Skip to content

AI Search Optimization: SEO for AI Overviews & Answer Engines · How AI search and answer engines work · lesson 2 of 17 · 13 min

The AI search landscape in 2026

A fast-moving market

AI search changes month to month. This lesson describes the landscape as of 2026 at a level that should age reasonably well; always verify specifics against each provider's current documentation.

The main surfaces

| Surface | What it is | Where sources come from (publicly stated or widely understood) | |---|---|---| | Google AI Overviews | AI-generated summaries above or within Google results for some queries | Google's search index and ranking systems; supporting links to web pages | | Google AI Mode | A conversational, AI-first search experience | Google's index, using query fan-out; supporting links | | ChatGPT search | ChatGPT answering with web results and citations | OpenAI's own search crawler (OAI-SearchBot) and third-party search partners | | Perplexity | Answer engine with prominent citations | Its own crawler/index plus other sources | | Microsoft Copilot | Assistant integrated with Bing and Windows/Edge | Bing's index | | Gemini app | Google's assistant | Google Search grounding and other Google systems | | Claude | Anthropic's assistant with web search | Web search and Anthropic's crawlers |

Several assistants also fetch pages on demand when a user shares a URL or asks the assistant to browse, using user-initiated agents distinct from their crawlers.

Google's position on optimising for AI features

Google's documentation for AI features in Search says, in essence, that there are no additional requirements and no special optimisation needed to appear in AI Overviews or AI Mode beyond existing SEO fundamentals: pages must be indexed and eligible to be shown with a snippet, and the usual advice on helpful, reliable, people-first content, technical requirements and good page experience applies. Google also notes that structured data is not required for AI features, though it remains useful generally. Sites can use existing controls — nosnippet, data-nosnippet, max-snippet and noindex — to limit how content appears.

Clicks from AI Overviews and AI Mode are counted in Search Console's Performance report within overall web search data. Since June 2026 Search Console also has a Search generative AI performance report, rolled out to all sites by the start of September 2026, showing impressions in AI Overviews, AI Mode and AI Overviews in Discover by page, country, device and date — but not queries, clicks or CTR. Google also added a property-level Search generative AI control in Search Console settings that lets a site opt out of those AI surfaces while staying in regular results (Lesson 1.3 covers both).

What changes for marketers

  1. More zero-click answers: for simple informational questions, users may get what they need without clicking. Some publishers report traffic declines on informational queries; others report that clicks from AI features are fewer but more engaged. Evidence is mixed and varies by sector — measure your own.
  2. Brand exposure without clicks: being cited or mentioned in answers builds awareness even when users don't click. This is harder to measure.
  3. Longer, more specific queries: conversational interfaces encourage detailed prompts ("I run a 12-person agency in Riyadh..."), which fan out into specific sub-questions.
  4. Comparison and recommendation queries become very important: "best X for Y", "X vs Y", "alternatives to X". Third-party comparisons and reviews heavily influence these answers.
  5. Bing matters more than its search share suggests, because its index supports Copilot and has been used by other assistants. Since February 2026 Bing Webmaster Tools has an AI Performance report showing citations of your pages in Copilot and Bing's AI answers, plus the grounding queries that retrieved them.

What stays the same

  • Being crawlable and indexable.
  • Creating content that genuinely helps people and demonstrates experience and expertise.
  • Earning a trustworthy reputation through independent sources.
  • Technical quality: fast, accessible, server-rendered content.

A practical priority list for 2026

  1. Ensure key pages are indexed in Google and Bing (verify both Search Console and Bing Webmaster Tools).
  2. Make critical content available in initial HTML (many AI crawlers don't render JavaScript).
  3. Decide your AI crawler policy consciously (Module 4) rather than blocking or allowing by accident — including CDN default settings.
  4. Strengthen content for comparison, pricing, "how to choose" and specific-use-case queries.
  5. Build third-party corroboration: reviews, expert coverage, credible comparisons.
  6. Set up measurement: AI referral traffic, sampled visibility, brand mentions.

