Will Kennard
Menu
Tools

AEO
Cheat Sheet.

New ways to be found.
The foundations still matter.

Your quick reference for answer engine optimisation. Keep it handy. Use it to find the next useful thing to improve.

The job, in one line

Help your brand be found,
understood and chosen.

Keep investing in SEO. Make your evidence easy to use. Learn how assistants represent you. Improve the work with agents.

  1. 01Question
  2. 02Interpretation
  3. 03Retrieval
  4. 04Answer
  5. 05Action

Retrieval and action are optional. Being fetched doesn’t guarantee being cited; being cited doesn’t guarantee being recommended.

Technical SEO

Make the information accessible.

Can the right systems actually find and read it?

  • Keep the foundations.

    Crawlable links, useful sitemaps, correct canonicals, intentional indexing rules and redirects that preserve intent.

  • Deliver the important content.

    Serve public answers, prices and specifications in the HTML. Check the raw response as well as the rendered page.

  • Test the whole path.

    Review verified crawler logs, CDN rules, bot challenges, timeouts and stale feeds. A working homepage proves very little.

  • Separate search from training.

    OAI-SearchBot is for search; GPTBot is for potential training use. Agree your access policy before changing bot rules.

RememberGoogle can render JavaScript. Other fetchers vary. Browser agents and search crawlers are different things.

In practice

Pick a commercially important product page. Can a plain HTML fetch establish the offer, current price and limitations? Then check whether the actual crawler can get through your infrastructure.

Customer decisions

Know who is asking. And why.

What decision does your customer need help making?

  • Keep keyword research. Add context.

    Use sales calls, support questions and lost-deal notes to understand real needs, budgets, locations and constraints.

  • Build a representative prompt library.

    Cover discovery, comparison, selection and follow-up questions for your actual customer groups.

  • Repeat your observations.

    Record platform, market, date and search use. Save answers and cited sources; a single result is an anecdote.

  • Follow the evidence.

    Read the sources behind recommendations. A model’s explanation of its choices is a research lead, not a ranking diagnostic.

RememberYour brand doesn’t have one universal AI ranking. The buyer’s context can change the answer.

In practice

Go beyond “best helpdesk”. Try a realistic buyer: a UK operations manager replacing an expensive tool for a team of eight. Then ask about integrations, migration effort and annual cost.

Content & evidence

Give people something worth using.

What can your business contribute that isn’t already everywhere?

  • Answer clearly, then substantiate.

    Use descriptive headings, useful comparisons and specific examples. Keep qualifications beside the claim they qualify.

  • Create human-led hero content.

    Original investigations, real tests, useful tools and expert opinions. AI can help with the work; it cannot invent the experience.

  • Treat your data as a marketing asset.

    Publish useful findings with a clear method, sample, period and limitations. Protect private information and check the interpretation.

  • Offer useful formats.

    Use feeds, CSVs, Markdown or APIs when a consumer needs them. Keep every format consistent with the maintained source.

RememberAn llms.txt file or special markup is no guarantee of citations. Useful information comes first.

In practice

A support platform could study which problems cause repeat customer contact. Publish the findings, methodology and expert interpretation, rather than another generic customer-service tips article.

Brand & corroboration

Be worth recommending.

Does the evidence beyond your website support your claims?

  • Get the actual offer right.

    Product quality, clear pricing and good customer service matter. Visibility cannot repair a disappointing experience.

  • Find influential sources.

    Inspect the publications, reviews, communities and partner pages that recur in answers for your category.

  • Earn relevant coverage.

    Work with PR, partnerships and customer teams. Contribute expertise and research rather than collecting empty mentions.

  • Keep the facts consistent.

    Correct outdated product details and listings. Make it clear who you serve, what you offer and what makes it credible.

RememberA citation can point to a third-party review. Your own page isn’t the only route into an answer.

In practice

If an integration partner describes an old version of your product, help them correct it. If customers need independent proof, a detailed customer review may be more useful than another claim on your homepage.

Measurement

Count the things that matter.

Did representation improve, and did it help the business?

  • Separate the outcomes.

    Track mentions, linked citations, recommendations and factual accuracy as different measures.

  • Show your working.

    Report the prompt set, sample size and repeat runs. Keep platforms and customer groups distinct where the data allows.

  • Connect to commercial value.

    Review referral visits, qualified enquiries, conversions and customer-reported discovery alongside search performance.

  • Test before claiming causation.

    Use stable baselines and matched groups where practical. Annotate content and platform changes; report uncertainty.

RememberA mention ≠ a citation ≠ a recommendation ≠ a sale. Crawler visits are not leads.

In practice

“Recommended in 18 of 60 eligible answers” is useful when the sample and scoring rules are clear. Pair it with enquiry quality and customer feedback, not a claim that 30% of the market saw you.

Agents & workflows

Automate the repetition. Keep the judgement.

Where can agents free you up to do more valuable work?

  • Build a maintained business brain.

    Capture positioning, product facts, customer research, metric definitions and previous decisions. Add owners, sources and review dates.

  • Connect knowledge, playbooks and skills.

    Knowledge supplies context. Playbooks explain the process. Skills package repeatable tasks, inputs, checks and outputs.

  • Make the work reviewable.

    Use code for exact checks and models for interpretation. Show evidence, flag uncertainty and define human approval points.

  • Measure the whole process.

    Include checking, errors and rework in time saved. Start with one useful workflow and improve it against known results.

RememberOKF makes knowledge portable. It doesn’t make it true, train a model or boost your public AI visibility.

In practice

For redirect mapping: match exact equivalents, suggest remaining candidates, flag uncertain matches, check destinations and export evidence for human review. Don’t automatically redirect every unmatched page to the homepage.

Keep an eye on

What changes next.

The agent is becoming another layer between a person and the web. Watch the useful changes. You don’t need to chase every announcement.

  • 01

    Retrieval & freshness

    Watch which sources appear and whether changed offers make it into answers. Google or Bing visibility is only part of the picture.

  • 02

    Personal context

    Test realistic follow-up questions. Budget, location and existing tools can change which brand fits.

  • 03

    Agents taking action

    Check whether an authorised agent can establish availability, understand terms and complete a supported task.

  • 04

    Formats & platform rules

    Follow changes to source attribution, paid placements and access. Invest when they affect your actual customer journey.

Need a hand with this?

Let’s make your
AEO strategy useful.

Technical SEO, AI search visibility, original content and workflows that help your team move faster.

Contact me about AEO