Chapter 4: AI Search and Generative Engine Optimization

Learning objectives

By the end of this chapter, you will be able to:

• Explain how AI search experiences find, select, and cite sources.
• Describe how Google’s AI Overviews and AI Mode, ChatGPT search, Perplexity, Gemini, and Claude differ for marketers.
• Build content that AI systems can understand, trust, extract, and cite.
• Strengthen the off-site signals that make AI systems mention your brand.
• Manage AI crawler access deliberately.
• Measure AI visibility and AI-referred traffic.
• Run a recurring AI visibility audit with an AI assistant.

The new front door to the internet

For twenty-five years, the search results page was a list of ten blue links, and SEO was the craft of ranking on it. That page is being rebuilt. A growing share of questions are now answered by AI: in Google’s AI Overviews and AI Mode, in ChatGPT search, in Perplexity, in Gemini, in Claude, and in Microsoft Copilot. People ask longer, more conversational questions (“What’s the best coffee subscription for someone who uses a pour-over and likes fruity flavors?”) and receive a synthesized answer with a handful of cited sources.

For marketers, this creates three shifts:

1. From rankings to recommendations. The goal is no longer only to rank on page one, but to be named in the answer and cited as a source.
2. From keywords to questions and contexts. AI systems handle long, specific, multi-part questions. Content that addresses real situations and trade-offs gets used.
3. From clicks to influence. Some AI answers satisfy the user without a click. Brand mentions inside answers shape consideration even when no visit is recorded, so measurement must expand beyond organic sessions.

This chapter builds on the SEO foundations of Chapter 3. Everything there still applies; AI search raises the bar for clarity, credibility, and brand presence.

How AI search systems choose sources

The details differ by product and change frequently, but most AI search experiences follow a broadly similar process:

1. Interpret the question. The model works out what the user is really asking, including implicit needs.
2. Retrieve candidates. The system searches a web index (Google’s own index for AI Overviews and AI Mode; other products use their own crawlers and partner search indexes) for relevant pages. Google has described a “query fan-out” technique in AI Mode, which issues many related sub-queries to gather a broad set of sources for one question.
3. Select and synthesize. The model reads the retrieved passages, chooses those that are relevant, clear, and trustworthy, and writes an answer.
4. Cite. The answer links to some of the sources used.

Two kinds of knowledge shape answers:

• Retrieved knowledge: pages fetched at the moment of the question. You influence this with crawlable, well-structured, authoritative content, just as with SEO.
• Trained knowledge: what the model learned about your brand and category during training, from the broad web. You influence this slowly, through consistent brand presence, reputable mentions, reviews, and discussion across many sites over time.

Expert insight. AI systems tend to cite pages that state facts plainly and that other trusted sources agree with. A page that hedges, buries the answer, or contradicts the consensus without evidence is less likely to be quoted. Clarity and corroboration are the two currencies of AI search.

The major AI search surfaces

Surface	How it works for marketers	What to prioritize
Google AI Overviews	Generated summary above traditional results for many queries, with source links; draws on Google’s index	Classic SEO fundamentals, clear answers, structured content, E-E-A-T signals
Google AI Mode	Conversational, multi-turn search with deeper answers and follow-ups, using query fan-out	Depth across a topic cluster, answers to follow-up questions, comparison content
ChatGPT search	Answers with cited sources inside ChatGPT, using OpenAI’s crawlers and partner search data	Crawl access for OpenAI’s search crawler, clear factual pages, brand mentions on reputable sites
Perplexity	Answer engine that shows sources prominently	Citable facts, data, freshness, authoritative pages
Gemini	Google’s assistant; uses Google Search for grounding	Same foundations as Google Search, plus presence in Google’s ecosystem (Business Profile, YouTube)
Claude	Anthropic’s assistant; can search the web and cite sources when web search is enabled	Crawl access for Anthropic’s search crawler, clear and well-sourced content
Microsoft Copilot	Assistant grounded in Bing search	Bing Webmaster Tools, indexing in Bing, clear content

Because each surface relies on search and crawling, technical accessibility (Chapter 3) is the price of admission everywhere.

Figure 4.1 [SCREENSHOT TO ADD]: An AI Overview citing sources. Capture a Google AI Overview for a category question, with the cited source links visible. Add a second capture of the same question answered in ChatGPT, Gemini, or Claude with citations.

Optimizing content for AI answers

1. Answer the question in the first lines

Lead every important section with a direct, self-contained answer of one to three sentences, then elaborate. Imagine the passage being lifted out of your page and placed in an AI answer: would it make sense on its own?

Weak: “Coffee freshness is a topic many people wonder about. There are lots of factors…”

Strong: “Whole-bean coffee tastes best between about 4 and 21 days after roasting. After that, aromatic compounds fade noticeably, especially once the bag is opened.”

