AI search optimisation method

AI SEO: Content Extractability Check

Page: https://addition-labs.com/resources/guides/ai-seo (fetched 2026-09-11, HTTP 200, 37.5 KB HTML, ~2,965 words of rendered body text) robots.txt: https://addition-labs.com/robots.txt (fetched 2026-09-11, HTTP 200) Target: citations for "ai seo" style queries in ChatGPT and Perplexity.

No .agents/product-marketing.md (or the older filenames) was found, so the audit below runs on the page and the request alone.

Step 3: Content Extractability Check

Score: 7 pass, 1 partial, 2 fail.

Two things the checklist does not ask but that matter for this page:

• Server-rendered. All body text, the table, the FAQ and the JSON-LD are in the raw HTML. The four <script type="module"> bundles are not needed to read the content. The header nav contains zero <a> elements in the HTML, so the H1 and standfirst are the first text a bot sees (H1 at character 7 of the rendered text).
• Discoverable. The URL is in /sitemap.xml and is listed with its description in /llms.txt (line 127, HTTP 200, 45 KB).

Step 4: AI Bot Access Check

/robots.txt in full:

User-agent: *
Allow: /

Sitemap: https://addition-labs.com/sitemap.xml

No Disallow lines at all, so every bot in the skill's list is allowed: GPTBot, ChatGPT-User, PerplexityBot, ClaudeBot, anthropic-ai, Google-Extended, Bingbot. CCBot (Common Crawl, training only) is also allowed; blocking it is a business decision, not a citation issue.

Live probe (HTTP status when fetching the guide with each user agent):

One correction to the skill's own bot list, from OpenAI's bots page (platform.openai.com/docs/bots, fetched 2026-09-11): the crawler that decides whether a page can appear in ChatGPT search results is OAI-SearchBot, not GPTBot. GPTBot is the training crawler, and OpenAI states the two settings are independent ("a webmaster can allow OAI-SearchBot in order to appear in search results while disallowing GPTBot"). Your robots.txt allows both, so nothing to change; it only matters if you later decide to block training.

The five highest-impact fixes (in order)

1. Add a named author with credentials, and switch the schema author to a Person

The only outright FAIL that both ChatGPT and Perplexity weight directly. Replace "By the Addition team" with a named person, a one-line credential (role, years, what they have run), and a link to an author page. In the JSON-LD change "author": {"@type": "Organization", ...} to {"@type": "Person", "name": "...", "jobTitle": "...", "url": "https://addition-labs.com/about/..."}, keep publisher as the Organization. The skill's own data: expert attribution is worth roughly +25 to 30% citation lift, and undated or unattributed content is named as an underperformer on every platform.

2. Rewrite the H2s to the query shapes, keep the editorial voice in the body

Retitle so the headings match the fan-out queries an assistant issues for "ai seo", without changing the argument underneath:

The H1 can stay as the brand headline; the <title> already carries "AI SEO" first.

3. Lead every section with a 40 to 60 word direct answer

Under each new H2, put the answer first and the scene-setting second. Example for "What is AI SEO?": open with a single definition sentence ("AI SEO is either using AI tools to produce search content, or getting your pages named inside AI answers; the two share a name, not a mechanism"), then the two-job explanation. Under "How much does AI SEO cost?", the first paragraph should carry the numbers that are currently buried in paragraph three (entry plans from $29 to $129 a month; Otterly tops out at $489). Perplexity in particular extracts atomic, self-contained paragraphs; the current openers ("Say you are asked to sort out AI SEO for a store...") cannot be lifted out.

4. Add the two comparison tables the fan-out actually asks for

"ai seo" queries fan out to "AI SEO vs traditional SEO" and "AI SEO vs GEO / AEO / LLM SEO". The page argues both points in prose (retrieval runs on the same index; the terms are the same second job) but has no table for either. Add a short comparison table under the "How does AI SEO work?" section: rows for index used, entry condition, what moves, how it is measured, with columns "Traditional SEO" vs "AI visibility". A second small table can map the synonyms (AI SEO / GEO / AEO / LLM SEO / AI visibility) to which of the two jobs each usually means, linking to your existing LLM SEO and AI Visibility guides.

5. Mark up the four-step triage as HowTo and expand the FAQ to cover the definition queries

The "How to buy AI SEO without buying two things" section is already four ordered steps (write down the outcome, ask the assistants signed out, open robots.txt, check indexing). Add a HowTo block to the existing @graph so the steps are extractable as a process. Then add three FAQ entries, on-page and in FAQPage, that match the remaining query shapes: "What is AI SEO?", "What is the difference between AI SEO and GEO?", "How much does AI SEO cost?" (answer with the sourced ranges already on the page). Keep answers at 40 to 60 words each.

Not a fix, but a note: robots.txt, freshness, sourcing, schema presence, server rendering, sitemap and llms.txt are all already in place. The gap on this page is attribution and heading/answer shape, not access.

Context still needed (from the skill's intake list; not required for the check above)

1. Have you run "what is AI SEO" / "AI SEO agency" / "AI SEO vs GEO" through ChatGPT and Perplexity signed out, and are you cited today? (One run is an anecdote; 3 to 5 runs per query gives a rate.)
2. Which competitors are being cited for those queries where you are not?
3. Do you have any third-party presence for the topic (LinkedIn articles, guest posts, review-site profiles)? The skill's data says brands are cited via third parties far more often than via their own domain, and this check only covers your own page.