The brand homepage is dying. Here is what replaces it.

I have spent fifteen years helping brands win SEO. I am writing this to tell you that the game has changed and most of the people selling you “AI SEO” are still selling you the old game with a new acronym.

Here is the change: buyers no longer visit your homepage to learn about you. They ask ChatGPT. They ask Perplexity. They ask the AI Overview at the top of Google. The answer they get is not a list of links to evaluate. It is a synthesis. They read the synthesis. They make a decision. Most of them never click through.

If your brand was not a citation inside the synthesis, you did not exist for that buyer. There was no second-place finish. There was no SERP position to optimize. You were absent from the conversation entirely.

This is the death of the brand homepage as a primary surface. The homepage is becoming a destination only the already-convinced visit. The buyer’s first encounter with your brand is now a sentence inside an AI answer that you did not write, sourced from signals you did not actively manage.

That is the bad news. The good news is that the signals are knowable, the tactics are buildable, and the brands that move first will compound a citation moat that competitors will not catch up on for years.

The numbers nobody is publishing in their pitch deck

Zero-click search has been climbing for five years. In 2026 it crossed 65 percent on Google in Spain and the UK. Search Engine Land’s measurement puts AI Overview presence at 25.11 percent of all queries. Perplexity’s daily query volume is doubling roughly every nine months. ChatGPT’s web search feature shipped to all users in late 2024 and is now the second-most-used research surface for B2B buyers under thirty-five.

The trendline is not subtle. The buyer is asking AI. The buyer is reading the AI’s answer. The buyer is making a decision based on whose name appears in the citation list at the bottom — and whose name does not.

Your homepage traffic is not down because your SEO got worse. It is down because the buyer never needed to visit.

Why this is not “SEO with new keywords”

The temptation, when faced with a new search surface, is to do what the SEO industry always does: keep the framework, swap the keywords, sell the same thing under a new acronym.

I have seen this pitch in 2026 from at least two-dozen agencies. It always ends the same way: the client buys “AI SEO,” gets a content calendar full of “what is X” articles, ranks fine, and then quietly notices that AI engines are still citing competitors instead of them. Six months in, the conversation gets uncomfortable.

The reason it does not work is that the underlying mechanic of selection is fundamentally different.

SEO ranks links in a list. The selection criterion is “which page is most relevant to this query.” The reward is a position. The user clicks or does not click — that is a separate question.

AI citation lifts content into an answer. The selection criterion is “is this brand a credible authority on this question, and is the content extractable enough to lift directly into the synthesis.” The reward is being in the answer. There is no second-place click-through traffic. There is no long tail of “we ranked seventh and got some clicks.” You are cited or you are absent.

The two surfaces share primitives — schema, crawl access, content quality — but they reward different attributes. SEO rewards keyword targeting and link authority. AI citation rewards entity identity, third-party trust signals, and content shape that is liftable.

Most “AI SEO” engagements I have seen are running the SEO playbook against the AI surface and being surprised when the surface does not respond.

The new attack surface — four pillars

At Citable we measure AI citation potential along four dimensions. We call this The CITE Framework — Crawl access, Identity, Trust signals, Extractability. Each one is a precondition; missing any one knocks you out of the citation pool.

Crawl is the first eligibility gate. If GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and the rest cannot reach your content, you do not exist. Most failures here are accidental — a robots.txt copied from a CMS template, a Cloudflare rule blocking unknown user-agents, a JavaScript-rendered site that the AI crawler cannot execute. If you have not audited your robots.txt against the actual list of AI crawlers in 2026, do that this week.

Identity is your entity definition. Schema.org, Wikidata Q-ID, Google Knowledge Graph, sameAs density. AI engines do not cite ambiguous entities. If your brand name collides with another brand or a generic term, the engine errs on the side of skipping you. The brand with a clean Person schema for the founder, an Organization schema with full sameAs, and a Wikidata entry with verified links to LinkedIn, Crunchbase, and GitHub wins the toss-up. Every time.

Trust is third-party authority. Wikipedia presence, citations from reputable press, .gov and .edu inbound links, listings in industry directories, profiles in analyst databases. AI engines weight third-party trust because self-claimed authority is cheap. The cost of getting cited by Wikipedia is high, and that scarcity is what makes it credible. Trust compounds slowly. Start now.

Extractability is the pillar everyone underweights. Even reachable, identified, trusted content gets skipped if it is not extractable. AI engines do not paraphrase well — they lift. A page that bundles a clean definition into a single paragraph beats a page that buries the same definition inside marketing language. FAQ schema, HowTo schema, semantic HTML, definitional sentence density, chunk size — these are what make content liftable. Most of your content is not.

What to do this week

You do not need a new agency to start. You need ninety minutes of focused work.

One. Run your domain through a free AI readiness checker. Citable’s is at /checker — six structural checks, ten seconds, no email. You will get a grade and a prioritized list. Do not pay for an audit until you have done the free check first.

Two. Open your robots.txt. Confirm GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and Bytespider are not blocked. If they are, fix it. If you have a Cloudflare WAF rule blocking unknown bots, audit the user-agent allow-list.

Three. Search for your brand in ChatGPT, Perplexity, Gemini, and Google AI Overviews. Use the prompts your buyers actually use — category questions, comparison questions, decision questions. Note which competitors are cited. Note whether you are. Note what the AI says about you when it does mention you. This is your baseline. It is more useful than any SEO ranking report you have ever read.

If after those three things you decide you need help, the paid AI Visibility Audit at Citable runs 50 prompts across four engines, scores you against all four CITE pillars, and gives you a 90-day implementation roadmap. It costs less than one month of a junior marketing hire and the report is yours to keep whether or not we work together afterwards.

The prediction

By 2028, “AI-citable” will be a standard line item in every B2B brand’s digital marketing strategy document. The brands that started in 2026 will have a two-year compounding lead on Wikipedia presence, on Wikidata sameAs density, on schema deployment, on extractable content. Those signals do not fast-follow well — Wikipedia editors do not approve articles for brands that did not exist in the canon two years ago.

The brands that wait until 2027 will spend the rest of the decade catching up. The brands that wait until 2028 will spend the rest of the decade explaining to their boards why they are absent from AI answers.

The brand homepage is dying. What replaces it is a citation footprint that lives across Wikipedia, Wikidata, the open knowledge graph, ChatGPT’s training corpus, Perplexity’s retrieval index, and every reputable third-party authority that AI engines weight.

Build that footprint now. Or wait, and explain later.

— Elizabeth S., founder, Citable