2. Structure for extraction

• Use question-style headings that mirror how people ask (“How long does coffee stay fresh after roasting?”).
• Use short paragraphs, numbered steps, bulleted lists, and comparison tables.
• Add a brief summary or “key takeaways” box near the top of long pages.
• Define terms explicitly (“Single-origin coffee is coffee sourced from one farm, region, or country”).
• Keep one main idea per section.

3. Cover the whole conversation

AI Mode and assistants handle follow-up questions. Build topic clusters (Chapter 3) that answer the next three questions a person would ask: comparisons, costs, alternatives, objections, and how-to steps. A page that compares options honestly, including when your product is not the best fit, is especially useful to AI systems summarizing choices.

4. Provide information gain

If your page says what a hundred other pages say, an AI system has no reason to cite you specifically. Add something new:

• Original data (“We tested 12 grind sizes and measured extraction”).
• First-hand experience with specific details, photos, and results.
• Expert commentary with named, credentialed authors.
• Proprietary frameworks, calculators, or tools.
• Up-to-date facts (prices, specifications, regulations) with dates.

5. Make facts consistent and verifiable

AI systems cross-check. Make sure your product names, prices, specifications, locations, and policies are identical across your website, Google Business Profile, marketplaces, social profiles, and directories. Cite sources for your claims, and keep “last updated” dates honest.

6. Use structured data

Schema markup (Organization, Product, Offer, Review, FAQPage, HowTo, Article, LocalBusiness) helps search engines understand entities and facts on your pages. It doesn’t guarantee citations, but it reduces ambiguity, and ambiguity is the enemy of being quoted.

7. Strengthen author and brand credibility

Author pages with real credentials, an informative About page, transparent contact details, editorial policies, and customer reviews all signal trustworthiness, which matters for both Google’s systems and the judgments AI models make about sources.

Prompt 4.1: AI-answer readiness review Act as an AI search optimization specialist. Here is the full text of a page targeting the question “[question]”: [paste]. Evaluate how likely an AI system is to quote or cite it. Score from 1 to 10 on: direct answer in the first lines, extractable structure, coverage of likely follow-up questions, information gain versus typical pages, factual specificity, and credibility signals. Then rewrite the opening 100 words for maximum clarity and list the five most valuable improvements.

Prompt 4.2: Follow-up question map A customer asks an AI assistant: “[core question]”. List the 15 most likely follow-up questions they would ask next, grouped by intent (compare, cost, how-to, objections, alternatives). For each, say whether our site currently answers it, based on this list of our URLs and titles: [paste]. Recommend which new pages or sections would make our coverage complete.

Building brand presence beyond your website

When an AI system recommends products or companies, it draws heavily on what other sources say. This is where GEO looks more like PR and community building than technical SEO.

• Earned media: coverage, reviews, and expert roundups in respected publications in your category.
• Review platforms: Google reviews, Trustpilot, G2 and Capterra for software, Amazon and retailer reviews for products, and industry-specific platforms. Quantity, recency, and the specifics in reviews all matter.
• Comparison and “best of” lists: being included, accurately, on credible third-party lists in your category.
• Communities and forums: genuine, helpful participation on Reddit, Quora, specialist forums, and Discord communities, where many AI systems find real user opinions. Never astroturf; fake participation violates platform rules and destroys trust.
• YouTube and podcasts: video and audio content (with transcripts) that explain your category.
• Knowledge sources: accurate entries on business directories, Wikidata where notable and appropriate, and industry databases.
• Partners and marketplaces: consistent listings on the platforms where your category is sold.

Prompt 4.3: Brand mention gap analysis I asked several AI assistants these questions about my category: [questions], and recorded the brands and sources they cited: [paste results]. Analyze which third-party sources appear most often, which competitors are mentioned and why they might be favored, and where my brand is missing. Recommend a 90-day plan of PR, review, community, and content actions to increase accurate mentions of my brand, prioritized by likely impact.

Managing AI crawlers

Many AI companies operate web crawlers, and some separate crawlers for different purposes, such as training models, powering search features, or fetching a page when a user asks. Publishers control access through robots.txt. Examples of crawler user agents include OpenAI’s GPTBot (training) and OAI-SearchBot (search), Anthropic’s ClaudeBot and Claude-SearchBot, PerplexityBot, and Google-Extended, which Google provides as a control for whether content is used for training its Gemini models. Notably, Google states that Google-Extended does not affect inclusion in Google Search, and AI Overviews draw on the regular Search index crawled by Googlebot.

