SEO for AI Search: Is SEO Still the Backbone?

Is SEO the Backbone of AI Search? What Still Matters, and What Doesn't

Yes, SEO is the foundation of AI search, but it's no longer the whole job. ChatGPT, Perplexity and Google's AI Overviews pull answers from search indexes, so a page that can't be crawled, indexed and trusted won't be cited. But ranking #1 no longer guarantees a citation. AI engines also weigh clear answers, brand mentions elsewhere and freshness.

That's the short version of SEO for AI search. The longer version matters, because most founders are hearing one of two wrong stories right now: "SEO is dead, you need AI optimisation instead" or "AI search is just SEO, nothing changes." The data supports neither.

Key takeaways

Google says there's nothing special you need to do for AI Overviews or AI Mode. The same SEO basics decide whether you're eligible to appear. But ranking and being cited are drifting apart. Studies now find a large share of AI citations go to pages outside Google's top 10. AI engines give heavy weight to what others say about you, such as mentions, reviews and community threads, not just what's on your own site. A practical way to think about it: be Eligible (SEO basics), Extractable (answer-first content) and Endorsed (mentioned by others). Track AI citations separately from rankings. They are now different scoreboards. How do ChatGPT, Perplexity and Google AI find content?

Each engine searches an index of the web, picks a handful of pages, and writes an answer from them. The engines differ in which index they use and how they choose.

Google AI Overviews and AI Mode

Google's AI features draw on Google's own search index. Google says there are no additional requirements or special optimisations needed to appear in them. A page needs to be indexed and eligible to be shown as a snippet, the same as any normal search result.

The twist is how the AI searches. AI Overviews and AI Mode can use a technique Google calls "query fan-out": they run multiple related searches across subtopics to build one response. So your page can be cited for a question it doesn't rank for, because it ranks for one of the sub-questions behind it.

ChatGPT

ChatGPT answers partly from what the model learned in training and partly from live web lookups. One analysis estimates roughly 60% of responses come from the model's memory, with the rest using real-time lookups through Bing's index.

OpenAI also runs its own search crawler, OAI-SearchBot. OpenAI states that sites which block OAI-SearchBot won't be shown in ChatGPT search answers, and that this is separate from GPTBot, its training crawler.

Perplexity

Perplexity runs its own index of more than 200 billion URLs. It leans heavily on community content: Reddit accounts for roughly 46.7% of Perplexity's top-10 source share in one analysis.

Is SEO for AI search still the backbone?

It's the backbone for eligibility. It's no longer the whole story for citation.

Yes: SEO decides whether you're in the running

If Google or Bing can't crawl your page, or your page is blocked from showing snippets, AI engines built on those indexes can't use it. Technical SEO, indexing, site speed and helpful content still decide whether you're even in the pool of pages AI can pick from.

No: ranking no longer decides whether you're quoted

This is the part most "AI search is just SEO" takes miss.

Ahrefs: the share of AI Overview citations from Google's top 10 fell from 76% in mid-2025 to 38% by early 2026. Ahrefs itself notes that part of the drop reflects better citation detection in its tools, so the two numbers aren't perfectly comparable. BrightEdge: only about 17% of the sources cited in AI Overviews also rank in the organic top 10, and that figure has barely moved over its tracking period. Conductor: across nearly 168 million AI Overview citations, the #1 organic page was cited about 25% of the time, and 57% of citations went to pages outside the top 10.

The studies use different methods, so the exact numbers vary. The direction doesn't: ranking helps, but it no longer carries you all the way.

Why ranking and citation drifted apart

Three things are going on:

Fan-out searches. The AI runs many related searches, so it pulls pages that answer sub-questions, not just the main one. Different selection logic. Search ranks whole pages for a query. AI picks passages that answer a specific part of the question cleanly. Randomness. AI answers vary run to run. A SparkToro study found ChatGPT and Google AI Overviews returned the same brand list less than 1% of the time on identical prompts. Any single check of "does ChatGPT mention us?" is a snapshot, not a ranking. What does AI search reward that classic SEO doesn't?

