Search used to be a list of ten links. Now a person can ask a question and get a written answer from Google, ChatGPT, Perplexity, Gemini or Claude, with a handful of sources attached. If you publish content, the question is no longer only “do I rank?” It is also “do I get used in the answer?”
That shift is what people mean by AI SEO. This guide covers what works, what is hype, and where to spend your time.
What AI SEO Really Covers
AI SEO has two parts. One is using AI tools to speed up research, drafting and analysis. The other is making your pages easy to discover, understand and cite in AI-powered search. This article is about the second part, because that is where most confusion sits.
Google Has Already Told Us the Core Answer
In May 2026, Google published its first dedicated guide to generative AI features in Search. It states that optimizing for AI Overviews and AI Mode remains part of SEO rather than a separate discipline. Google’s own framing is that optimizing for generative AI search is optimizing for the search experience, and so it is still SEO. letsdatasciencemedresponsive
The practical reasoning is simple. AI Overviews draw on Google’s core Search index, while AI Mode runs a “query fan-out” process that fires several related sub-queries through the same ranking signals as classic search. There is no hidden second door. A page that cannot be crawled, indexed and shown with a snippet has no realistic path into those answers. A page blocked by noindex or nosnippet is not eligible.
Google was just as direct about what you can skip. There is no special schema for AI Overviews, and no need to chop articles into artificial chunks for Google’s sake. Google also says there is no ideal page length for these features.
Step 1: Let the Right Machines In
Each AI platform has its own crawlers, and a careless robots.txt file can quietly remove you from the answers you want.
Reports based on vendor documentation describe OpenAI’s setup like this. OAI-SearchBot powers ChatGPT’s search results and citations, while GPTBot relates to training. That means you can decline training use and still stay eligible for citation. One audit story describes a site that blocked every AI bot it could find, including the exact crawlers behind citations. growwithsakibgrowwithsakib
Perplexity adds a wrinkle. PerplexityBot builds the index behind its results, and Perplexity recommends allowing it. Its separate user-triggered fetcher generally ignores robots.txt rules. So a disallow line controls the index, not every fetch. it
Also check your firewall. Aggressive bot protection through services like Cloudflare can label these crawlers as “unknown” and block them. Your robots.txt can look perfect while your security layer turns the bots away. Confirm both, and verify real crawler activity in your server logs.
Step 2: Write for the Question Behind the Question
Fan-out matters more than most people realise. When someone asks a broad question, the system splits it into smaller ones. A page that answers only the headline query loses to a page that also covers the follow-ups a careful reader would ask.
Here is a practical way to build that coverage:
- State the direct answer early. Put a plain two-sentence answer under each important heading, then add depth.
- Cover the neighbouring questions. Include costs, limits, exceptions, comparisons and “when this does not apply.”
- Add something that did not exist before. A dataset you collected, a test you ran, a real case with numbers, or an expert opinion with a name attached. Summaries of summaries rarely get cited, because the engine already has plenty of them.
- Name your claims’ sources. Pages with checkable facts are safer for a system to quote.
At Sarkar SEO, we apply a simple test before publishing: could a reader use this page to make a decision without opening another tab? If the answer is no, the page is not finished.
Step 3: Be Known Beyond Your Own Domain
Ranking on your site is only half the picture, because different engines weigh sources differently. One analysis of citation overlap found that Perplexity’s citations matched Google’s top ten 28.6% of the time, against roughly 8% for Gemini, Copilot and ChatGPT in-text citations. That study comes from a third-party site, so treat the exact numbers cautiously. The direction still makes sense: being cited by ChatGPT or Perplexity does not depend only on where you sit in Google’s results.
Commentary on these engines also points to freshness and community presence. Perplexity searches in real time on nearly every query, so fresh and specific content tends to win there. The takeaway is to keep key pages updated and build a real footprint in the places your audience already discusses your topic: industry publications, forums, video and professional communities. Mentions from credible third parties help every engine decide you are a recognised source.
Step 4: Measure Honestly
Clicks alone now understate your visibility. Track these instead:
- Indexed and snippet-eligible status for your top pages in Search Console.
- AI-specific reporting. One source notes that a Search Console AI search report began rolling out from June 2026 to show performance in AI Overviews and AI Mode. Check whether it has reached your account.
- Manual citation checks. Ask ChatGPT, Perplexity and Gemini the ten questions your customers ask most, and record whether you appear, who does, and what they say that you do not.
- Brand search and direct traffic, which often rise when you are being mentioned in answers even without a click.
Expect lower click-through on informational queries. Some studies claim very high zero-click rates in AI Mode, but figures vary by source, so use them as a warning rather than a forecast. The better response is to make your pages the ones worth clicking for what an AI summary cannot give: tools, original data, templates and depth.
What to Ignore
Skip anyone selling a secret AI ranking trick. Google’s guidance does not support special markup for its AI features, and rewriting every page into tiny fragments adds effort without evidence of benefit. Filling pages with AI-generated text is a bigger risk. Google’s spam policies explicitly apply to generative AI responses, including AI Overviews and AI Mode. AI tools are fine for assistance, but the published page must still meet quality rules.
A llms.txt file is another open question. Some guides recommend it as a low-cost signal, but I found no evidence that Google uses it, and Google’s guidance does not require it. Treat it as optional housekeeping, not a strategy.
A Simple Starting Plan
Begin with a technical audit: crawlability, indexation, snippet settings and bot access. Then pick your five most valuable pages and rebuild each around the full set of questions a buyer would ask. Add original evidence. Finally, test your visibility across several AI engines monthly.
This is the approach we follow at Sarkar SEO, because it holds up across algorithm changes. Teams that treat AI SEO as better SEO, with clearer answers and stronger proof, usually do better than teams chasing every new acronym.
If you want a second pair of eyes on your site’s AI readiness, the team at Sarkar SEO can run that review with you.
