AEO / AI Visibility Audit — find out why AI engines aren't citing you
A hands-on audit of how ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews answer the questions your buyers ask — who they cite instead of you, why, and what to change first.
No checkout and no subscription — tell us about your site below and we'll reply with a quote to confirm.
What you get
Ten deliverables, in one report you can share with your team. Every finding is evidenced — screenshots of the actual AI answers, the actual robots rules, the actual schema — not a scorecard we generated in the dark.
Publishing is not the same as being seen. Across the five sites we track, of 2,386 pages that had already earned an impression, 94.9% never got a single click in the observation window. What an audit answers is exactly this: which pages are stuck at that step, and why.
How this was measured- 01
Brand AI mention baseline
5-10 core buyer questions run across ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews, with screenshots of each answer — your starting position, on the record.
- 02
Competitor citation analysis
Who the engines cite instead of you on those same questions, and how often each competitor shows up.
- 03
AI answer source map
The domains and page types the answers actually pull from in your category — the shortlist of places worth being. For each tracked query it also names the third-party pages every engine on our sweep roster actually cited, grouped by host with citation counts. We do not broker paid placements.
- 04
Reddit / forum placement opportunities
The specific threads and communities the engines are already citing in your space, and which are realistic for you to earn.
- 05
Technical accessibility check
Whether GPTBot, PerplexityBot and Claude-Web can actually reach your content — robots rules, server-side rendering and page speed.
- 06
Content AI-readiness score
How extractable your key pages are: answer-shaped openings, self-contained sections, and the passages an engine could quote.
- 07
Schema audit
Which structured data you have, which is malformed, and which types are worth adding for your page templates.
- 08
Brand entity consistency
How consistently your brand is described across your own site and the third-party sources engines read — the naming and description drift that confuses retrieval.
- 09
Vertical new-question radar
The emerging questions being asked in your vertical that nobody has answered well yet — the openings worth writing into.
- 10
30-day action plan
A prioritised, sequenced list of what to change first, with the expected effect and effort of each item.
See a full one, before you buy one
We ran the whole thing on fdmini.com — a shop we own — and published the result unedited: real citation counts from every engine we sweep, the crawler and schema findings, the page-one rankings no engine cites. The four sections a person writes are left blank on purpose, so you can see exactly which half is a probe and which half is an analyst.
See a full sample auditNew to any of these? AEO · AI citations · brand entity · share of voice
Two promises
Full refund if we find nothing actionable
If the audit surfaces no actionable opportunities for your site, you get a full refund. We would rather hand back the fee than sell you a report you cannot act on.
A shareable report link
Delivery includes a shareable report link, not just a PDF in your inbox — so your team, your agency or your client can open the same evidence you did.
We run TriRank on ourselves first
Every figure below was measured by TriRank on a site it runs on, with the date it was read next to it. Some of these sites are ours; the numbers from the others are published with those sites' permission.
Three sites, three different measurements. That is evidence we run the product on real sites — it is not a sample anyone should generalise from.
Or have us run the fixes
A $399 audit ends with a plan and someone on your side has to execute it. The Founding tier is the other half: our pipeline ships the fixes the audit turns up, a founder reviews the site with you once a month, and you get a white-label weekly report to hand to whoever asked for one.
For comparison: B2B Gaps, an agency selling a comparable done-for-you SEO/AEO program, publishes its entry tier at $5,000 per month (b2bgaps.com/pricing, last modified 2026-02-15, checked 2026-08-31).
| Option | Published price | Source |
|---|---|---|
| People doing the work, at an agency | From $5,000 per month | B2B Gaps entry tier (b2bgaps.com/pricing, page last modified 2026-02-15, checked 2026-08-31) |
| An AI-visibility monitoring tool — it measures, and it writes nothing | $99 per month and $399 per month | Profound's two published tiers, Starter and Growth (tryprofound.com/pricing, read 2026-08-29; its third tier is still Custom) |
| TriRank | $69 one-time for the market report · $399 one-time for this audit · $999 per month per site for done-for-you | This site's own current prices, set out on this page and on the pricing page |
Three prices for three different problems, not one problem at three price points. The first buys people's time. The second buys measurement and writes nothing. We are not going to tell you we are some number of times cheaper than either, because these columns do not divide into each other.
We ship the fixes
The items your audit turns up go into our pipeline and get executed, rather than back to you as a checklist.
A founder reviews it with you, monthly
One call a month, with a person who can change what we are doing — not a report generated on a schedule.
White-label weekly report
A weekly write-up with your name on it, for the person who has to justify the spend.
Who does what
We do
- Run the crawler and citation probes on a schedule and report what came back, including the weeks where nothing moved.
- Draft the content the findings call for, instead of handing you a list of things to go and write.
- Deliver each article into your CMS — WordPress, Ghost or Webflow — as a draft post, so what is waiting for you is a post to approve rather than a document to paste.
Still you
- Connect Google Search Console once, so the ranking half of your report is your own data rather than an estimate.
- Read the drafts and decide what goes live. They arrive as drafts, never as published posts.
The right-hand column is short on purpose, and it is not short because we ran out of automation. Both of those steps are places where somebody has to decide whether a sentence about your business is true, and that judgement is yours to make. We would rather hand you a draft you throw away than publish something you would not have put your name to.
The Founding rate is capped at three sites. It is a founding price for the first three customers, not a discount that expires — once the three are taken, the fourth site onwards is $1,499/month.
