Agency Pilot — $499

Answer engine optimization

Answer Engine Optimization (AEO): Visibility in AI Answers, Without the Myths

AEO is the practice of improving how well your brand and content can be discovered, understood, used and cited by systems that generate direct answers. The term is useful. Several of the myths attached to it are contradicted by Google’s own documentation.

Updated & reviewed Google Search Central primary documentation, quoted directly

What AEO covers

AEO focuses on systems that respond with an answer rather than only a list of links.

Depending on context that includes Google AI Overviews and AI Mode, ChatGPT-style search, Gemini, Perplexity, Copilot and other assistant interfaces. In practice the work involves answering questions clearly, creating original evidence, improving entity and product clarity, keeping important pages crawlable and indexable, earning credible third-party references, and monitoring whether the brand is actually mentioned, cited or recommended.

TermThe question it asks
SEOHow do we become discoverable and competitive in search?
AEOWhen an answer system synthesises information, does it use, cite or recommend us?
GEOThe same job, framed around generative engines specifically. See generative engine optimization.

For Google the overlap is explicit. Its AI features and your website documentation states that “the best practices for SEO remain relevant for AI features in Google Search” and that there are “no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary”. A page still has to be indexed and eligible to appear in Google Search.

So if somebody sells you AEO on the premise that SEO no longer matters, that premise contradicts the documentation of the largest answer engine they are selling access to.

Four myths worth dropping

Each of these is contradicted by a page Google publishes, linked below.

Myth 1 — you need llms.txt to appear in Google’s AI results

For Google Search, no. The same Google documentation states: “You don't need to create new machine readable files, AI text files, or markup to appear in these features. There's also no special schema.org structured data that you need to add.” Other tools may choose to consume such files, and you can publish one for your own reasons — but presenting it as a Google requirement is inventing a requirement.

Myth 2 — schema markup guarantees AI citations

Structured data can help systems understand page entities and can make a page eligible for specific search features. Eligibility is not entitlement: Google is explicit that meeting every requirement and best practice still does not mean content will be crawled, indexed or served. Use accurate structured data because it improves machine-readable clarity, not as a switch you flip for AI.

Myth 3 — FAQ schema is a secret AI-search tactic

Google announced in August 2023 that FAQ rich results would be shown only for well-known, authoritative government and health sites, and would no longer be shown regularly for everyone else. An FAQ section can still be genuinely useful to a reader. This site follows its own guidance: the pages you are reading have FAQ sections and deliberately ship no FAQPage markup.

Myth 4 — more content means more AI visibility

More pages create more crawlable surface area. That is not the same as more value, and Google's spam policies define scaled content abuse as “when many pages are generated for the primary purpose of manipulating search rankings and not helping users”. The better question for any new page is: what evidence or perspective does this add that the existing web does not already contain? If the answer is nothing, publishing it is a cost, not an asset.

A workflow that can actually be checked

Six steps, the last of which is the one most AEO advice skips.

1 · Pick the questions

Buying questions in the buyer’s own words. If a question cannot change a decision, it does not belong in the set.

2 · Measure the current state

Run them before you change anything. Without a baseline you cannot tell improvement from drift.

3 · Read the winning answers

What the answers that beat you actually emphasise — an observation, not a ranking factor.

4 · Check the claim is true

Verify the product fact before writing it on a page. This step is the one most AEO advice skips.

5 · Change one thing

A named page, a named change. Several simultaneous changes make the remeasurement uninterpretable.

6 · Remeasure

Same questions, same engines, same protocol — and report movement, not causation.

  1. Map real questions. Category, use case, comparison, alternatives, purchase criteria, branded, strategic positioning — and separate buyer questions from informational ones.
  2. Establish a baseline. Run the questions across the engines you care about and record mention, recommendation, citations, competitors and the provider and model context.
  3. Identify the gaps that matter. A high-intent “best vendor for X” loss usually matters more than a definition question. Do not prioritise every absence equally.
  4. Inspect the evidence layer. Winning-answer themes, cited sources, the relevant pages you own, and the verified product facts behind any claim you are about to make.
  5. Choose one addressable action. Clarify a verified capability on a relevant page; create a missing comparison or use-case asset; correct outdated product information; pursue a legitimate third-party route; or decide not to compete on a capability you do not have.
  6. Remeasure. Use a compatible question and protocol, record the measured change, and do not claim causality from a simple before and after.

The part most AEO advice skips

Almost all of it is about what to change. Very little is about how you would know it worked — and AI answers drift on their own, so something will always appear to have changed.

  • the same questions, unchanged and versioned, on both sides of the comparison
  • the same engines, with the model that actually answered recorded per run
  • failed calls excluded from the denominators on both sides
  • enough runs that you are comparing states rather than single samples
  • an explicit statement that a measured change is correlation, not proof of cause

Before / after≠Causality

Our reports do the first four and say the fifth out loud. The mechanics are on AI search monitoring; the metric definitions are on AI search analytics.

What makes content useful to an answer system

There is no universal formula, but content that gets used shares six properties.

Specific

It answers a defined problem rather than a category.

Evidence-backed

Facts, examples, methodology, first-hand experience.

