Is SEO Dead? What AI Search Really Changed

SEO is not dead. The part of SEO that gets a page crawled, rendered and indexed is now the gate to most AI answers as well.

Google is explicit about this. Its AI features documentation says a page must be "indexed and eligible to be shown in Google Search with a snippet" to appear as a supporting link in AI Overviews or AI Mode. The same page says "There are no additional technical requirements." Google's separate guide to optimizing for generative AI features says you don't need new machine-readable files, AI text files or special markup to appear in its AI features.

So the question "is SEO dead with AI" has a plain answer. The foundations of search engine optimization still apply. What changed is the surface on top: answers instead of ten links, fewer obvious clicks, and thinner reporting. For a change-by-change breakdown, see how AI is changing SEO.

The next move is not a new discipline. It's checking that the old gates are open for every engine that matters to you.

It helps to split an AI answer into four stages.

1. Parametric: what the model knows from training, with no web access. It's frozen at a knowledge cutoff. Nothing you publish this quarter reaches it.
2. Access: whether a crawler may fetch and store your page. You and your CDN control this.
3. Retrieval: whether your page enters the model's context for a given question. The engine controls this, partly based on its index.
4. Generation: which retrieved pages the model actually uses and cites. The model controls this.

SEO lives mostly in stage 2, and stage 2 is a prerequisite for stages 3 and 4.

Engine by engine, as of October 2026

• Google AI Overviews and AI Mode draw on Google Search's index. There is no robots.txt token that blocks AI Overviews separately. Googlebot rules are the control.
• ChatGPT search relies on OAI-SearchBot. OpenAI says sites opted out of OAI-SearchBot "will not be shown in ChatGPT search answers." ChatGPT also rewrites your question into one or more search queries before it retrieves anything.
• Microsoft Copilot grounds its answers in Bing's index.
• Perplexity uses PerplexityBot to build its search index.
• Claude uses Claude-SearchBot for search quality.

Otterly's 2026 experiment with two AI-generated blog sites shows the dependency clearly. When Google deindexed both sites, Google's AI surfaces almost stopped citing them. Bing kept them in its index, and Copilot kept citing them. Being indexed is necessary. It isn't sufficient: in the same test, ChatGPT barely cited the sites even while they ranked in Bing.

Ranking matters, but it's a loose proxy, not a guarantee.

Ahrefs found that 37.9% of URLs cited in Google AI Overviews also appeared in the first 10 results (March 2026). The rest came from deeper positions. Ahrefs changed its method between studies, so don't read its older and newer numbers as a measured decline. They come from different instruments.

Two things follow:

• A top-10 page is more likely to be in the pool Google's AI draws from. That's a reason to keep doing SEO.
• A top-10 page is not promised a citation, and a page outside the top 10 is not excluded. Google may run several related searches for one question, so the page that gets cited may rank for a sub-question you never tracked.

The practical read: keep ranking work for the queries that matter, and stop treating rank position as your AI visibility number. Our guide to how AI search features affect organic results covers the overlap in more detail.

The real shift is in what you can see, not in what gets you included.

Clicks

AI answers often resolve a question on the results page or in the chat. That changes how many visits a mention produces, and nobody publishes a reliable, general number for how much. Treat any single traffic-loss figure you see quoted with suspicion unless you can check its method.

Referrals

Referral tracking is uneven. OpenAI says ChatGPT adds utm_source=chatgpt.com to referral links, so those visits are easy to spot. Google AI Overviews emit no separate referral, so their clicks blend into ordinary Google organic traffic. That's why claims about ChatGPT's share of AI referral traffic mostly measure which engines tag their links.

Reporting

As of October 2026, Google Search Console's generative AI report shows impressions for AI Overviews and AI Mode. It shows no clicks, no click-through rate and no position. Bing Webmaster Tools has an AI Performance report with citations and grounding queries (the searches the AI actually ran), but no API for it yet.

No engine publishes an "AI rank." Any tool reporting your position inside a Google AI answer has no ground truth to check it against. Measure presence, accuracy and citations, engine by engine, and say which numbers are samples.

Most of what decides AI visibility sits outside any publisher's reach. A short list sits inside it, and those items have the strongest evidence.

Two items on the left deserve emphasis.

Rendering. Vercel's crawler study found ChatGPT and Claude crawlers fetch JavaScript files but don't execute them, so they can't read content that only appears client-side. Googlebot does render JavaScript. A pricing table built in the browser can be visible to Google and missing for ChatGPT.

Brand mentions elsewhere. When engines cite other sites instead of yours, being named on those sites is associated with appearing in the answer. That's an association in the evidence, not a promised effect.

Work from the gates inward. Each step removes a block that would make the later steps pointless.

1. Confirm Google indexing and snippet eligibility. Check key pages for noindex, nosnippet, max-snippet:0 and stray data-nosnippet attributes, often inherited from a template. In Search Console, check whether the property has opted out of generative AI features, including by inheriting a parent property's setting.
2. Read robots.txt bot by bot. Allow the search bots you want: OAI-SearchBot, PerplexityBot, Claude-SearchBot, Bingbot, Googlebot. Decide training bots (GPTBot, ClaudeBot, Google-Extended) separately. OpenAI says each setting is independent.
3. Test what your firewall actually returns. A robots.txt that allows a bot means nothing if your CDN answers it with a 403. Cloudflare changed its AI crawler defaults on 15 September 2026, so crawler access can break without you changing anything. The AI Crawler Checker requests a page as each AI crawler and shows which ones get turned away.
4. Put load-bearing facts in the initial HTML. Prices, plan names, locations and one-line definitions should be readable before JavaScript runs.
5. Check Bing separately. Verify the site in Bing Webmaster Tools. A healthy Google index tells you nothing about Bing, and Bing backs Copilot.
6. State specific facts plainly. A clear sentence with a price or number gives an engine something it can quote. Fix places where your own site states the same fact two different ways.
7. Measure per engine. Ask real buyer questions on each platform and record whether you appear and whether the facts are right. The free checker on this site runs up to three buyer questions as a snapshot.

If you want a compact version, the one-hour AI SEO fix list covers steps 1 to 4.

The "SEO is dead" story pushes teams toward work the evidence doesn't support, and away from work it does.

• Rebuilding for AI with markup. Google says structured data isn't required for its generative AI features and that Google Search ignores llms.txt. Ahrefs' matched test of 1,885 pages adding schema found no major lift in AI citations. Existing markup does no harm. It just isn't a lever for AI answers.
• Rewriting for an "AI-friendly" style. Google says you don't need to write in a specific way for generative AI search, and it rejects chunking as a requirement. The C-SEO Bench study found most citation-rewriting methods ineffective and often harmful.
• Mass-publishing AI pages. In Otterly's test, Google removed both 1,000-post AI sites with no manual action notice. Google's spam policy targets scaled content that adds no value, however it's made.
• Blanket-blocking AI bots. Blocking OAI-SearchBot removes you from ChatGPT search answers. Allowing GPTBot does not get you into ChatGPT search, because GPTBot governs training only.
• Trusting one blended score. Google says AI Overviews and AI Mode "may use different models and techniques." An average across engines hides which one is failing. See whether one dashboard can track every AI engine before you buy one.
• Assuming Bing equals ChatGPT. Ranking in Bing produced almost no ChatGPT citations in Otterly's test. Check each engine directly.

The pattern behind all of these: they add work at the generation stage while the access stage is still unchecked. Fix access first.