Risk · 4 min read
Good evidence can tell you to stop
A delivery model that cannot produce a recommendation to stop is not producing evidence. It is producing momentum.
Every stage-gated process claims that stopping is possible. Very few make it survivable. By the time a decision point arrives, the commitment is large, the sunk cost is visible, and the organizational cost of saying stop exceeds the cost of continuing quietly.
The structural fix is to make the first commitment small enough that stopping is unremarkable. A decision to stop after a short, bounded first stage is a normal governance outcome. A decision to stop eighteen months into a programme is an incident.
This changes what evidence is for. If the only acceptable finding is confirmation, the stage was theatre. Useful evidence has to be capable of surprising the organization that commissioned it.
The practical marker is whether anyone can describe, in advance and in specific terms, what result would cause the initiative not to continue. If nobody can answer that, the checkpoint is decorative.