AEO vs GEO vs SEO: Does Ranking on Google Get You Cited by AI? We Checked 189 Page-One Results

People search "geo vs seo" about 2,900 times a month in the US, "aeo vs seo" about 1,900 times and "aeo vs geo" about 1,600 times, according to Google Ads volumes from DataForSEO. Most of the pages ranking for those searches are definitions. They tell you AEO and GEO are about AI answers and SEO is about rankings, then list tactics.

The question I wanted answered sits underneath: if your SEO is already good, how much of AEO and GEO do you get for free? So I wrote down a method, asked 24 software buying and how-to questions on Google and on four AI surfaces on the same day, and checked how many of the pages each AI surface cited were already on Google's first page.

24

questions, one day

189

Google page-one results

732

AI citations on 4 surfaces

$0.51

DataForSEO spend

Collected 15 October 2026 through DataForSEO, US English, desktop: Google organic + AI Overviews, Google AI Mode, ChatGPT (search forced on) and Perplexity Sonar. The spend includes keyword research and a one-question pilot.

Download the dataset (CSV, 925 rows): every page-one Google result and every page each AI surface cited, with the question, its type, the source type and whether the cited page ranked.

AEO, GEO and SEO in a few lines

AEO, answer engine optimization, is the older term. People used it for featured snippets and voice assistants, which read one answer aloud. GEO, generative engine optimization, comes from a 2023 paper by Aggarwal and colleagues that tested which content edits made a page more visible in AI-generated answers. Today the two words describe the same job, and the vendors who sell it pick a side mostly on branding. Profound, for one, wrote that "it is our professional opinion that GEO is a bad term" in its own AEO vs GEO post and argues the two are the same thing. I agree with the conclusion and don't care much about the name.

SEO is the one that actually differs. Here is how I would separate them:

Google's own position is that the two overlap on its surfaces. Its page on AI features and your website says "the best practices for SEO remain relevant for AI features in Google Search (such as AI Overviews and AI Mode)," and that a page has to be indexed and eligible to be shown to be a supporting link. It also says both features "may use a 'query fan-out' technique," issuing "multiple related searches across subtopics and data sources." Google makes no claim about ChatGPT or Perplexity, and neither do I without data.

What page one for these searches says, and what it leaves out

I pulled Google's top 10 for "aeo vs geo", "geo vs seo" and "aeo vs seo" on 15 October. The ranking pages come from Adobe, Profound, Jasper, StoryChief, Foundation and CXL, a string of agencies and AI-SEO tool makers, and a few Reddit threads and YouTube videos. Most of them define the terms and give tactics. StoryChief's GEO vs SEO post offers two anecdotes, a Perplexity answer that cited the pages ranked 6 and 10 ahead of the one ranked first, and an AI Overview whose most-cited page was outside the top 50. Profound's post mentions its research found "a surprisingly small overlap between Google and ChatGPT" but gives no figure there.

The search results page made the point better than any of them. Google's own AI Overview for "geo vs seo" cited 13 pages, and none of them was among the 9 ordinary results under it. For "aeo vs seo" it cited 3 pages, none ranking. For "aeo vs geo", 3 of its 10 citations ranked.

What I could not find on page one was a measurement across many questions and several engines. That is what the rest of this post is.

What I ran

I wrote the method down before collecting anything. Four B2B software categories (CRM, project management, email marketing, help desk) and six questions in each: two best-X ("best crm for b2b sales teams"), two A-vs-B ("zendesk vs freshdesk") and two how-to ("how to warm up a new email domain"). They are phrased the way people type into Google, short and lowercase, and the same string went to every surface.

On 15 October 2026, in US English, each question went once to:

• Google search, desktop, keeping the ordinary results on page one (7 to 9 per question, since video and forum blocks take some slots) and the AI Overview on the same page. An overview appeared for all 24.
• Google AI Mode.
• ChatGPT, through DataForSEO's capture of chatgpt.com with search forced on. Every answer reported the same model.
• Perplexity, through its Sonar API with web search on. This is the API, and the perplexity.ai app can cite differently.

Everything went through DataForSEO, and the whole thing cost $0.51 including keyword research and a one-question pilot. A cited page counts as "in the top 10" only if it is the same URL as a page-one result for that question, after removing the query string and the fragment. YouTube links match on the video ID. Separately, I counted a citation as "same site" when its domain had any page in that question's top 10.

Finding 1: how much each AI surface overlaps with Google

Google AI Overviews17.7(35 of 198)

Google AI Mode11.8(13 of 110)

Perplexity (Sonar API)7.9(19 of 240)

ChatGPT with search3.3(6 of 184)

Share of cited URLs, %. 24 B2B software questions (best-X, A-vs-B and how-to), each asked once on every surface on 15 October 2026, US English. "Top 10" is the ordinary results on the first page of the same Google search, 7 to 9 per question. A match is the same URL with query string and fragment removed.

