Proof
What a pilot measures
We have no customer result to show yet, and we will not invent one. This is the template a pilot fills in: measures, baselines and pass thresholds fixed in writing before the first user signs in.
No results yet. Every result cell below is empty on purpose. The first pilot will fill it, misses included.
The ten measures
| # | Measure | Definition | Source |
|---|---|---|---|
| M1 | AuCs reaching confirmed | Share of pilot assets under construction confirmed by week 4 | Decision audit trail |
| M2 | Ties to the cent | Proposals whose lines total the AuC cost, every amount from a cost source, approver differs from proposer | Evidence checks |
| M3 | Time, proposal to approval | Median hours from submit to approve | Audit trail timestamps; customer baseline |
| M4 | Rule acceptance | Share of proposals submitted without changing the rule's shares | Audit trail (override reasons) |
| M5 | Read-back match | Share of read-backs that match; each mismatch explained | Read-back result |
| M6 | Labour per project | Analyst time per AuC, with and without Attesta | Customer time sample (reported by customer) |
| M7 | Days from in-service to decision | Median, pilot AuCs | Worklist data |
| M8 | Regulated base effect | Only where a regulatory regime applies; otherwise not applicable | Customer finance (reported) |
| M9 | Support | Open severity 1 cases at exit | Support log |
| M10 | Usability | Feedback thumbs and comments; usage events per role | Usage and feedback view (opt-in) |
How it is run
- Week 0: record baselines, the regime answer for M8 and consent choices.
- Weekly: export usage, read feedback, log mismatches and reasons.
- Week 4: fill each result as pass or miss, label it cash or accounting-only, and write one paragraph on what did not work.
Anything not recorded by the product is labelled "reported by customer". Figures from one pilot are not a promise of your results, and you own the accounting outcome.