Regional notes

AI features roll out unevenly by country and language. Arabic and Urdu support, feature availability and source pools can differ from English-language US results. Test in your markets and languages rather than relying on US-centric reports.

How each surface finds sources (and what you can check)

| Surface | Crawler / index to allow | First-party data available to you | |---|---|---| | Google AI Overviews / AI Mode | Googlebot (Search index) | Search Console: Performance (clicks within web totals) and the generative AI impressions report | | Microsoft Copilot / Bing AI answers | Bingbot | Bing Webmaster Tools: AI Performance (citations, grounding queries) | | ChatGPT search | OAI-SearchBot (search), ChatGPT-User (user fetches) | Referrals tagged utm_source=chatgpt.com in analytics | | Perplexity | PerplexityBot, Perplexity-User | Referrals from perplexity.ai | | Claude with web search | Claude-SearchBot, Claude-User | Referrals from claude.ai | | Gemini app | Google systems incl. Search grounding | Referrals from gemini.google.com |

Worked example: a Riyadh clinic group picks its battles

A dental clinic group in Riyadh (illustrative) has one SEO specialist and a small budget. Instead of chasing every assistant, she checks where patients actually come from: GA4 shows AI referrals mostly from ChatGPT and Gemini; Search Console shows the clinic's service pages earning AI Overview impressions in Arabic; Bing Webmaster Tools shows a handful of Copilot citations for an English "cost of dental implants in Riyadh" guide. Her priorities follow the evidence: Arabic service pages first (Google), the implant cost guide second (Copilot and ChatGPT), and she confirms OAI-SearchBot and Bingbot are not blocked by the clinic's CDN.

Hands-on: a 30-minute landscape check for any site

  1. Google: Search Console → Performance → last 3 months; then open the generative AI report and note which pages get AI impressions.
  2. Bing: verify the site in Bing Webmaster Tools (import from Search Console is fastest) and open AI Performance.
  3. Analytics: in GA4, build an exploration with session source matching the AI regex from Lesson 5.1.
  4. Access: curl your top 5 pages with JavaScript off and confirm key facts are in the HTML.
  5. Policy: open robots.txt and your CDN's bot settings; list which AI bots are allowed.

Write the answers into a one-page "AI surface map" for the business — it becomes the baseline for everything else in this course.

Common mistakes

  • Treating one provider's behaviour as universal.
  • Ignoring Bing Webmaster Tools.
  • Assuming AI Overviews need special markup or a separate content strategy from SEO.
  • Believing traffic trends from other industries predict yours.

Video lecture: The AI search landscape in 2026

Lecture coming soon · 13 chapters · about 9 minutes. Read the full transcript below.

  1. The AI search landscape in 2026
  2. Google: AI Overviews and AI Mode
  3. New in Search Console (2026)
  4. Microsoft: Copilot and Bing
  5. Independent answer engines
  6. What changes, what stays
  7. Example 1: a freelance photographer
  8. Example 2: Riyadh clinic group (illustrative)
  9. Common mistakes
  10. Recap and try this now
  11. Watch me do it: the 30-minute check
  12. Choosing priorities
  13. Test where you sell

Lecture transcript

The AI search landscape in 2026

Here's a question I hear from almost every business owner right now. Which AI should we optimise for? ChatGPT? Google's AI Mode? Copilot? Perplexity? Claude? The honest answer is: the ones your customers use, and the ones you can measure. In this lecture you'll get a clear map of the AI search landscape as it stands in twenty twenty-six, where each surface gets its sources, what data each one gives you, and how to pick your priorities without chasing every shiny new launch.