The Google surfaces overlapped most with Google's rankings, which is what you would expect. Even there it was a minority: 35 of the 198 pages AI Overviews cited (17.7%) were on page one for the question, and 12 of the 24 overviews cited at least one page-one result. AI Mode was at 11.8%. Perplexity cited exactly 10 pages per answer, and 19 of its 240 were on page one.

ChatGPT barely touched Google's first page. It cited 184 pages and 6 of them ranked there, and in 20 of 24 answers it cited no page-one result at all. The pages its searches retrieved before answering tell the same story: 345 distinct pages, 7 of them on Google's page one for the question.

Seven of the 24 questions got no page-one citation from any of the four surfaces. "hubspot vs salesforce" and "best crm software for small business" are two of them.

Looked at from the other side, most of what ranks went uncited:

Cited by none73(138 of 189)

Cited by one18(34 pages)

Cited by two7.4(14 pages)

Cited by three0.5(1 page)

Cited by all four1.1(2 pages)

Share of 189 ordinary page-one results, %. Surfaces: Google AI Overviews, Google AI Mode, ChatGPT with search, Perplexity Sonar. Both pages cited by all four were a vendor's own A-vs-B comparison page (ActiveCampaign vs HubSpot, Zendesk vs Freshdesk).

Of the 189 page-one results, 138 were cited by nothing. Two were cited by all four surfaces: ActiveCampaign's comparison page for "activecampaign vs hubspot" (ranked 2) and Zendesk's for "zendesk vs freshdesk" (ranked 3). Both are a vendor's own head-to-head page on an A-vs-B question.

The AI surfaces also disagreed with each other. They cited 611 different URLs in total, and 519 of those (85%) were cited by one surface only. The two Google surfaces shared the most, 53 URLs. ChatGPT shared 7 with AI Mode and 10 with AI Overviews.

Finding 2: it depends on what you ask

A-vs-B questions gave the most overlap on every surface except AI Overviews. When someone types "asana vs monday", the vendors' own comparison pages and a few review-site comparisons rank, and the AI surfaces cite that same small set of pages. For "asana vs monday", every one of the 8 page-one results was cited by at least one surface.

Best-X was the odd one. ChatGPT cited 85 pages across the eight best-X questions, and none was on Google's page one. Its searches explain most of it. Across the 24 answers it ran 45 searches, 28 of them containing "pricing" and 15 containing "official". For "best crm for b2b sales teams" it searched for official CRM pricing pages and cited HubSpot, Salesforce, Zoho, Pipedrive and Microsoft product and pricing pages, which is not what ranks for the question a person types. Our five-engine run found the same habit two weeks earlier: 63% of ChatGPT's cited pages were on the site of a brand the answer named.

How-to questions were the other surprise. AI Mode gave no links at all on 4 of the 8 how-to questions, including "how to set up a sales pipeline in a crm", and wrote a step-by-step answer instead. I can't tell from the capture whether that is what a person would have seen on screen.

Same page or same site

Counting by site instead of by page makes every number bigger. That matters if you think in terms of "our domain ranks, so we're probably fine."

Google AI Overviews

same URL17.7%

same site28.3%

Google AI Mode

same URL11.8%

same site22.7%

Perplexity (Sonar API)

same URL7.9%

same site10.4%

ChatGPT with search

same URL3.3%

same site12.5%

ChatGPT cited 85 pages for the 8 best-X questions. None of them was on Google's page one.

Share of cited URLs, %. Same site means the cited page's registered domain had any page in the top 10 for that question. YouTube and Reddit count as one site each, which flatters the Google surfaces a little.

For ChatGPT, the jump from 3.3% to 12.5% is mostly vendors. On "best help desk software for small business", Zoho's help desk page for small businesses ranked, and ChatGPT cited Zoho Desk's pricing page instead. On "activecampaign vs hubspot" it cited both companies' pricing pages and home pages next to the comparison page that ranked. Same site, different page. If your site ranks for a question, that is some evidence the AI surfaces know your domain. It says little about which of your pages they will cite.

Ranking first did not help much

If AI citations followed Google's rankings, the top three results would be cited most often.

Google AI Overviews

positions 1-318.1%

positions 4-1018.8%

Google AI Mode

positions 1-38.3%

positions 4-106.0%

Perplexity (Sonar API)

positions 1-36.9%

positions 4-1012.0%

ChatGPT with search

positions 1-35.6%

positions 4-101.7%

Ranking in the top three did not raise the AI Overview citation rate (13 of 72 against 22 of 117).

Share of page-one results the surface cited, %. 72 results at positions 1 to 3 and 117 at positions 4 to 10, across 24 questions. Small counts: ChatGPT's 5.6% is 4 pages.

