Answer engine optimization
Being the answer, not the tenth blue link.
Answer engine optimization is the work of getting a business named and cited when somebody asks an AI assistant a question, rather than when they type one into a search box. It shares most of its groundwork with SEO and it is judged by a different rule: there is no page two in an answer.
How it differs from SEO
Search rewards being on a list. Ten results appear, and a business at number seven still gets seen by somebody who scrolls. Answers do not work that way — an assistant names two or three sources and stops. The difference between third and fourth is the difference between existing and not.
The second difference is mechanical. A search crawler indexes a site on its own schedule and serves what it stored. Several assistants fetch pages at the moment the question is asked, using a separate crawler with a separate name that a site can block independently — and very often does, without the owner knowing.
The groundwork is shared. A site that is fast, crawlable, clearly written and honestly structured does well at both. There is no separate AEO stack to buy.
The crawlers that decide it
These five fetch pages to answer questions. Blocking them is not the same choice as blocking training — and the two are routinely confused in a single setting.
- GPTBotChatGPT
- OAI-SearchBotChatGPT search
- ClaudeBotClaude
- PerplexityBotPerplexity
- Google-ExtendedGemini, AI Overviews
You can check your own in about ten seconds: open yourdomain.com/robots.txt and look for a Disallow: / under any of those names.
What actually moves it
- Do this
Let the retrieval crawlers in
GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot and Google-Extended fetch pages at the moment a question is asked. A robots.txt that disallows them makes a site unquotable no matter how good it is. This is the single most common cause of zero AI visibility, and almost nobody chose it deliberately.
- Do this
Answer the question in the page, near the top
An assistant lifts a passage. A page that buries its answer under six paragraphs of preamble gives it nothing clean to lift, so it quotes whoever wrote plainly.
- Do this
Say who you are in a form machines read
Structured data stating your services, area, hours and prices removes the guesswork. An assistant that has to infer these will sometimes infer them wrong, and it will not tell you it did.
- Do this
Be citable somewhere other than your own site
Assistants lean on sources they already trust. Being mentioned where those sources look matters more than another page on your own domain.
- Do not bother
Publishing an llms.txt file
It does nothing. Google has said so publicly, no major assistant reads it, and selling one is theatre. We refuse to generate them.
- Do not bother
Mass-producing near-identical pages
A page per town from a template helps nobody who lands on it. In March 2024 Google removed sites from its results entirely for this, and an assistant will not cite thin copy either.
How anyone honestly measures it
Assistants are not deterministic. Ask the same question twice and the answer can name different sources, so a single run tells you almost nothing — a “rank” in an AI answer is not a real number.
The honest unit is a share across many runs, published with its sample size and a confidence interval, per assistant rather than blended into one figure. Anyone reporting a single AI ranking position without a sample size is reporting noise.
Every AI-visibility number Ansverra shows carries the number of runs behind it, the interval around it, and the question set it came from. Monitoring is free on every plan.
Questions people ask
- What is answer engine optimization?
- Answer engine optimization is the work of getting a business named and cited when someone asks an AI assistant a question, rather than when they type a query into a search box. It overlaps with SEO — both need a site that is crawlable, clear and credible — but it is judged differently: there is no page two in an answer, so a business is either one of the few sources cited or it is absent.
- How is Ansverra different from an AI visibility tracker?
- A visibility tracker reports how often a brand appears in AI answers. Ansverra reports the same thing free, and then changes the website so the number moves — editing titles, descriptions, structured data, internal links and crawler access directly in the customer's CMS. Each change is snapshotted before it is made and can be undone in one click.
- Why can ChatGPT not see my website?
- The most common reason is a robots.txt file that disallows the retrieval crawlers — GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot and Google-Extended. Many sites block these by default, often through a hosting or CDN setting the owner never chose. A blocked site cannot be quoted by an assistant no matter how good its content is, and nothing notifies the owner.
- Does Ansverra publish content automatically?
- Generated pages and rewrites pass a quality gate that can refuse them, and publishing prose always waits for a person. Low-risk technical and metadata changes — the reversible ones — are handled without asking, in the same way an agency would not ask before fixing a broken redirect.
- What happens to the work if I cancel?
- Every change stays. Improvements are written into the customer's own CMS as first-class content, not injected by a script that disappears when the subscription ends. What is lost on cancellation is the accumulated model of the market, the action history and the archived search data.
- Which platforms does Ansverra write to?
- WordPress through a plugin, Shopify and Webflow through their APIs, any Git-backed site through pull requests, and anything else through an open connector protocol a developer implements once. Where a platform cannot be written to safely, the product says so rather than claiming the capability.