Google: AI Overviews and AI Mode

Let's start with Google, because it's still the biggest door. There are two AI surfaces. AI Overviews are summaries that appear above or within normal results for some queries. AI Mode is a conversational, AI-first search experience that uses query fan-out. Both draw on Google's search index and ranking systems. That means the crawler you need is plain old Googlebot. There is no separate AI crawler for these features. If you're indexed and eligible for a snippet, you're eligible to be a supporting link.

New in Search Console (2026)

And here's what changed in twenty twenty-six. Search Console now has a generative AI performance report. It shows impressions for your pages in AI Overviews, AI Mode, and AI Overviews in Discover, broken down by page, country, device and date. It does not show queries or clicks. Clicks from those features are still counted inside your normal web search totals. Google also added a Search generative AI control in Search Console settings, which lets a site opt out of those AI surfaces while staying in regular results. We'll look at that trade-off carefully later in this module.

Microsoft: Copilot and Bing

Next, Microsoft. Copilot and Bing's AI answers run on the Bing index, so Bingbot is your crawler. And Bing has given site owners something genuinely useful. Since February twenty twenty-six, Bing Webmaster Tools has an AI Performance report. It shows how often your pages were cited in Copilot and Bing AI answers, which pages, and the grounding queries Copilot generated to find them. Bing's index has also been used by other assistants over the years, which is why Bing matters more than its search share suggests.

Independent answer engines

Then the independent answer engines. ChatGPT search uses OpenAI's OAI-SearchBot for its search index, plus ChatGPT-User when someone asks it to open a page. Clicks usually arrive tagged with utm source equals chatgpt dot com. Perplexity uses PerplexityBot and a user-initiated fetcher. Claude's web search uses Claude-SearchBot, and Claude-User for fetches a user asks for. The Gemini app draws on Google systems including Search grounding. None of these gives you a Search Console-style report, so analytics referrals and sampling are your main evidence.

What changes, what stays

So what changes for marketers? Four things. More zero-click answers on simple questions, so some informational traffic shrinks. Brand exposure without a click, which is real but harder to measure. Longer, more specific prompts, like I run a twelve-person agency in Riyadh, which fan out into precise sub-questions. And comparison queries, best X for Y, X versus Y, alternatives to X, become hugely important, because third-party reviews shape those answers. What stays the same? Crawlability, genuinely helpful content, reputation, and technical quality.

Example 1: a freelance photographer

Worked example one, a simple one. A freelance photographer in Birmingham asks, should I worry about all of these? We check three things in half an hour. Search Console shows her portfolio pages earn a few AI Overview impressions. Analytics shows a trickle of referrals from ChatGPT. Bing Webmaster Tools isn't set up at all. Her priority list writes itself: verify Bing by importing from Search Console, which takes minutes, and make sure her pricing and service areas are in plain HTML text, not just inside images.

Example 2: Riyadh clinic group (illustrative)

Worked example two, more realistic, with illustrative details. A dental clinic group in Riyadh has one SEO specialist. GA4 shows AI referrals mostly from ChatGPT and Gemini. Search Console shows Arabic service pages earning AI Overview impressions. Bing shows Copilot citing an English guide to implant costs in Riyadh. So the plan follows the evidence. Arabic service pages first, for Google. The implant cost guide second, for Copilot and ChatGPT. And a quick check that the clinic's CDN isn't blocking OAI-SearchBot or Bingbot. Notice what she didn't do: chase every assistant equally.

Common mistakes

Common mistakes. Treating one provider's behaviour as universal. Ignoring Bing Webmaster Tools, which is free and now shows AI citations. Assuming AI Overviews need special markup. They don't. And believing traffic trends from another industry predict yours. Also remember regional rollouts. Features, languages and source pools differ between the US, the UK, the Gulf and Pakistan. Test in your own markets and languages, in Arabic or Urdu where that's how customers search.

