Search stopped being one thing. A buyer asks ChatGPT the open question, checks Google's answer box while comparing, and sends Perplexity after the details. Three engines, three sets of rules, one brand that has to show up in all of them. Most organic programs are still optimizing for the middle one.
Why search visibility is now three problems, not one
Most organic programs were built for a world with one destination: a ranked
list, a click, a session, an attributed conversion. That world still exists, but
it is now one lane of three, and it is the lane losing share.
The other two behave differently. An answer box takes your content and returns
it to the buyer without the visit. A generative engine reads your site, decides
whether you are credible enough to name, and either cites you or assembles the
answer from a competitor. In both cases the traffic report understates what
happened — which is why so many teams believe organic is declining at exactly
the moment their content is being read more than ever.
The work is to stop optimizing for the session and start optimizing for the
answer.
What the foundation actually requires
Everything downstream depends on a few unglamorous conditions being true.
- The engines can reach you. AI crawlers are frequently blocked in
robots.txt by a decision no one remembers making. You cannot be cited by an
engine you have locked out.
- The engines can read you. Most crawlers do not execute JavaScript. If your
page is a client-rendered shell, the content you are proud of does not exist
to them.
- The engines can identify you. Structured data — Organization, Person,
Service, FAQPage, and the
sameAs links that connect you to your profiles —
is how a model resolves you as an entity rather than a string of words.
- The engines have something to lift. Answer-first content means real buyer
questions as headings with direct, quotable answers underneath. Not a preamble
and a brand story before the substance.
None of this is exotic. It is just rarely anyone’s job, so it does not get done.
The per-engine play
Once the foundation holds, the lanes separate.
For ChatGPT, the lever is corroboration. It introduces your brand into the
conversation, and it does that on the strength of what third parties say about
you. That makes it a digital PR and citation problem more than an on-site one.
For Gemini and AI Overviews, the lever is entity authority. Buyers are
comparing here, inside the answer, and the engine leans on structured facts and
brand mentions to decide who belongs in the consideration set.
For Perplexity, the lever is depth in public. It rewards dense, citable,
specific material and will happily route around anything gated. If your best
thinking is behind a form, you have opted out of the engine your most serious
researchers use.
How I work this lane
I build it as one system with three outputs rather than three campaigns. The
technical and structural work is shared. The content strategy is written to
serve both a human reader and a model assembling an answer. The measurement runs
per engine, on a schedule, so the program is judged on share of answer and
citation growth alongside the traditional organic numbers.
And I do the work, not just the plan. The audit, the schema, the content
architecture, the query set — those are things I build, which is the only way to
know whether the strategy survives contact with the site.
Frequently asked questions
What is the difference between SEO, AEO, and GEO?
SEO earns a ranked position on a results page a person then clicks. AEO structures your content so an engine can lift a direct answer out of it, into a featured snippet or answer box, whether or not a click follows. GEO is about being cited inside a generated answer from a model like ChatGPT, Gemini, Perplexity, or Claude. They share a technical foundation and then diverge sharply. Treating them as one workstream is the most common mistake I see.
Do I need a different strategy for ChatGPT, Gemini, and Perplexity?
Yes, because they sit in different places in the buyer journey. ChatGPT tends to act as the awareness layer, rewarding third-party corroboration and citation — your brand gets introduced there, and you should think in impressions rather than clicks. Gemini and AI Overviews sit in consideration, rewarding entity authority and brand mentions while buyers compare and shortlist. Perplexity is the researcher’s engine, dense with visible citations, where gated expertise effectively does not exist. One universal “AI SEO playbook” underperforms in all three.
Can you build AI-search visibility from nothing?
That is usually the starting condition. Most brands have no GEO function, no crawler policy written with AI engines in mind, and no measurement of whether they get cited. The early work is unglamorous and fast: confirm the AI crawlers are not blocked in robots.txt, get real server-rendered text on the page, add the structured data that tells engines what you are, and build genuinely answer-first content. Then the slower work starts, which is earning the off-site corroboration that citations depend on.
How is AI-search visibility measured if there are no clicks?
By auditing the engines directly. You build a query set that mirrors how buyers actually ask about your category, run it per engine on a schedule, and record whether you appear, how you are characterized, and who gets cited instead of you. That gives you share of answer, sentiment, and a competitive gap — none of which show up in a traditional rank tracker. Paired with branded search lift and direct traffic, it is a real measurement program rather than a vibe.
Does traditional SEO still matter if AI answers are taking the clicks?
It matters more, not less, because it is the substrate. Engines assemble answers from content they can crawl, parse, and trust. A site that is technically sound, well structured, and cited by others is the same site that gets pulled into generated answers. What changes is the scorecard: optimizing purely for sessions will make an AI-visible brand look like it is losing while its influence grows.
How long before this shows results?
The access and structure fixes — crawler permissions, server rendering, structured data, llms.txt — can land in weeks and sometimes change citation behavior quickly, because they remove a hard block rather than compete for a position. Content and off-site corroboration follow a normal organic curve, typically a couple of quarters before the trend is unambiguous. Anyone promising faster than that on the authority side is selling something.