The offer
AI SEO audit for WordPress plugins: find out why AI engines recommend your competitor
Thirty checks, four engines, two weeks. At the end you know why the models recommend someone else and which three fixes are worth doing first. Every check I run is published on this page, before you pay for any of them.
What arrives at the end of two weeks?
Five things: a prompt run of 30 buyer questions across ChatGPT, Perplexity, Google AI Mode and Copilot with screenshots; your share of voice against three competitors you name; the citation-source map; the 30-check report with evidence and severity for each; and a 90-day roadmap ordered by effort against likely effect.
Then a 45-minute walkthrough where we go through it live and you can argue with me.
What is the citation-source map, and why does it matter most?
It is the list of specific URLs the engines pulled from when they recommended someone other than you. On a typical audit that is five or six pages, and they are rarely the ones founders expect — a three-year-old roundup post, one Reddit thread, a review-site profile nobody has updated since 2024.
You cannot fix a gap you cannot see. This is the part clients say changed how they think about it.
What are the thirty checks?
Published in full, in seven groups. Several clients have run them themselves instead of hiring me, which is fine — it is the same reason the Index is public.
- Is your content in the HTML, or does it need JavaScript?
- Do robots.txt and your CDN or WAF both let OAI-SearchBot, PerplexityBot and Googlebot through?
- Are your key pages indexed in Bing, not just Google?
- How fast does a cold fetch return?
- Are your key pages indexable and snippet-eligible, with no stray noindex, nosnippet or max-snippet:0?
- Are the pages you care about within two clicks of the homepage?
- Do your headings ask the questions buyers ask?
- Is the answer in the first 30% of the page?
- Can a paragraph be lifted and still make sense?
- Do you name the product, or say “we” and “it”?
- Is there a statistic with a linked source?
- Do you have a real comparison table anywhere?
- Which “best X plugin” listicles include you?
- What does Reddit say about your plugin?
- Is there a YouTube video that names you?
- Do review-site profiles exist and are they current?
- How many branded mentions does the web carry?
- Is your support forum answering or embarrassing you?
- Does your readme describe the problem or the features?
- Is the changelog current enough to signal maintenance?
- Does “tested up to” match the current WordPress?
- Do your screenshots and banner load?
- Is your category placement the one buyers search?
- Is the product name and description the same on your site, Crunchbase, LinkedIn and G2?
- Does Organization schema carry a sameAs cluster?
- Is there a real person attached to the product?
- When were your key pages last substantively updated?
- Is lastmod in your sitemap honest?
- Can an AI agent complete your trial, demo or contact form: real labels, real buttons, no layout shift?
- Is your key action offered to agents as a tool (WebMCP) or a documented API?
Aligned with Google’s guidance. The checks that touch Google Search follow what Google’s own AI features guidance asks for: pages that can be crawled, indexed and shown as snippets, with content written for people. None of them rely on things Google has ruled out, such as llms.txt, chunking pages for AI or rewriting content “for AI”. What Google’s guide confirms, and what our data adds.
Third-party presence means mentions you earned: a roundup that included you on merit, a Reddit answer from a disclosed account, a review from a real user. I do not buy placements, plant threads or trade links, and the audit flags any that already exist. Why manufactured mentions do not work.
Agent readiness is new and marked emerging. AI agents that fill in forms for buyers are early, so these two checks carry less weight than the rest. Softrecs runs the same setup on its own site first: what Lighthouse found on ours.
Why is retrievability weighted critical?
Because every other check assumes the engine can read the page. GPTBot, ClaudeBot and PerplexityBot do not execute JavaScript, while Googlebot and Bingbot do. A plugin site can rank first in Google and be entirely invisible to ChatGPT, Claude and Perplexity for the same query.
Roughly one site in five fails at least one of the three pages that matter. It is usually the cheapest thing on the list to fix.
Is this an AI visibility audit?
Yes, with one difference. An AI visibility audit usually gives your site a score. This one also runs your buyers’ real questions on ChatGPT, Perplexity, Google AI Mode and Copilot, and maps which pages the engines cited instead of you. The score tells you something is wrong. The citation-source map tells you where to fix it.
Can I use the thirty checks as a GEO checklist?
Yes. The list above is the whole checklist, and nothing is held back for paying clients. Start with the retrievability group: if an engine cannot fetch your pages, nothing in the other six groups helps. For a wordpress.org plugin, the free AI visibility checker runs several of those checks in under a minute.
What a checklist cannot give you is the prompt run and the citation-source map, because both depend on your category’s live answers. That part is what the $750 pays for.
Who is this not for?
Free-only plugins with no revenue — the maths does not work for either of us. Products with no site content at all, where the honest answer is “write some pages first” and you do not need $750 to hear it. And anyone who wants a guarantee that ChatGPT will recommend them by a given date. Nobody can promise that, and the people who do are why this field has a credibility problem.
Movement shows around month three, not month one. I put that in the proposal in writing. If you need pipeline this quarter, an audit is the wrong purchase — say so and I will tell you what I would do instead.
If you would rather see the method before paying for it, the prompt set builder generates the same fifteen prompts for your category, free.
Before you buy, read the GEO case study results: what has moved, and what has not yet.
Start with twenty minutes instead
Book a review and I will run your category live on the call. You see the actual answers rather than my summary of them, and you decide afterwards whether the full audit is worth $750.