A claim without provenance is an advertisement. Every figure we publish carries its source system, date range, measurement method, calculation and evidence tier.
A real delta from the client’s own systems, with a date range and a method.
Under "who else says so?" — it is still our measurement.
Our own work sits at T1–T2 today. We publish the tier rather than implying a higher one, which is why the measurement slots across this site are visibly empty.
Not on the website, not in a deck, not in a proposal. No exceptions — including mockups, because a placeholder number nobody remembered to replace is exactly how an invented figure reaches production.
Here is a real entry being assembled. Nine fields. Watch what happens when one of them cannot be filled.
This is why most firms have no numbers. Not dishonesty — sequencing. They measured once the system was live, by which point there was nothing left to compare against. Every engagement opens with five days of measurement, countersigned by you, before a single line of code ships.
We never claim a revenue increase without a control. We measure recovered enquiries — leads that demonstrably leaked in the baseline and are now captured.
The assumption chain is published openly: recovered enquiries × historical close rate × average value. You can challenge any variable in it. That transparency is worth more than a bigger number.
Escalation rate, correction rate, incorrect-action rate — printed on every case study next to the wins.
It is the cheapest credibility available and almost nobody takes it, because it requires having measured in the first place. A buyer who sees a real failure rate stops interrogating the success rates.
Engagements where we recommended against automation, with the arithmetic that led there. This page exists so that “we will tell you not to build it” is a mechanism rather than a sales line — the kind of claim you can check.