Government
How do you run a public-sector proof of concept that actually produces a decision?
A pilot that cannot fail cannot inform a decision. Design the stop condition first.
The answer
Write down, before it starts, the question the pilot answers and what result would end it — including the result that says stop. A proof of concept scoped to "explore the technology" cannot fail and therefore cannot inform anything; one scoped to a single workflow with a stated success measure and a decision date produces a defensible recommendation either way, which is the only output a public body can actually use.
What the pilot must produce besides a demo
A written readiness assessment of the agency’s own systems against what the technology needs. An honest operating-cost estimate for year two, not just the build cost for year one. And the evidence trail the eventual production system would need, exercised at small scale, because a pilot that skips it has not tested the part that makes the system defensible.
The result that ends it
Name it in advance: the measure, the threshold, and who decides. Pilots that quietly continue are the most expensive outcome in public-sector technology, because they consume budget and produce no decision anyone is accountable for having made.
Talk to the practice
Tell us what you are trying to build and what has to be true for it to work.
Contact