Loading…
The Decionis Index
What governed execution actually looks like.
Every figure here is measured, aggregated across organisations, and published only above a privacy threshold. Where we cannot measure something yet, it is listed as unmeasured rather than estimated.
What the industry actually governs
The distribution of governed actions across execution patterns — the real shape of operational risk, not an opinion about it. Each row is published only where at least 3 organisations contributed to it.
1.Safe code deployment
4 actions · 100.0%
531 governed actions matched no pattern rule and are excluded from the shares above rather than folded into a bucket — a distribution that absorbed them would misstate exactly what it exists to report.
3 patterns are withheld because too few organisations contributed to them. A per-pattern row is more identifying than the site-wide total.
Classifier version 1.0.0. Published with the figures, because changing the rules changes the numbers.
Measured
Aggregated across at least 3 organisations. Each figure is counted at the action, not estimated from it.
81
Actions governed
last 7 days
High-stakes actions that asked permission before executing. The denominator for everything else here.
532
Actions held short of execution
last 30 days
Actions that were not permitted to run as requested — the ones a report would have described afterwards.
$15,020
Spend stopped before it moved
last 30 days
Monetary value of actions refused or held. Counted at the amount on the action, not an estimate of consequence.
No figure this period
These are instrumented, but the source returned nothing for this window. Shown as absent rather than as zero — they are different claims.
- Policy violations refused — last 30 days
- Override rate — last 30 days
- Approval time saved — last 30 days
Not yet measured
Listed openly, with what measuring each would require. These are the figures the Index is missing, not figures it is withholding.
Most copied policy
Which executable policy teams fork most often.
Needs — Fork and copy telemetry on the Policy Exchange. The catalogue is public and unauthenticated by design, so this needs counting without identifying who copied.
Fastest-growing policy
Which policy is being adopted fastest week over week.
Needs — A retained time series of per-policy adoption; today only current totals exist.
Average approval delay
How long an escalated action waits for the authority it needs. Distinct from time saved — this is the cost of escalation, not its benefit.
Needs — Escalation open-to-resolution timestamps. Time saved is measured today; the wait itself is not.
Where these numbers come from
Every figure is derived from decisions the protocol actually took, each carried in a signed Decision Dossier that can be verified independently. The methodology behind the monetary figures — including what it deliberately does not claim — is published separately.
Prevented-loss methodologyPublic drift dashboardA signed dossierThe Learning Loop