Skip to main content
Free toolkit

The AEO Toolkit

Everything we use to make this site citable by answer engines — templates, a scored page audit, and a two-week rollout. Free, complete, no email required.

What is in the toolkit?

The AEO toolkit is a free set of templates, checklists, and scoring rubrics for making a website citable by answer engines: an answer block template, a 24-point page audit, an llms.txt starter, JSON-LD entity snippets, a 50-question citation-share tracker, and a two-week rollout sequence.

The templates

Copy each block, replace the bracketed slots, and ship. They are written to be used verbatim rather than adapted.

Answer block template

Fill the five slots and delete the brackets. Keep the finished paragraph between 40 and 70 words.

[TERM] is a [CATEGORY] that [WHAT IT DOES], used to [PURPOSE].
It [KEY MECHANIC OR SCOPE: number, duration, or boundary].
Unlike [NEAREST NEIGHBOUR CONCEPT], it [THE ONE CONTRAST THAT SEPARATES THEM].

Rules:
- Sentence one names the term. No pronouns pointing outside this block.
- State the quantity if the question implies one.
- No "In today's", no rhetorical questions, no brand voice.
- Place it immediately after the H1, before any setup.

llms.txt starter

Serve at /llms.txt as text/plain. Annotate every URL with the question it answers.

# [Organization Name]

> [One sentence defining the organization: category + differentiator, under 30 words.]
> Primary experts: [Name] ([role]), [Name] ([role]).
> Contact: [email] | Canonical site: https://[domain]

## Core definitions
- https://[domain]/[slug]: Answers "[the exact question]".
- https://[domain]/[slug]: Answers "[the exact question]".

## Research and original data
- https://[domain]/[slug]: [What the dataset or study contains and its sample.]

## People
- https://[domain]/author/[slug]: [Name] — [role], [topics they can be cited on].

## Methods and frameworks
- https://[domain]/[slug]: [Named method] — [what it measures or produces].

## Usage
Content may be quoted with attribution to [Organization Name] and a link to the source URL.

Person + Organization JSON-LD

The minimum entity kit. Emit the Organization once on the root, then reference its @id from every article.

{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "Organization",
      "@id": "https://[domain]/#organization",
      "name": "[Organization Name]",
      "url": "https://[domain]",
      "description": "[The canonical one-sentence description.]",
      "sameAs": ["https://www.linkedin.com/company/[handle]"]
    },
    {
      "@type": "Person",
      "@id": "https://[domain]/author/[slug]#person",
      "name": "[Full Name]",
      "url": "https://[domain]/author/[slug]",
      "jobTitle": "[Title]",
      "affiliation": { "@id": "https://[domain]/#organization" },
      "knowsAbout": ["[topic]", "[topic]", "[topic]"],
      "sameAs": ["https://www.linkedin.com/in/[handle]", "https://[publisher]/[author]"]
    },
    {
      "@type": "Article",
      "@id": "https://[domain]/[slug]#article",
      "headline": "[Headline]",
      "description": "[Meta description.]",
      "author": { "@id": "https://[domain]/author/[slug]#person" },
      "publisher": { "@id": "https://[domain]/#organization" },
      "datePublished": "[YYYY-MM-DD]",
      "dateModified": "[YYYY-MM-DD]",
      "mainEntityOfPage": "https://[domain]/[slug]"
    }
  ]
}

FAQPage JSON-LD

Only emit this for questions that are visibly answered in the page body, in the same words.

{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "[The question, phrased the way users ask it]",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "[Two to four sentences that fully answer it, matching the on-page text.]"
      }
    }
  ]
}

Citation-share tracker (CSV)

Freeze the question set, run it monthly across every engine, and compute citations divided by question-engine pairs.

question,intent,engine,run_date,cited,description_accurate,competing_domains,notes
"what is an incentive audit?",definitional,chatgpt,2026-09-01,1,1,"domain-a.com;domain-b.com",
"what is an incentive audit?",definitional,perplexity,2026-09-01,0,0,"domain-a.com",
"how do you redesign a compensation incentive?",procedural,ai-overviews,2026-09-01,,,,
"incentive design vs change management",comparative,claude,2026-09-01,,,,
"who does cultural performance audits?",commercial,copilot,2026-09-01,,,,
"who is [expert name]?",entity,chatgpt,2026-09-01,,,,

# citation_share = sum(cited) / count(rows)
# accuracy_rate  = sum(description_accurate) / sum(cited)

Representation audit prompt

Run this in each engine, unmodified, and paste the raw output into your tracker before editing anything.

You are auditing how you represent an organization.

