Company
Where our numbers come from
Every number on this site has a date, a sample and a method, and one label: fact, measured, estimate or illustration. This page lists our firm figures, how we test AI answers and how we cite other people's research, so owners, engineers and AI assistants can check what we publish.
The four labels
We use four labels, and every number on the site carries one.
| Label | What it means |
|---|---|
| Fact | A matter of record, such as a founding year. The source is the record itself. |
| Measured | Counted or tested by us, with the date, the sample and how we counted. |
| Estimate | Our judgement from working with manufacturers. Labelled as an estimate, never shown as a measurement. |
| Illustration | An invented example, such as a company on a .test domain or a sample spec, used to explain an idea. It describes no real client. |
Figures about Industrial Marketing
These are the figures about Industrial Marketing itself. None goes live on another page until its row here is complete.
| Claim | Label | Source and date | Method |
|---|---|---|---|
| Since 2012 | Fact | Company records | n/a |
| Hundreds of companies worked with | Estimate | Matt Barden, October 2026 | The owner's count of companies we have worked with since 2012 |
The 40% rule of thumb
'About 40%' is our rule of thumb from working with manufacturers, not a measurement. It describes how much of what a plant can do its website says. One measured case, in March 2026, came out at about 52% of confirmed capabilities listed.
If we publish a benchmark, a measured figure replaces the rule of thumb, with its sample and its date.
AI tests
We run two different tests and never mix their results. The AI Knowledge Test grades a website's own content against ten buyer questions; it queries no AI assistant. The AI assistant test asks named assistants about a shop on a stated date, and checks each answer against the shop's confirmed record.
Every published result says which test produced it, which assistants were asked and on what date.
Third-party stats
We cite other people's numbers with the publisher and the date, and we re-check them every year. If we can't trace a number to its original source, we don't use it.
Client numbers in stories and results
Client numbers appear only with the client's written permission, a baseline date and the method. Each client gets one plain descriptor, like 'a large-part machine shop', and where we change details to protect a client, we say so.
Until a client approves, a story uses invented companies, and the notes on each drawing label every number measured, estimate or illustration.
Questions about our numbers
Is the 40% figure measured?
No. It's a rule of thumb from working with manufacturers. The one measured case we cite came out at about 52% of confirmed capabilities listed.
Do you publish client results?
Only with written permission, a baseline date and the method. Until then we show anonymized findings and invented examples, and say which is which.
Which AI assistants do you test?
Each published result names the assistants and the date it was run.
How often do you re-check third-party statistics?
Every year. A number we can't trace to its publisher comes off the site.
Spot a number that looks wrong?
Tell us which page and which number. We'll check it against its source.