How to book
Fill in the form below with your site and the buyer questions you care about, and we will confirm the scope and start date. Search Console access is optional — it sharpens the ranking half of the analysis, but the AI-citation half runs without it.
Frequently asked questions
Who is this audit for?+
Teams that already get traffic from search and want to know why AI assistants recommend someone else. It suits SaaS, e-commerce and service businesses with a live site and a defined set of buyer questions. If you have no site content yet, the audit will mostly tell you to write some — start there instead.
What do you need from me?+
Your domain and the questions your buyers actually ask, plus the competitors you care about. Google Search Console access is optional: with it we can cross your real rankings against AI citations, without it we still run the full AI-citation, technical and content analysis.
What is the refund policy?+
If we find no actionable opportunities for your site, you get a full refund of the $399 fee. The judgement is about actionability, not about whether the findings are flattering — an audit that says 'you are already cited everywhere' with evidence is a valid result and is not refundable.
Can I upgrade to ongoing monitoring?+
Yes. The audit is a one-time engagement and stands alone, but most teams pair it with a TriRank subscription so the citation sweeps on your tracked queries keep running every 3 days after the plan is delivered. See the pricing page for the monthly plans.
The AI Search Market Report — $69
The $399 audit is about your site. This is about your market: how AI assistants are summarising it, which sources they keep citing, and where the gaps are. It is a document, not an engagement — a smaller thing to buy, and an honest one to buy instead.
- You pay $69 now. Stripe collects the card and the email we send the report to; you do not need an account.
- The first batch ships after our first audit cycle. We would rather tell you that than invent a date.
- The full contents list and the sample section are published below.
Rather not pay up front? Reserve a copy below and pay on delivery.
- Reserve now, pay on delivery. We are not asking for a card and nothing is charged today.
- There is no refund policy because there is nothing to refund — if you do not want it when it ships, you do not pay.
- The full contents list and the sample section are published below.
- The first batch ships after our first audit cycle. We would rather tell you that than invent a date.
The full contents list+
This is the document's own text, reproduced without editing. The report is written in English.
- Market summary — the shape of the finding, in one page.
- Engine coverage, stated plainly — how many engines this edition measures, and what that costs the conclusions.
- What gets cited, by what was asked — the page shapes an answer engine reaches for, split by the question it was answering. (This is the section published free as the sample.)
- Whose pages get cited — your own domain against everybody else's, and what kind of page each side contributes.
- How wide the source pool is — concentration, the long tail, and how far apart two neighbouring questions really are.
- Brand visibility baseline — how the sampled brands score across all three questions, and why the spread matters more than the average.
- Does domain authority predict citation — the correlation, its confidence interval, and the honest limit of a null result.
- AI crawler behaviour, observed — volume against cadence on a live property.
- What to do with this — five moves the measurements support, each with the number behind it.
- Method, limits, and what we did not measure — the frame, and the things this edition does not speak for.
The sample section, in full+
This is the document's own text, reproduced without editing. The report is written in English.
What gets cited, by what was asked
This is the section published free as the report's sample.
Every cited URL in the sample is sorted into one of five page shapes, decided by the URL path alone. In precedence order — first rule that matches wins:
| Shape | The rule |
|---|---|
root | The path is exactly the front door. |
alternatives | The path names a comparison: alternativ*, competitor, a vs segment, compar*. |
review | The path names a review, a rating or a testimonial. |
article | The path sits under blog / article / resource / guide / news / post, or carries a year-month date segment. |
other | Everything the four rules above do not match. |
Precedence is load-bearing and overlap is normal, not exceptional: a URL such as a blog path ending in a review slug matches both article and review, and this report reads it as a review — the person who asked about reviews landed on a review, and the fact that the host files it under a blog is a publishing detail. Reordering the rules would change every share below, which is why the order is printed here rather than left in the code.
Nothing in this classification looks at the host. There is no list of "review sites" or "forums" anywhere in the pipeline, which means the rules apply just as well to a host nobody has seen before, and there is no roster to go stale.
Across all 2,469 citations:
| Page shape | Share of all citations |
|---|---|
other | 38.3% |
alternatives | 29.6% |
review | 21.3% |
root | 8.4% |
article | 2.5% |
Read other first and read it as a limit: the four named rules speak for the rest and not for that 38.3%. It is dominated by profile and listing pages — company pages, encyclopedia entries, app-store listings, deep product paths — which have no single path signature to key on.
The same citations, split by the question that produced them:
| Page shape | Brand name | Reviews | Alternatives |
|---|---|---|---|
root | 20.8% | 4.5% | 2% |
alternatives | 0.8% | 2.1% | 81.5% |
review | 5.1% | 54.9% | 0.8% |
article | 2.8% | 2.1% | 2.7% |
other | 70.6% | 36.5% | 13% |
This is the strongest single result in the report. The three columns are not variations on one distribution; they are three different distributions.
- Ask about alternatives and 81.5% of what comes back is a comparison page. Review pages contribute 0.8% — effectively nothing.
- Ask about reviews and 54.9% is a review page, while comparison pages fall to 2.1%.
- Ask about the brand name and neither dominates: 70.6% lands on
other— profiles and listings — and 20.8% on somebody's front door, the only column where the front page is a serious contributor.
The practical reading: a page earns citations for the question its URL and its subject match, and gets very little credit outside it. A single well-written page covering "reviews and alternatives" is competing in two neighbourhoods with one address, and the measurement says the engine treats those neighbourhoods as almost separate. The next section puts a number on how separate.