Measured under Broadcastwell method.
An open standard for reporting what AI engines answer. Seven requirements, published so anyone can meet them, and so a buyer can tell in one look whether a number can be checked.
Any tool or agency may state the phrase. The condition is meeting all seven requirements on the page that prints it.
Most published AI visibility figures cannot be checked by the person reading them. There is no denominator, no interval, no named engine and no way back to the answer the number came from. That is not a hard problem to fix; it is only a convention nobody wrote down. This is the convention, written down, under a licence that lets a competitor adopt it.
The seven requirements
| Requirement | What it means in practice |
|---|---|
| Receipts | Every number is traceable to the observed answer text and the URLs that answer cited. A figure with no retrievable answer behind it is not measured under this method. |
| Engines named exactly | ChatGPT, Claude, Perplexity, Google AI Overviews and Google AI Mode. An engine is named as itself; a vendor name is not a substitute for the surface that was queried. |
| Scheduled runs, with the repeat count stated | Runs are scheduled rather than taken when convenient, and the number of repeats per question and engine pair is printed beside the result. Extra runs taken because verdicts disagreed are labelled separately and never folded into the headline. |
| Counts before points, and every rate with its base and interval | A count and its denominator come first. A percentage is shown with the base it came from and a 95 percent confidence interval. |
| No composite score | Mentions, recommendations, recommendation position and citations are reported separately. They are different questions and a single blended number hides which one moved. |
| The method version stated | The version in force when a figure was produced is printed with the figure. Results produced under different versions are not compared. |
| A public dispute route | Anyone named in a published measurement can contest it in public, and the outcome is published beside the original rather than replacing it. |
Licence and attribution
The standard is published under Creative Commons Attribution 4.0. You may use it, quote it, adapt it and build a product on it, commercially included, with attribution to Broadcastwell.
The measurement method the standard is drawn from is published in full on the methodology page, including its version history, the scoring rules and the limits it states about itself.
Adopting it
Print the phrase where the figures are, not in a footer, and put the seven items within reach of the reader: the base and the interval beside each rate, the engine names in full, the repeat count, the method version, a route to the receipts and a route to dispute the result.
Broadcastwell's own published record works this way. The Absence Index prints a count, its base and a 95 percent interval on every vendor record, carries the release it was measured in, and publishes its dispute policy and log.
Questions
Who may use the phrase?
Any tool, agency or in house team that meets all seven requirements on the page it prints the phrase on. There is no application, no fee and no permission to ask for. The licence is CC BY 4.0, so attribution to Broadcastwell is the only condition.
Does using it mean Broadcastwell checked your work?
No. It is a statement about how a measurement was produced, made by whoever produced it. It is not an endorsement, an audit or a certification, and Broadcastwell does not review or approve anyone's use of it.
Why publish it rather than keep it?
Because the numbers in this market are mostly unfalsifiable. A buyer cannot check a figure with no base, no interval, no receipts and no named engine. A shared standard is worth more than a private one, including to a company that has to live inside it.
What if we can only meet some of it?
Then say which parts. A partial claim stated honestly is more useful than the phrase used loosely, and using the phrase while failing a requirement is the one thing the licence does not cover.
Start with a measured baseline
See which buyer questions leave you out, the sources visible in those answers, and the three changes worth testing first. The Category Audit covers ten questions; the Diagnostic extends to 35 questions and 525 scheduled observed answers, adaptive runs extra. AI answers vary, so repeated observations and disagreements are reported.
Start with the $490 Category AuditResults appear in your account; a sign-in link arrives by email. Credits in full against the Diagnostic within 30 days.
Get the Diagnostic, $990Results appear in your account; a sign-in link arrives by email. Credits in full against the program.