Crawler names and behaviors change, so check each company’s current documentation. The strategic decision for most brands is straightforward:

• Most businesses benefit from allowing search-related AI crawlers, because being visible in AI answers is valuable and blocking them makes you invisible on those surfaces.
• Publishers whose content is the product (news, research, premium content) may make more nuanced choices, such as allowing search crawlers while blocking training crawlers, or pursuing licensing agreements.

You may also see proposals such as an llms.txt file that summarizes a site for AI systems. Adoption among major AI companies is limited and uncertain, so treat such files as optional experiments, not priorities.

Measuring AI search visibility

AI search measurement is immature, so combine several imperfect signals.

1. AI referral traffic in GA4

Visits from AI assistants appear as referrals from domains such as chatgpt.com, perplexity.ai, gemini.google.com, claude.ai, and copilot.microsoft.com. Create a custom channel group in GA4 with a rule that matches these sources (using a regular expression on session source) and name it “AI assistants,” placed above the Referral channel. Then you can track sessions, engagement, and key events from AI traffic over time. Visitors from AI assistants often arrive with high intent, so watch their conversion rate closely.

2. Search Console

Google includes AI Overviews and AI Mode activity in Search Console’s Performance report as part of overall search data, without a separate breakdown at the time of writing. Watch for patterns such as impressions rising while clicks fall on informational queries, which suggests answers are being satisfied on the results page.

3. Prompt tracking

Build a fixed list of 20 to 50 questions that matter to your business (category questions, comparisons, problem questions, and branded questions). Each month, run them in the main assistants and record:

• Whether your brand is mentioned.
• Your position and how you’re described (accurately? favorably?).
• Which competitors appear.
• Which sources are cited.

Use a spreadsheet, or one of the AI visibility tracking tools now offered by SEO platforms. Answers vary between runs and users, so look at trends across many questions rather than single results.

4. Brand demand

Track branded search volume, direct traffic, and “how did you hear about us?” survey responses (add “AI assistant such as ChatGPT, Gemini, or Claude” as an option). These capture influence that never shows up as a referral.

Prompt 4.4: AI visibility tracker setup Create an AI visibility tracking plan for [business] in [category]. Include: 30 tracking questions across category, comparison, problem, and branded intents; a spreadsheet structure with columns for date, assistant, question, brand mentioned (yes/no), position, sentiment, competitors mentioned, and cited sources; a GA4 regular expression for an “AI assistants” custom channel group; and a monthly reporting template that summarizes trends and actions.

Toolkit: The AI Visibility Tracker sheet gives you a ready-made monthly tracking log for questions, mentions, competitors, and cited sources.

An AI visibility audit, step by step

1. Baseline. Run your tracking questions in at least four assistants and record results.
2. Technical check. Confirm key pages are indexable, fast, and accessible to search-related AI crawlers.
3. Content check. Review your 20 most important pages with Prompt 4.1 and fix the opening answers, structure, and coverage gaps.
4. Entity consistency. Audit your brand facts across your site, Google Business Profile, marketplaces, directories, and social profiles.
5. Off-site presence. Use Prompt 4.3 to identify missing third-party sources and plan PR, reviews, and community participation.
6. Measurement. Set up the AI assistants channel group and add AI to your attribution survey.
7. Repeat monthly and report trends quarterly.

Hands-on exercises

Exercise 4.1: Four-assistant test (30 minutes). Ask the same five buying questions about a category you know in ChatGPT, Gemini, Claude, and Perplexity. Record which brands and sources each cites. How consistent are they?

Exercise 4.2: Rewrite for extraction (30 minutes). Take one of your pages (or a competitor’s) and rewrite the first 150 words and three headings using the principles in this chapter. Compare before and after with Prompt 4.1.

Exercise 4.3: GA4 channel group (20 minutes). In the GA4 demo account or your own property, draft the regular expression and rules for an “AI assistants” custom channel group.

Strategic assignment: AI search visibility plan

Deliver an AI Search Visibility Plan for your business: a baseline across at least four assistants using 30 questions; a technical and crawler access review; rewrites of five key page openings; a follow-up question map and content gap list; a brand mention gap analysis with a 90-day off-site plan; a measurement setup (custom channel group, survey option, tracking sheet); and an AI log.

Key takeaways

• AI search shifts the goal from ranking to being recommended and cited.
• Systems retrieve, select, synthesize, and cite. Clarity and corroboration win.
• Answer first, structure for extraction, cover follow-up questions, and add information gain.
• Off-site brand presence (reviews, earned media, communities, comparisons) strongly shapes AI recommendations.
• Allow search-related AI crawlers unless you have a clear reason not to.
• Measure with AI referral traffic, prompt tracking, Search Console trends, and brand demand signals.

---

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Comment *

Name *

Email *

Website

Δ