Marketing, decoded · Updated weekly
The ideas behind the marketing that actually moves markets.

Analyst relations, positioning, and AI — written plainly for the people who have to make the decisions.

Find your reading path → Subscribe free
Analyst Relations Marketing Strategy AI & Technology Digital Transformation B2B Marketing Thought Leadership

Start here.

What brings you in today? Pick a path and the essays below will filter to match.

Browse by topic

Stay in the loop.

Insights on analyst relations, marketing strategy, and technology — delivered when it matters.

Built to be read — by people and machines.

01 — Skim-ready

Every essay opens with a TL;DR and key takeaways, so a reader — or an assistant — gets the point in five seconds.

02 — Structured

Semantic HTML and JSON-LD describe each piece, so search engines and LLMs cite the source instead of guessing at it.

03 — Navigable

Filter by topic or pick a reading path built around your role — no scrolling a wall of posts hoping something fits.

Browse by topic

More Posts

Google Finally Wrote Down Its Rules for AI Search. Here Is What They Actually Say.


Google published an official guide to optimizing for AI Overviews and AI Mode. It is the closest thing to a primary source this discipline has. Marketing leaders should read it before buying another generative engine optimization pitch.

For two years, advice on AI search visibility has come from everywhere except the companies running the AI search systems. Consultants filled the vacuum. Some of the advice was sound. Much of it was guesswork sold at retainer prices.

That changed when Google Search Central published "Optimizing Your Website for Generative AI Features on Google Search." It is official documentation, updated June 29, 2026, and it covers AI Overviews, AI Mode, and the agent experiences Google expects to follow. This primer walks through what the guide says, what it dismisses, and where a marketing leader should stay skeptical.

Google's AI Answers Run on the Same Search Index You Already Optimize For

The most useful part of the guide is the plumbing explanation. Google confirms that its generative artificial intelligence (AI) features sit on top of the core Search ranking and quality systems, not beside them. Two mechanisms matter.

The first is retrieval-augmented generation (RAG), which Google also calls grounding. When someone asks a question in AI Mode or triggers an AI Overview, the system uses Google's existing ranking systems to retrieve relevant pages from the search index, then generates a response from what those pages say, with links back to the sources. The AI does not answer from memory alone. It answers from the index.

The second is query fan-out. The model takes the original question and spins off a set of related queries to gather more material. Google's own example: a search for "how to fix a lawn that's full of weeds" might fan out into queries about herbicides, chemical-free removal, and prevention. Your page can surface for a fan-out query the user never typed.

The practical implication is direct. If your content cannot rank in conventional Google Search, it cannot be retrieved for an AI answer. Visibility in AI Overviews is downstream of indexation and ranking, not a separate channel with separate rules.

Google Says GEO Is Just SEO. That Is True for Google and Only for Google.

The guide addresses the terminology fight head on. Answer engine optimization (AEO) and generative engine optimization (GEO) are, in Google's framing, the same work as search engine optimization (SEO) because the AI features are part of the search experience. Google also points readers to its guidance on evaluating third-party SEO advice, a polite way of telling site owners to pressure-test vendor claims.

Take that position seriously, but notice its boundary. This document describes how Google's systems work. It says nothing about how ChatGPT, Perplexity, Copilot, or Claude retrieve and cite web content. Those platforms run their own crawlers, their own retrieval systems, and their own citation behaviors. A tactic Google ignores may still matter somewhere else. The guide is authoritative for the largest AI search surface on earth, and silent on the rest.

I covered this fragmentation problem in my Info-Tech Research Group blueprint, Stay Relevant in the Era of AI-Powered Search. Each answer engine is a distinct distribution channel with distinct mechanics. Google just documented its channel. The others have not been this explicit.

The Content Bar Is "Non-Commodity." Google Even Defines It.

The guide's content advice centers on a term worth adopting internally: non-commodity content. Google's example of commodity content is a piece like "7 Tips for First-Time Homebuyers," material based on common knowledge that anyone could produce. Its example of non-commodity content is "Why We Waived the Inspection & Saved Money: A Look Inside the Sewer Line." Specific. Experiential. Not reproducible by a language model summarizing the internet.

Google states the reasoning plainly: its AI systems look across many sources, so a unique viewpoint stands out. Content that recycles what others have said, or that a generative model could produce on its own, adds nothing to the retrieval pool and gets treated accordingly.

The guide also warns against manufacturing pages for every conceivable query variation, including fan-out queries. Doing that at scale to manipulate rankings violates Google's scaled content abuse spam policy. This is the death notice for the "publish 400 programmatic pages targeting AI queries" pitch some agencies are still selling.

Supporting guidance is familiar but worth restating for your content team. Organize pages with clear headings and sections written for human readers. Add high-quality images and video where they genuinely support the text, because AI features pull visual assets too. If you use generative AI tools in production, the output still has to meet the Search Essentials and spam policies. The tool is not the problem. Commodity output is.

Technical Requirements Are Table Stakes, Not Differentiators

To be eligible for generative AI features, a page must be indexed and eligible to appear in Google Search with a snippet. That is the whole technical entry requirement. The rest of the technical section is standard hygiene: keep content crawlable, follow JavaScript SEO practices if your site depends on client-side rendering, provide good page experience, reduce duplicate content, and verify the site in Search Console so you catch problems early.

