How we measure
You are not buying a chatbot workshop. You are buying a lab result: a measurement stack run on your files, plus a human who will not let a number become a claim without a class.
Three tracks
- 1
Public site
Proves: Friction on one host — form, price, buried CTA.
Never: Hours or euro. Label: Website hypothesis.
- 2
Owner export
Proves: Volume, waiting time, and measured hours if duration exists.
Never: A ranking from a conversation. This is the only euro line.
- 3
Decision call
Proves: Exceptions, a different rate, GO / PARK / NO. Label: You stated.
Never: Hours invented on the call. Decision Scan only.
Five labels
| Label | Meaning |
|---|---|
| Measured hours | Time you actually worked, if the file contains duration or start–end. |
| Waiting time | Calendar delay between emails — not multiplied by €40. |
| You stated | A number you declared (rate, deal value) — dated, not “measured in session”. |
| Website hypothesis | Friction on the public site — no hours, no euro. |
| Not enough data | We will not invent a ranking. |
Hours per week × €40/h, unless you declare another rate in writing. We do not write “pays back in weeks” — that needs a build price, which this scan does not set.
When a client number is published
- Baseline. Hours from the owner export — duration or start–end — not from a feeling and not from the website crawl.
- Rate. Hours given back × €40/h unless the scan agrees a different rate in writing.
- Publish. A client case and its numbers publish only when the client verifies them.
Edge cases
- Client did not confirm → we do not count.
- Scope changed → recount from zero for that client.
- Dispute → the number is held with a note.
Rounding: hours to tens, euro to hundreds — when a verified number exists.
Back to proof →