after we build · part nine · trust and stopping
Client trust as the ultimate lagging signal
No metric captures everything. But the closest single signal to "is the chain working?" is client trust — whether the client believes the team understands their situation and is making it better.
Trust is lagging. It builds slowly over cycles where the prediction was checked and the result was shared honestly. It erodes in a single cycle where the team shipped something that didn't work and didn't acknowledge it.
The signal reading shared with the client is the most important trust-building artifact. Not because it shows success — because it shows honesty. A team that says "we predicted 15 minutes, it came in at 22, here's what we learned and what we'll adjust" builds more trust than a team that says "we shipped the feature" and moves on.
The client is not a recipient of the chain's output. The client is a participant in the loop — and their experience of the prediction check is what decides whether the next cycle gets funded.
When to stop
Not every feature should be continued. Not every product should be maintained. The portfolio view also asks: when should the team stop?
Stop investing in a feature when:
- The prediction was checked and the value didn't materialize, and
- The model update doesn't reveal a clear path to a different approach.
Stop maintaining a product when:
- The VRI has been below a threshold for multiple cycles, and
- The cost of continuing exceeds the value being delivered.
These are hard decisions. The chain doesn't make them easier to feel. It makes them easier to justify — because the data is there. The predictions were checked. The signal readings are written. The decision is grounded in evidence, not in opinion.
Enough to know what is alive.
DORA signals are tracked and read. VRI is current. Rework is broken down by chain level. The portfolio review has happened in the last 90 days and produced at least one decision — including, where warranted, a kill.