Extractable

Clear headings, precise definitions and useful tables — without losing meaning when a fragment is lifted out.

Context-rich

Conditions, limitations and trade-offs, not just the happy path.

Original

It adds something beyond restating the current results page.

Maintained

Volatile pricing and product facts carry dates and get updated.

A page summarising ten existing articles adds little new evidence. A page publishing a benchmark, a real experiment, a transparent methodology or a first-hand dataset can become something other pages have a reason to reference — and answer systems may then meet the same evidence through several independent paths. Use public or explicitly permitted data; do not repurpose private customer evidence.

Third-party evidence matters — do not manufacture it

AI answers frequently use sources outside the vendor’s domain. The action is not to spam every site an answer cited.

Source relevance≠Source addressability

First determine relevance, legitimacy, addressability, eligibility, and whether your participation would even be truthful. That is the difference between a research finding and an outreach spreadsheet.

What AEO cannot guarantee

No honest AEO strategy can guarantee a ranking, a citation, a recommendation, permanent model behaviour, or a causal explanation of a black-box answer.

Google is unusually blunt about this market

“No one can guarantee a #1 ranking on Google”, and “Google doesn't evaluate or endorse third-party SEO tools, and these tools don't have access to Google's internal ranking data.” That applies to us as much as to anyone else selling in this category. Measure outcomes; qualify claims.

What is live today, and what is not. The measurement and the intelligence described on this page are real and running: buyer-question cohorts, three engines, recommendation separated from mention, competitor occupancy, citation evidence, bounded owned-site inspection, ProductTruth, commercial gap and action priority scored separately, and compatible before/after comparison. What does not exist yet is the software around it — no login, no dashboard, no alerts, no automated schedule, no API. During a founding pilot the workflow is run by a person and delivered as a reviewed report file.

AEO as an agency service

AEO becomes a strong recurring service when it has a repeatable loop.

A client does not need another monthly deck saying “AI visibility is up 4%”. They need to know what changed, which buyer question matters, what work is justified, what evidence supports it, and what happened after the work. That loop is the service layer, and it is what an agency pilot is built around.

A checklist before calling a page “optimised for answer engines”

If the last item has no answer, everything above it is unverifiable.

  • Is it indexable and crawlable?
  • Does it satisfy a real question somebody actually asks?
  • Does it contain non-commodity value?
  • Are the factual claims sourced, and are the product facts current?
  • Is the brand or entity unambiguous?
  • Does the page include useful context and stated limitations?
  • Is there first-hand evidence where first-hand evidence is possible?
  • Can you measure whether the relevant answer outcomes change?

Common questions

What is answer engine optimization (AEO)?

AEO stands for answer engine optimization. It is the work of getting a brand named, cited and recommended inside the answer a system generates, rather than ranked in a list of links a person then chooses from.

Is AEO replacing SEO?

No. For Google’s own AI features, Google states that SEO best practices remain relevant and that a page still has to be indexed and eligible to appear in Google Search. AEO adds a focus on the generated answer layer on top of that.

What is the difference between AEO and GEO?

AEO is the broader term for systems that answer questions; GEO emerged specifically around generative engines. In practice the workflows overlap almost entirely, and arguing about the acronym is not a strategy.

Does schema markup improve AEO?

Accurate structured data can improve machine-readable understanding and eligibility for supported search features. It does not guarantee an AI citation or recommendation, and Google states that no special structured data is needed to appear in its AI features.

Do I need llms.txt for Google AI search?

Google states you do not need to create new machine-readable files, AI text files or markup to appear in its AI features. Other tools may choose to use such files, but presenting llms.txt as a Google requirement is not supported by Google’s own documentation.

How do you measure AEO?

Track a frozen set of relevant questions across the answer systems you care about and measure mentions, recommendations, citations, competitors and change over time — with failed calls excluded from the denominators on both sides of any comparison.

Sources

Official documentation and first-party measurement, in that order. Third-party facts carry the date we last read them from the vendor's own page — verify before you act on a number.

How we measure, and how this page is sourced

What we actually measure

Every result comes from the official OpenAI, Google Gemini and Perplexity APIs with web search or grounding enabled. In our data those surfaces are recorded as openai_api_web_search, gemini_api_google_search and perplexity_api_sonar, and reports name them explicitly.

That is not the same thing as a logged-in person using the ChatGPT app. These are official interfaces with web search on, but they are measurement surfaces rather than consumer sessions: no personalisation, no memory, no app-only features. Results can differ from what any individual user sees, and we do not claim the two are identical. We would rather put that on the pricing page than in the small print.

How this page is sourced

  • Vendor pricing and features come from each vendor's own public pricing page, with the date we checked it. We do not copy figures out of other people's comparison posts.
  • Google-specific claims come from Google Search Central, not from SEO commentary about Google.
  • Research claims name the paper and separate what was demonstrated in a benchmark from what holds in the open web.
  • First-hand numbers come from our own measurement runs. Where the brand measured is founder-owned, the page says so in the same sentence as the finding.

This page is maintained by the person who built the measurement engine. There is no editorial team and no external reviewer behind it, and we are not going to invent one for a byline. How the measurements are collected.