AEO & SEO

SEO earns a ranking; AEO earns a citation. SEO optimizes a page to appear in a ranked list of links a person then clicks, while AEO optimizes content so a language model quotes and attributes it inside the answer it generates. The tactics overlap (both reward crawlable, well-structured, genuinely useful pages) but the measurement is completely different: SEO is measured in position and clicks, AEO in share of citation across engines. The practical reason to run both is that page-one rankings increasingly produce no clicks, because an AI Overview and paid results push the first organic listing well below the fold.

You get cited by AI search by making a specific, verifiable claim easy for a model to extract and safe to attribute. In practice that means answering the exact question in the first two sentences of a section, marking it up with FAQPage or Article schema, keeping the facts current and sourced, allowing GPTBot, PerplexityBot, and ClaudeBot in robots.txt, and being corroborated somewhere the engine already trusts. Hedged, padded, keyword-stuffed copy gets skipped because there is nothing quotable in it. The engines reward the same thing a careful reader does: a direct answer with something concrete behind it.

You measure AI search visibility by polling a fixed set of buyer questions across each answer engine on a schedule, parsing which domains get cited, and tracking your share of those citations over time. A ranking tool cannot do this: AI answers are generated per query, vary between engines, and often cite sources that do not rank on page one at all. RevAlign runs exactly this loop, a weekly scoreboard over a few hundred buyer prompts across ChatGPT, Gemini, Perplexity, and Exa, and the questions where a competitor is cited and you are not become the work queue for the following week.

Answer engine optimization is generally sold either as a measurement subscription (a dashboard that reports where you are cited, commonly a few hundred to a few thousand dollars a month) or as a done-for-you engagement where someone also does the work the measurement implies. The distinction matters more than the price: most of the category sells tracking, and tracking alone does not change a single citation. RevAlign runs it as one closed loop, measurement plus the page and schema work the loss list calls for, priced as a monthly engagement rather than per-page.

Hire on whether the engagement closes the loop, not on which acronym is on the invoice. A traditional SEO agency optimizes for rankings and will report position and impressions, which is a problem when page-one rankings produce no clicks because an AI Overview sits above them. An AEO-focused engagement should tell you which buyer questions you are cited on today, which ones a competitor owns, and what specifically changed after the work shipped. If a prospective agency cannot show you that before-and-after, it is selling SEO deliverables with newer vocabulary.

AEO is worth it for an early-stage B2B startup when its buyers research vendors through AI tools and the category is not already dominated by a few entrenched sources. The advantage early-stage companies have is that AI citation is far less winner-take-all than search rankings: engines cite specific, well-structured answers from small domains regularly, because they are optimizing for a quotable claim rather than domain authority. It is a poor fit if nobody is asking an AI about your category yet, which is exactly what a prompt-set baseline tells you before you spend anything.

Expect a measurable baseline in the first week and the first citation movement in roughly four to eight weeks. The baseline is fast because it is measurement: poll the prompt set, see where you stand. Movement takes longer because it depends on engines recrawling changed pages and on your answers being corroborated elsewhere, and it is uneven, some engines pick up a new answer within days while others lag by a month or more. Anyone promising citations in week one is describing a paid placement, not optimization.

llms.txt is a proposed convention, not a standard any major answer engine has committed to honoring, so treat it as cheap insurance rather than a lever. It costs an hour to publish a clean summary of what your company does and which pages matter, and it does no harm. What demonstrably does matter today is more boring: allowing AI crawlers in robots.txt, valid structured data, fast and server-rendered pages, and answers written to be extracted. If a vendor sells llms.txt as the centerpiece of an AEO program, that is a signal about the rest of the program.

If you want to be cited in AI answers, yes: blocking GPTBot, PerplexityBot, ClaudeBot, or Google-Extended in robots.txt removes you from the pool of sources those engines can quote and attribute. The tradeoff people worry about is training data, and it is a real decision for publishers whose product is the content itself. For a B2B company whose content exists to get discovered by buyers, the calculation is one-sided: being unquotable is not protection, it is invisibility. Check your robots.txt, since many site templates block these agents by default.