One nuance deserves attention. Google says semantic HTML is not required for its own systems but recommends it anyway because it helps other consumers of your pages, including screen readers and the browser agents discussed below. Clean markup is becoming an accessibility investment and an agent-readiness investment at the same time.

For commerce and local businesses, the guide points to Merchant Center feeds and Google Business Profiles as the pipes that feed product listings and business details into AI responses. It also mentions Business Agent, a conversational experience that lets customers chat with a brand inside Google Search. If you sell products or drive foot traffic, your structured feeds now do double duty.

Five Things Google Tells You to Stop Paying For

The mythbusting section is where the guide earns its keep. Google names specific tactics that do nothing for visibility in its AI features.

LLMS.txt files and special AI markup. Google Search does not use them. Creating an llms.txt file will neither help nor hurt your Google visibility. If another platform in your mix reads these files, fine, maintain one for that reason. Just do not attribute any Google result to it.

Chunking content into fragments. There is no requirement to break pages into small pieces for AI comprehension. Google's systems can find the relevant passage inside a longer page. Page length should serve the reader, not a theory about retrieval windows.

Rewriting copy in a robotic "AI-friendly" register. The systems understand synonyms and intent. You do not need to enumerate every long-tail phrasing of a question. Write the way your audience reads.

Buying mentions. AI features can surface what blogs, forums, and videos say about your brand, which has spawned a cottage industry of inauthentic mention placement. Google says its ranking systems reward quality and its spam systems block manipulation, and the AI features depend on both. Earned discussion counts. Manufactured discussion is spam with extra steps.

Overinvesting in structured data for AI reasons. No special schema.org markup unlocks AI Overviews. Structured data remains worthwhile for rich results eligibility, so keep it in the program, but stop funding schema projects justified purely by AI visibility claims.

The Agent Section Is the Early Warning

The guide closes with a short section on agentic experiences that reads like a preview of the next three years. AI agents complete tasks for people, such as booking reservations or comparing product specifications. Browser agents do this by rendering your site, inspecting the document object model (DOM), and reading the accessibility tree. Google points to agent-friendly website practices published on web.dev and flags the emerging Universal Commerce Protocol (UCP), which will let Search agents transact.

Google labels this optional, something to explore "if you have extra time." Treat that framing the way experienced operators treated mobile-friendliness guidance in 2012. Optional is a phase.

Where the Guide Deserves Pushback

Two cautions before you circulate this internally as settled law.

First, the guide answers "how do I appear in AI features" and never touches "what happens to my traffic when I do." Appearing as a grounding source in an AI Overview is not the same as earning the click. The guide's opening frames AI search as an opportunity to reach people more inclined to engage and convert. That is Google's characterization of its own product, and your analytics should be the judge of it, not this document.

Second, Google is a party with interests, not a neutral referee. The advice to ignore llms.txt is accurate for Google and convenient for Google. Keeping SEO and AI optimization defined as one discipline keeps the discipline oriented around Google's index. That is a reasonable position from the market leader. It is not a reason to skip measuring how your brand shows up in the other answer engines your buyers actually use.

What to Do Monday

Kill the hacks line items. Review active SEO and content vendor engagements for llms.txt deployments, content chunking projects, AI-specific rewrites, mention-buying, and schema work justified only by AI visibility. Google has now stated in writing that none of it moves the needle in its systems. Reallocate that budget.

Run a commodity audit. Pull your 25 highest-traffic pages and sort them into commodity and non-commodity using Google's own definitions. Anything a competitor or a language model could have written gets a rework plan built on first-hand experience, proprietary data, or a named expert's viewpoint.

Confirm the entry requirements. Have your team verify in Search Console that priority pages are indexed and snippet-eligible. This is the entire technical gate for AI feature eligibility, and it takes an afternoon to check.

Assign the agent question. Give one owner 30 days to assess your site against the agent-friendly practices Google references, starting with semantic markup and accessibility tree quality. If you sell online, add UCP to your commerce platform roadmap conversation.

Measure beyond Google. Google documented its channel. Your buyers also ask ChatGPT, Perplexity, and Copilot. Track how your brand is retrieved and cited across each, and hold each channel to its own playbook rather than assuming Google's rules travel.

Works Cited

Bellamkonda, Shashi. Stay Relevant in the Era of AI-Powered Search. Info-Tech Research Group, www.infotech.com/research/ss/stay-relevant-in-the-era-of-ai-powered-search.

"Creating Helpful, Reliable, People-First Content." Google Search Central, Google, developers.google.com/search/docs/fundamentals/creating-helpful-content.

"Optimizing Your Website for Generative AI Features on Google Search." Google Search Central, Google, 29 June 2026, developers.google.com/search/docs/fundamentals/ai-optimization-guide.

"Spam Policies for Google Web Search." Google Search Central, Google, developers.google.com/search/docs/essentials/spam-policies.

Shashi Bellamkonda

Marketing and analyst relations practitioner. Writing about the ideas behind the marketing that actually moves markets in technology. Views are my own.