Before any engagement, we measure how often AI engines actually recommend a company when its buyers ask. Three briefs from recent measurements, anonymized because the findings belong to the companies. The numbers are real, live, and re-runnable.
A pain management practice outside Atlanta. Respected physicians. Real patients. A fully booked clinic. By every measure that mattered for the last twenty years, a healthy business.
We ran 54 live AI searches using the exact questions their patients ask. Best pain management doctor nearby. Specific treatments for specific conditions. Three engines, three runs each, live web search turned on.
They appeared zero times. Not buried on page two. Not mentioned and ranked low. Zero. The AI engines recommended their neighbors by name, down to individual doctors a few miles up the road.
One of North America's largest specialty steel producers. Generations of manufacturing. Fully integrated mills. A category leader by volume, the kind of company whose products are inside things you use every day.
Here is what their homepage gives an AI engine to work with: a signal 13 characters long. That is the entire machine-readable case for a company that leads its category.
We ran 63 live buyer searches for their exact product category. Their Share of Model came back at 7%. Ninety-three percent of the time, a buyer asking AI about the products this company leads in gets pointed somewhere else.
A Wisconsin manufacturer of vehicle control systems. Their electronics run inside equipment across construction, agriculture, and fire and rescue. Engineers actively source exactly what they make. This is not a demand problem.
We asked three AI engines about their category, 63 times. They surfaced in none of the answers. 0% Share of Model. That part, we had seen before.
So we asked by name. Nine branded queries, the kind where the AI just has to find the company the user already named. The engines couldn't. 0 of 9. An engineer who heard about this company at a trade show and asked AI to look them up would come back empty.
Every number on this page comes from live queries run against AI engines with web search enabled, the same way a real buyer or patient would ask. Each question runs three times per engine across three engines, because a single response is an anecdote and a repeated one is a measurement.
Nothing here is modeled, estimated, or projected. If you want to see the method applied to your own company, we will run it. The results are re-runnable on demand, which is the point. You should be able to check our math.
A free scorecard with your own live numbers: how often AI engines recommend you, who they recommend instead, and where you stand.
Get Your Free ScorecardFree. No call required. Numbers you can verify.