AI engines reward being mentioned and trusted across the web, and being easy to quote. Classic SEO has always touched these, but AI weighs them more heavily.

What others say about you

Ahrefs studied 75,000 brands and found the strongest predictor of showing up in AI answers was branded web mentions, meaning references to your brand on sites you don't control. It wasn't backlinks or domain rating. Another analysis found brand search volume correlated more strongly with AI citations than backlinks did.

For a business, that means press coverage, reviews, industry lists, podcast appearances and community discussion now feed directly into AI visibility.

Third-party sources the AI already trusts

Each engine has favourite sources. Wikipedia is ChatGPT's most-cited source at 7.8% of its citations, and Perplexity leans on Reddit and review sites. If your category is discussed on a few trusted sites, being present there matters as much as your own blog.

The mix also shifts. ChatGPT's share of Reddit citations fell from roughly 60% to 10% in mid-September 2025. Don't build your whole strategy on one platform.

Freshness

AI engines skew towards recent content. One 2026 research roundup found about half of cited content was under 13 weeks old. Pages with current numbers and a clear "last updated" date have an edge.

Passages that can be quoted on their own

This one is our take rather than a published study. The pages we see cited most often answer the question in the first few lines, use headings phrased as questions, and write sections that still make sense when lifted out of the page. An AI can't quote a paragraph that only makes sense after reading the three before it.

The Eligible, Extractable, Endorsed framework

This is how we think about SEO for AI search at Outrank. It's our own framework, not an industry standard.

1. Eligible: can AI find and use your page? This is classic technical SEO. It covers indexing, crawl access, snippet eligibility, speed and a sound site structure. Without it, nothing else counts.

2. Extractable: can AI lift a clean answer from your page? This is content structure. It means a direct answer up top, question-led headings, self-contained sections, specific numbers, tables and lists, and a named author with real experience.

3. Endorsed: do other sources back you up? This is off-site reputation. It covers mentions in press and industry lists, reviews, community discussion, and consistent information about your business across the web.

Classic SEO covered layer 1 well, layer 2 partly, and layer 3 mostly through backlinks. AI search raises the weight of layers 2 and 3.

How to do SEO that also wins in AI search: a playbook

These steps work on top of normal SEO, not instead of it.

Step 1: Fix eligibility first Make sure your key pages are indexed in Google Search Console and in Bing Webmaster Tools. ChatGPT leans on Bing, so don't skip it. Check your snippet settings. Snippet controls like nosnippet and max-snippet, and of course noindex, can also cut you out of AI Overviews. A setting added years ago for another reason may be blocking you now. Allow OAI-SearchBot in robots.txt if you want to appear in ChatGPT search. Watch for broad rules or firewall settings that block it by accident. Know what Google-Extended does. Blocking it affects Gemini training and grounding, not AI Overviews, which run on ordinary Googlebot crawling. You don't need to allow it to appear in AI Overviews. Don't wait for an "AI file". Google says you don't need new machine-readable files, AI text files or special markup to appear in its AI features. An llms.txt file won't hurt, but it won't substitute for the basics. Step 2: Write for extraction Open every page with a 40–60 word answer to the main question, in plain language. Phrase headings as the questions people ask, and open each section with a one- or two-sentence answer. Make every section stand on its own. Avoid "as mentioned above". Use specific numbers, dates and named examples, and cite sources for them. Show who wrote it and why they know, with a real author bio. Add structured data (Article, and FAQPage where you have real FAQs) that matches what's visible on the page. It helps machines understand the page but doesn't guarantee inclusion. Step 3: Cover the fan-out, not just the keyword

List the 5–10 follow-up questions someone would ask after your main query. Answer each clearly, either within the post or in closely linked posts. For "how much does performance marketing cost in India", those follow-ups include agency fee models, minimum ad budgets, what's included in a retainer, and how long before results show. This is what the AI's fan-out searches look for.

