We write usage into the contract.
Training firms deliver attendance. Implementers deliver go-live. Both walk away before the thing is used. We sign a usage number. It is baselined before delivery starts, and measured at day 30, 60 and 90.
Attendance is not adoption. Go-live is not adoption.
A room full of trained staff who never open the tool is a cost, not a return. A system that goes live and sits unused is worse, because it was expensive. The gap between delivery and use is where most of this market quietly fails.
So we close it in the contract. Before delivery begins, we baseline where you are. Then we commit to numbers at three checkpoints. If the usage is not there, the standard is not met, and that is our problem to fix.
Three shapes of work. Nine checkpoints.
| Work | Day 30 | Day 60 | Day 90 |
|---|---|---|---|
| The AI EdgeMeasures whether people use the tool. | Named participants using the tool weekly in live work | Hours returned per function against baseline | Function-level workflows running without us |
| Core · System of recordMeasures whether the operation runs on the system. | Named users transacting in the system | Transactions posted inside the system versus outside it | Time to close the month against baseline |
| Core · Systems that actMeasures whether the tool no longer needs the people. | Containment rate, cases closed without a human | Escalation quality and error rate on closed cases | Cost per resolved case against the human baseline |
The exact thresholds are set per engagement, against your baseline, so they are numbers we will stand behind rather than numbers that sound good in a brochure. For systems that act, the standard inverts: it measures whether the tool has stopped needing the people.
Ask us to sign the number.
Nobody else in this market will. Bring us the work that keeps failing to stick, and we will tell you what we can commit to, in writing.