For AI Overviews, positions 1 to 3 and positions 4 to 10 were cited at almost the same rate, 18.1% and 18.8%. The result at position 1 was cited in 2 of 24 overviews. That fits what we saw in the 100-overview study, where position 1 was cited less often than position 4. ChatGPT cited 4 top-three pages and 2 lower ones, which is too few to read anything into.

What kind of pages got cited

I labelled every cited domain by hand as a software vendor (a company selling any software product), a publisher (media, review sites, agencies, consultants and personal blogs) or user content (YouTube, Reddit, Quora, LinkedIn, Medium). The labels are in the CSV.

The two Google surfaces cited video and forums far more than Google's page one did. AI Overviews linked YouTube 45 times, in 16 of the 24 overviews, and Reddit 8 times. AI Mode linked YouTube 17 times and Reddit 10 times. None of those 62 YouTube links was a video from the ordinary results under it, so ranking did not bring them in. ChatGPT and Perplexity cited no YouTube, Reddit, Quora or LinkedIn page at all, which matches the zero we found for ChatGPT in the earlier five-engine run.

Perplexity's 88% vendor share needs a footnote. Many of those are small software companies' blog posts ranking tools in their own category, so "vendor" here often means a listicle written by a vendor, the same pattern we saw in our Claude citation study.

How this compares with other measurements

Two weeks ago our AI Overviews study found 35% of 683 overview links were top-10 pages. This run found 17.7% of 198, and 21.6% (11 of 51) on best-X questions alone, the type that study used. I don't think the earlier number was wrong. The questions here are short and keyword-style rather than "what is the best ..." sentences, they cover three intents rather than one, the run is a quarter the size and it is two weeks later. I can't tell which of those moved the number. Ahrefs, with 863,000 keywords and 4 million AI Overview URLs, reported 37.9% in March 2026.

Semrush's July 2025 comparison of 5,000 keywords is much further from us. It reported 82% URL overlap with Google's top 10 for Perplexity, 67% for AI Overviews and 35% for AI Mode, with ChatGPT lowest. The order matches ours except for Perplexity. The levels don't come close. Their study is 15 months older, used different questions, and I don't know exactly how they counted a match or which Perplexity product they queried, so I would not try to reconcile the two. What both agree on is that ChatGPT overlaps with Google least.

So how much of AEO and GEO is just SEO?

On Google's AI surfaces, some of it. A page that ranks has a chance of being cited, roughly one in five in this run for AI Overviews, and Google says its normal SEO guidance applies. On ChatGPT, close to none of it, at least on page-one terms: it runs its own searches, often for brand-specific pages, and cited a Google page-one result 6 times out of 184. Perplexity sat between them.

If I were deciding what to do with this, it would go roughly in this order:

1. Keep doing SEO for Google. It is the only one of these channels where ranking and citation visibly move together, and AI Overviews and AI Mode sit on the same results page.
2. Make your own product, pricing and comparison pages answer the obvious questions in plain words. 28 of ChatGPT's 45 searches mentioned pricing, and the only two pages cited by all four surfaces were vendor comparison pages.
3. For each question you care about, look at what each AI surface actually cites before you write anything. The free AI Overview checker lists an overview's sources, and the ChatGPT visibility checker runs one search-grounded question and shows what it cited. If the overview for your category is mostly YouTube, a blog post is competing for the smaller share.
4. Track the surfaces separately. With 85% of cited pages cited by one surface only, a single "AI visibility" number would hide most of this.

The wider playbook, from passage-level writing to measurement, is in the GEO guide.

Tracking it with CueScout

CueScout runs the questions you choose every day on ChatGPT, Perplexity, Google AI Overviews and, depending on plan, Gemini and Google AI Mode, with Claude and Grok as add-ons. For each answer it stores the cited URLs in order and the brands named, and for AI Overviews it also stores where each cited page ranked in the ordinary results on the same page, which is the comparison this post is built on. If you want one reading without signing up, the AI visibility checker asks ChatGPT, Perplexity, Gemini and Google AI Overviews a single question at once.

Limits

• 24 questions, all about B2B software, asked once per surface on one day in US English. Every percentage here is a snapshot, and several rest on small counts.
• I wrote the questions. They are short and keyword-style. Longer, conversational prompts may produce different citations, especially on ChatGPT.
• ChatGPT came from DataForSEO's capture of chatgpt.com with search forced on, and Perplexity from the Sonar API, which returned exactly 10 citations per answer. The perplexity.ai app and a logged-in ChatGPT user may see other sources.
• "Top 10" means the ordinary results on Google's first page for the exact question, 7 to 9 per question here. A cited page that ranks 11th, or ranks for a related search, counts as not ranking. Fan-out could explain some of the gap and this run does not test it.
• AI Mode returned no links for 4 how-to answers. They are excluded from the AI Mode citation counts, and I can't tell whether links failed to load or were never shown.
• Source types were labelled by hand at the domain level, and the vendor and publisher line is a judgement call for agencies and review sites. The CSV has every label.