Recap and try this now

Recap. Google's AI features run on Googlebot and the Search index, and Search Console now reports AI impressions. Copilot runs on Bing, and Bing reports citations and grounding queries. ChatGPT, Perplexity and Claude run their own search crawlers and user fetchers, measured through referrals. Try this now. Spend thirty minutes on the landscape check in the lesson text: Search Console, Bing, analytics, a raw HTML test, and your robots and CDN settings. Write the results on one page. That's your AI surface map, and your baseline.

Watch me do it: the 30-minute check

Watch me do it. Here's my thirty-minute landscape check on a real site, in order. Minute zero: Search Console. I open Performance for the last three months and export the top pages. Then I open the generative AI report and export pages with AI impressions. I put both lists side by side and mark overlaps. Minute eight: Bing Webmaster Tools. If the site isn't there, I use import from Search Console, which takes a couple of minutes. I check Sitemaps for errors, then AI Performance for cited pages and grounding queries. I copy the top ten grounding queries into my notes. Minute fifteen: GA4. I build a quick exploration with session source matching the assistant regex and look at sessions and key events by source. Minute twenty-two: access. I run curl on the top five pages with JavaScript off and search the output for the price, the city and one key feature. Minute twenty-six: policy. I open robots.txt and the CDN's bot settings, and list which AI bots are allowed. Minute thirty: I write one page. Where AI demand shows up, which surfaces cite us, what's blocked, and the first three actions. That page is the baseline for everything else in this course.

Choosing priorities

So how do you choose priorities without chasing every launch? Use three questions. One: where is there evidence of demand? That means referrals in analytics, AI impressions in Search Console, citations in Bing. Two: which surfaces can you influence cheaply? Being indexed in Bing, allowing the right crawlers, and putting facts in plain HTML help almost everywhere at once. Three: which markets and languages matter? A Karachi retailer selling in Urdu and English has different priorities from a London SaaS firm. Score each surface on those three questions, and you'll usually find that two or three fixes cover most of the landscape.

Test where you sell

There's also a regional point worth pausing on. AI features roll out unevenly by country and language. What a US marketer sees in AI Mode may not be what a customer in Jeddah or Islamabad sees. Source pools differ too. An Arabic-language answer needs Arabic sources, and if your Arabic pages are thin machine translations, you're unlikely to be one of them. So always test in the market and language you actually sell in, and note the location for every test you run.

Video transcript

Let's map the AI search landscape as it stands in 2026, and what it means for your work. On Google, there are AI Overviews, the summaries that appear above many results, and AI Mode, a conversational search experience. Both draw on Google's own index and ranking systems. And here's the important part: Google says there's no special optimisation for them. Your pages need to be indexed and eligible to show a snippet. The usual fundamentals apply. Outside Google, there's ChatGPT search, Perplexity, Microsoft Copilot, which runs on Bing, Gemini, and Claude. Each retrieves web content in its own way, with its own crawlers. So Bing matters more than its market share suggests, and your crawler settings matter too. What changes for marketers? Simple questions increasingly get answered without a click. Comparison queries, like "best tool for a small agency" or "X versus Y", become critical, and they're heavily shaped by third-party reviews and comparisons. Prompts get longer and more specific. What stays the same? Being crawlable and indexed. Content that genuinely helps people. A reputation that independent sources confirm. And technically sound pages, ideally with content in the initial HTML, because many AI crawlers don't run JavaScript. One last thing: features roll out unevenly by country and language. Test in your own markets, in English, Arabic or Urdu, instead of relying on someone else's screenshots.

Key takeaways

  • Major surfaces include Google AI Overviews and AI Mode, ChatGPT search, Perplexity, Copilot, Gemini and Claude.
  • Google says no special optimisation is needed for its AI features beyond SEO fundamentals; indexing and snippet eligibility are required.
  • AI Overviews/AI Mode traffic sits within Search Console web data and isn't separated at the time of writing.
  • Comparison and recommendation queries, Bing indexing and third-party corroboration matter more in AI search.

Try it

Verify your site in Bing Webmaster Tools if you haven't, and compare indexed page counts for key sections with Google Search Console.