There’s a simple test for anyone selling AI visibility services: ask them to run their method on themselves and publish the result. Almost nobody will do it. Scores are easy to sell when the client can’t check them, and the industry has settled comfortably into that arrangement.
We took the test. Before writing another word of advice, we pointed our own citation pipeline, the same one behind our $995 Citation Audit, at revenueexperts.ai, and we’re publishing what it found. Not because the results were flattering. Because a measurement method you’d hide from is a method you shouldn’t sell.
I’ve spent 25+ years in digital marketing and finance, and the finance half is why our audits work the way they do: numbers that can’t be traced are numbers that don’t count. That standard is written into everything below.
What does a proper AI citation audit require?
Four things most tools skip: real buyer questions, repeated runs, verified citations, and evidence grades.
Our pipeline asked 50 questions our actual buyers ask, from “what is AEO” down to pricing and vendor comparison, and put each one to the AI engines three separate times, on documented dates. Repetition matters because engines answer differently run to run; a single-run check can call a fluke a fact. Then the step that defines the method: for every page an engine cited, the pipeline opened that page and checked whether it actually supported the claim. Engines mis-cite more often than most marketers would believe. Finally, every finding got graded by evidence strength: verified, pattern, or hypothesis. Nothing in the report pretends to be stronger than it is.
That’s the full methodology, published, step by step. We built it because we couldn’t find a tool that met it.
What did the audit show about our own visibility?
Two findings that hold up, and one pattern that matters more than either.
The findings that hold up: every AI engine we tested cited revenueexperts.ai at least once, and when a question produced a citation for us, it repeated across nearly every re-run, usually at or near the front of the engine’s source list. Where we exist, we exist reliably and prominently. That’s the profile of a domain the engines trust.
The pattern that matters more: our citations cluster on branded questions, the ones that already contain our name. On the open category questions, where a buyer hasn’t heard of us yet, the engines answer with other companies’ links. Roughly half of our visibility also flows through third-party pages about us rather than pages we control.
Here’s the expert read, and it’s the reason this finding is worth publishing: this is the most common pattern we see in B2B AI visibility. We’ve now made it the first thing we check in every client audit. Companies build identity content for years, so the engines learn who they are. But nobody builds answer content for the category questions where shortlists actually form. The result is a business that AI can describe but never recommends. If a company that builds citation tools for a living shows this pattern, it’s worth finding out whether yours does too.
Why publish results that don’t flatter us?
Because in this market, honesty is a technical requirement, not a virtue.
The AI visibility industry runs on numbers buyers can’t verify, and buyers have noticed. Distrust of visibility dashboards is the most consistent complaint in our own market research. There’s exactly one way to be the exception: build a method that survives exposure, then expose it. Our reports grade every finding by evidence, our methodology is public, and our own baseline is in this article. A vendor whose numbers can’t survive that treatment has a marketing problem. A vendor who won’t try has told you something.
The other reason is professional discipline. A citation audit is a snapshot, not a prophecy. It proves where you stood on documented dates, and nothing else. Anyone who tells you a specific change “will” raise your citations, without measuring before and after, is selling confidence they don’t have. We wrote that rule into our client reports. It would be strange to exempt ourselves.
What we’re doing about it: the same playbook we sell
The audit’s question map doubles as a work order: every question where an engine cites a competitor and not us is a page assignment. So the plan is the one our clients get, run on ourselves.
Target the winnable questions first: the ones where engines already hand out citations, just not to us. Build answer-first pages for each, structured the way engines extract. Publish owned, canonical pages for everything engines currently source from third parties, so describing us becomes easiest through us. Keep the technical foundation clean against the 36 factors that both our site and our free audit are built on. Then re-run the same 50 questions and publish the before-and-after here, whichever way it goes.
That last step is the commitment that separates a method from a pitch. It’s already made.
What this means for your company
Start with the question this article should have raised: do you know your own pattern?
Most companies discover they have one of three profiles. Invisible entirely, where the engines can’t even parse the site. Branded-only, where you’re described when asked by name but absent from the shortlist questions; that’s our profile, and probably the most common. Or category-cited, the profile everyone wants, where engines recommend you to buyers who’ve never heard of you.
Finding out costs nothing to start. The free 60-second audit tells you whether AI engines can technically read and trust your site, scored against all 36 published factors. The $5 tracker tests five real buyer questions across ChatGPT, Claude, Gemini, and Perplexity, with repeated runs, in about 90 seconds. And when you’re deciding where next quarter’s content budget goes, the $ 1495 Citation Audit maps all 50 questions in your category, verified and graded, with the prioritized roadmap. The fee is credited in full toward any fixed engagement.
Frequently asked questions
What is an AI citation audit? A measurement of which websites AI engines cite when answering the questions your buyers ask. A rigorous one uses questions built for your specific category, repeats every measurement, verifies each cited page against the claim attached to it, and grades findings by evidence strength.
Why would a company be cited on branded questions but not category questions? Branded questions hand the engine your name; it only has to describe you. Category questions force the engine to choose the best answer among many sources, and it picks pages structured as direct, verifiable answers to that exact question. Years of identity content build the first kind of visibility. Only deliberate answer content builds the second, and the second is where buying decisions form.
Can any vendor guarantee AI citations? No, and the claim itself is a red flag. Engines change models, sources, and behavior without notice. What can be guaranteed is process: documented measurement before, verified work during, and the same measurement re-run after, so the change is proven rather than promised.
The re-measurement gets published here when it’s in. That’s the standard we’re asking you to hold every vendor to, starting with us. If you want your own baseline first, the free audit takes sixty seconds.