. Whitespark's 2026 survey of 47 local search experts weighted AI search visibility as on-page 24%, reviews 16%, citations 13%, links 13% and Google Business Profile 12%, and found "3 of the top 5 AI Search visibility factors are citation factors" [43]. Its top-rated factor was presence on expert-curated "best of" lists [43]. It is a survey of opinion, not a measurement of any engine.

Reviews. In BrightLocal's 2026 survey of 1,002 US adults, ChatGPT was the tool most used for business recommendations (31%), followed by Google's AI Mode (23%) [44]. Among AI users, 63% trusted the recommendations, and 88% fact-checked the reviews the AI cited [44]. SOCi found ChatGPT recommended just 1.2% of the brand locations it studied, against a 35.9% average appearance rate in Google's local 3-Pack, and the locations it did recommend averaged 4.3 stars [45].

The guardrails are the same as for any review program. Google's policies ban incentives for reviews and review gating [46], and Google says inauthentic mentions aren't as helpful as they might seem [1]. Earn reviews by asking every real customer; earn mentions by being worth mentioning. See do Google reviews help SEO and local citations.

What the research says about writing for AI answers

The setup matters for how far you can trust it. The researchers built "GEO-bench, a benchmark consisting of 10K queries from multiple sources," and for each query "only the top 5 sources are fetched from the Google search engine"; answers came from "the gpt3.5-turbo model" [6]. They measured visibility two ways, Position-Adjusted Word Count and Subjective Impression, across nine rewriting tactics [6].

Two findings stand out. Lower-ranked pages gained the most: in the lab, pages ranked fifth that cited their sources gained visibility, while the top-ranked page lost some on average. And the authors are clear about the limit: "owing to the black-box nature of search engine algorithms, we didn't evaluate how GEO methods affect search rankings." [6]

AI tools also get things wrong. The Tow Center at Columbia Journalism Review tested eight AI search tools on news excerpts and found that, collectively, they gave incorrect answers to more than 60 percent of queries, and that generative search tools fabricated links [47]. Check what the major assistants say about your business, and fix the sources behind any error.

Local businesses in AI answers

Google's guide says: "Where appropriate, generative AI responses can include product listings, product information, and information about local businesses." [1] What the data shows so far:

• SE Ranking found that when AI Mode links to Google, 97.9% of those links point to Google Maps business profiles; such links appeared in 9.2% of responses [48].
• Sterling Sky found AI local packs on about 7% of the keywords it tracks (US, mobile), surfacing about 32% as many businesses as traditional 3-packs [49].
• SOCi's figures for ChatGPT, 1.2% of locations recommended against 35.9% appearing in the 3-Pack, point the same way: AI answers name few businesses [45].
• In our own check on September 27, 2026, 12 commercial searches made from Las Vegas showed the map results on 11, organic results on all 12 and an AI Overview on 2 ("dental implants" and "botox"); neither AI Overview named a business from the map results [23].
• In a US-wide capture for "near me seo" the same day, the AI Overview named two of the three map-pack businesses with their star ratings and review counts, and linked to neither. It is one observation, but it shows profile data surfacing inside an AI answer [23].

Whitespark's guide to AI Mode adds two cautions: each user may see different AI Mode results for the same query, and AI can pull from outdated or unrelated sources [50]. Keeping your profile, site and listings current is the defense.

Measuring AI visibility honestly

Counting rules worth knowing: an AI Overview occupies a single position in Search Console, and every link in it gets that position; clicking an external link in AI Mode counts as a click; and a follow-up question in AI Mode counts as a new query [54]. The generative AI report launched on June 3, 2026 and reached all sites by August 31, 2026 [51].

Why samples, not rankings: SE Ranking ran the same AI Mode searches three times and found only 9.2% of URLs matched across all three tests [48]. Google also warns: "Be wary of third-party tools that promise ranking success or claim to use "internal" Google metrics." [1]

Clicks, honestly: Pew found users clicked a regular result on 8% of visits when an AI summary appeared, against 15% without one, and clicked a link inside the summary on 1% [20]. Seer Interactive found brands cited in an AI Overview earned 120% more organic clicks per impression than when not cited [55]. Google says clicks from results pages with AI Overviews are "higher quality," without publishing the data behind that claim [2].

What the first page guarantee counts in AI answers

The first page guarantee is a written promise with narrow terms. We guarantee up to 50% of the key phrases you could rank for will rank on the first page of Google. The three places are your Google Business Profile in Google's local map results, page one of the organic results, and AI suggestions and answers. The milestones are at least 20% of the agreed phrases by day 90, 40% by day 180 and 50%+ by the end of the first year, and at each one the share is measured separately in each of the three places.

ChatGPT, Perplexity and Copilot answers don't count toward that share. They are real channels, and the foundations in this guide help with them, but the written terms count only Google's own AI answer on the results page.

Google shows an AI answer for only some searches. In our check of 12 commercial searches made from Las Vegas on September 27, 2026, two showed one [23], and Google itself says AI Overviews "often don't trigger" [2]. So we never promise an AI answer on any particular search; under the written guarantee terms, the AI share is worked out over the phrases on your list for which Google shows an AI answer when we check. The first page guarantee does not promise a position, or traffic, calls, leads or sales, and it is offered after an assessment, only if you qualify.

How we measure the AI place in practice: for each phrase on the agreed list, we capture the results page from your agreed market and record whether an AI answer shows and whether it names or links your business. See how the first page guarantee works and our client results.

Start with what Google can already see

Most AI visibility work is SEO work that was skipped: pages Google can't index, profiles that are out of date, and no reason for anyone else to mention you. A free SEO audit is the place to start, and our organic SEO guide covers the groundwork. If you want the result in writing, read how the first page guarantee works.