1. What is [Organization Name]? Answer in three sentences.
2. Who leads it, and what are they known for?
3. What frameworks or methods is it associated with?
4. Who are its closest alternatives, and how do they differ?
5. List the sources you used, with URLs.
6. State plainly what you do not know about this organization.

Do not speculate. If you are uncertain, say so.

How do you score a page for answer engine readiness?

Score each item all-or-nothing and total it out of 100. Below 60, the page is invisible to retrieval. Above 85, it competes on substance rather than structure.

Extraction (40 points)

Whether a machine can lift a complete answer out of the page without the rest of the document.

  • 8Answer block within the first 70 wordsA 40-70 word self-contained paragraph sits directly under the H1 and answers the page's core question in full.
  • 5No orphan pronouns in the openingThe first sentence names the subject rather than opening with 'This', 'It', or 'That approach'.
  • 6Question-shaped headingsSubheads are phrased as the questions real users ask, not as one-word labels like 'Overview' or 'Pricing'.
  • 5One question per sectionEach chunk answers a single question so a retrieval match is unambiguous.
  • 8Specific, committed claimsNumbers, durations, and boundaries are stated rather than deferred to a sales conversation.
  • 4FAQ section with real questionsThree to six questions taken from actual buyer language, each answered in two to four sentences.
  • 4Tables and lists for comparisonsComparative content is structured, not buried in prose, so it survives chunking as a unit.

Entity and trust (30 points)

Whether the machine can decide who is making the claim and why they can.

  • 7Named author with a hub pageThe byline links to a stable author URL that exists on the same domain.
  • 6Person schema with sameAsjobTitle, knowsAbout, affiliation, and at least two external profile links are present.
  • 5Organization schema with a stable @idArticles reference the organization node rather than repeating it inline.
  • 4Consistent entity phrasingThe same bio sentence and credential wording appear everywhere the entity does.
  • 5Original materialThe page contains data, a method, a case, or firsthand practice not available elsewhere.
  • 3Dates and revisions visibledatePublished and dateModified are in both the prose and the structured data.

Machine access (20 points)

Whether retrieval systems can reach and interpret the page at all.

  • 6Server-rendered contentThe answer text is in the initial HTML, not injected after hydration.
  • 4llms.txt publishedA curated, annotated map of the site is served at /llms.txt as plain text.
  • 3Sitemap includes the pageThe URL appears in an indexed sitemap with a realistic lastmod.
  • 3AI user agents allowedrobots.txt does not block the crawlers whose engines you want citations from.
  • 4Canonical and clean URLOne canonical URL, no redirect chain, no tracking parameters in internal links.

Cluster strength (10 points)

Whether the page sits inside a body of work rather than standing alone.

  • 4Links to adjacent questionsThe page links to three or more related definitions or guides on the same domain.
  • 3Inbound internal linksAt least three other pages link to this one with descriptive anchor text.
  • 3External corroborationThe claim or framework appears on at least one domain you do not own.

The two-week rollout

  1. Day 1
    Freeze the question set

    Collect 50 questions from sales calls, Search Console, and comparison queries. Version the list and stop editing it.

  2. Day 2
    Baseline the representation

    Run the audit prompt in five engines. Record absence, misdescription, and misattribution separately.

  3. Days 3-4
    Fix the entity layer

    Author hubs, Person and Organization schema, one canonical bio sentence reused everywhere including external profiles.

  4. Days 5-7
    Rewrite the top ten pages

    Add an answer block under each H1, convert labels to question-shaped headings, and commit to specific numbers.

  5. Day 8
    Publish the machine layer

    llms.txt, sitemap refresh, robots.txt review for AI user agents, canonical cleanup.

  6. Days 9-11
    Build one definition cluster

    Five to eight linked definition pages around the term you want to own, each answering exactly one question.

  7. Day 12
    Earn external corroboration

    One podcast transcript, one guest piece, or one dataset release that repeats the canonical description off-domain.

  8. Days 13-14
    Score and schedule

    Run the 24-point audit on every rewritten page, log the first tracker row, and set a monthly re-run.

Questions about the toolkit

Is the AEO toolkit really free?
Yes. Every template, checklist, and rubric on this page is published in full, copyable, and free to use in commercial work. No email is required.
How long does it take to see citations?
Entity fixes and answer blocks typically show up in engine answers within two to six weeks, because retrieval indexes refresh faster than classic ranking. Representation changes in model training data take far longer and may never fully update, which is why corroboration off-domain matters.
Do I need special software to run this?
No. The audit is a spreadsheet, the tracker is a CSV, and the schema is hand-written JSON-LD. Tooling helps at scale but adds nothing to the first two weeks.
Does The Incentives Lab sell AEO services?
No. The Incentives Lab is an incentive design practice. This toolkit exists because it is the exact process used on this site, and publishing it is cheaper than explaining it.