Step 4: Earn mentions off your site Get into the lists your buyers read, such as industry roundups, "best agencies in Bangalore" pieces, and directories like Clutch or GoodFirms for service businesses. Publish something only you can publish: original data, a benchmark, or a teardown. That's what earns citations and mentions from others. Show up where your category is discussed: podcasts, LinkedIn, relevant Reddit communities and industry newsletters. Participate genuinely. Planted promotional posts tend to get removed and damage trust. Keep your business details consistent everywhere: name, services, location and what you're known for. Step 5: Measure AI visibility separately Build a prompt set. Write 20–30 questions your buyers would ask an AI, and run them monthly in ChatGPT, Perplexity, Gemini and Google. Record who gets mentioned and cited. Watch referral traffic in GA4 from chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com. Check Search Console's generative AI report. It now has a performance report for AI features, but it shows impressions only, with no click or query data. Judge trends over months, not single checks, because AI answers vary run to run.

Checklist: SEO for AI search

• Key pages indexed in Google and Bing
• No accidental noindex, nosnippet or max-snippet limits on pages you want cited
• OAI-SearchBot allowed in robots.txt and not blocked by your firewall or CDN
• Every priority page opens with a 40–60 word direct answer
• Question-based H2s, each with a short answer up top
• Specific, sourced numbers and a visible "last updated" date
• Named author with a real bio
• Article and FAQPage schema that match visible content
• Follow-up questions covered in the post or linked posts
• Brand listed on the directories and roundups your buyers read
• Monthly prompt-set check across 3–4 AI engines
• AI referral traffic tracked in GA4

Frequently asked questions Is SEO still relevant now that people use AI search?

Yes. ChatGPT, Perplexity and Google's AI features all draw on search indexes, so a page that can't be crawled, indexed and trusted won't be cited. What has changed is that ranking well no longer guarantees a citation. SEO is now necessary but not sufficient on its own.

Do I need to do anything special to appear in Google AI Overviews?

Google says no. There are no extra requirements, special files or markup for AI Overviews or AI Mode. A page must be indexed and eligible to show a snippet, so avoid noindex or nosnippet on pages you want cited, then follow normal SEO best practice.

How does ChatGPT decide which websites to cite?

ChatGPT combines what the model learned in training with live web searches that lean on Bing's index and OpenAI's own crawler, OAI-SearchBot. Studies suggest it favours sources that are widely mentioned and trusted across the web, such as Wikipedia, review sites and well-known publications. Pages that give a clear, specific answer near the top are easier for it to quote.

Does ranking #1 on Google mean AI will cite my page?

Not necessarily. A large 2026 analysis by Conductor found the #1 organic result was cited in AI Overviews only about a quarter of the time, and 57% of citations went to pages outside the top 10. Ranking improves your chances, but citation also depends on how clearly your page answers the question and how much the web trusts your brand.

What is the difference between SEO, AEO and GEO?

SEO (search engine optimisation) aims to rank pages in search results. AEO (answer engine optimisation) and GEO (generative engine optimisation) are newer terms for being cited or mentioned in AI-generated answers. In practice they overlap heavily: AEO and GEO build on SEO foundations and add a stronger focus on answer-first content and off-site brand mentions.

Should I block AI crawlers like GPTBot?

It depends what you want to block. GPTBot is OpenAI's training crawler, while OAI-SearchBot powers ChatGPT search, and blocking OAI-SearchBot keeps you out of ChatGPT search answers. Similarly, Google-Extended controls Gemini training and grounding but doesn't affect AI Overviews. Many businesses allow the search crawlers and decide on training crawlers separately.

How can I tell if AI engines are citing my website?

Run a fixed set of 20–30 buyer questions through ChatGPT, Perplexity, Gemini and Google every month, and record which brands and pages are cited. Also track referral traffic from AI platforms in GA4, and check Search Console's generative AI report, which shows impressions from Google's AI features. Judge the trend over several months, because AI answers vary from one run to the next.

The bottom line

You were right to think of SEO as the backbone. It decides whether AI engines can find and use your content at all. But the part that decides whether you get quoted now lives in two other places: how clearly your pages answer questions, and how often the rest of the web vouches for you.

If you want to know how visible your business is in AI answers today, and what's holding it back, our SEO and AEO team can run a prompt audit across ChatGPT, Perplexity and Google and map the gaps across eligibility, content and reputation.