Pilot and rollout

A pilot is a controlled experiment with a decision attached. It is not a smaller version of the whole programme, and it is not complete until someone decides what happens next.

Purpose

• Prove the approach on real work, with real constraints, before committing the organisation to it.
• Surface the organisational frictions that a plan cannot anticipate.
• Produce credible evidence for the people who have not yet been convinced.
• Reduce the cost of being wrong.

Selecting a pilot

Representative

Uses the same technologies, constraints and dependencies as the wider estate — not the easiest team, and not the hardest.

Bounded

One product or service with a clear owner and a delivery cadence that will generate feedback within the pilot window.

Willing

The team volunteers. A coerced pilot produces compliance and no learning.

Visible

The outcome can be seen by the rest of the organisation. A pilot nobody notices cannot create momentum.

Reversible

Failure does not damage customer-facing commitments or regulatory standing.

Pilot phases

1. Define success before starting Agree the metrics, the baseline, the targets and the decision rule — what result triggers scale, adjustment, or stop. Written down and signed by the sponsor.

2. Baseline Measure current performance on the chosen metrics: deployment frequency, lead time, change failure rate, restore time, and any workload-specific KPI.

3. Implement Apply the target practices against the standards in this library, with a two-week feedback cadence.

4. Observe and collect feedback Capture both quantitative results and qualitative experience from the team and their downstream consumers.

5. Review and adjust Compare against baseline. Identify what worked, what did not, and what needs to change before scaling.

6. Decide Scale, extend the pilot, adjust the approach, or stop — against the rule agreed at the start.

Metric set

Dimension	Metric	Baseline	Pilot target	Result
Throughput	Deployment frequency
Throughput	Lead time for change
Stability	Change failure rate
Stability	Mean time to restore
Quality	Defect escape rate
Workload	Service-specific KPI
Experience	Team confidence (survey)

Rollout

Once the pilot is validated, rollout is a repeatable process rather than a second programme.

1. Codify the pilot. Turn what worked into templates, pipeline definitions, and documentation — not a narrative account.
2. Train the trainers. The pilot team becomes the reference for the next teams, so knowledge spreads horizontally rather than through consultants.
3. Sequence by readiness. Roll out to teams with the closest profile first; each wave reduces the remaining distance.
4. Keep the feedback loop. Each wave has the same baseline, measure and review cadence as the pilot.
5. Publish the evidence. Every wave's result updates the organisation's view of what the change delivers.
6. Retire the old path. Once a team is on the new process, remove the old tooling and permissions. A parallel manual path will be used.

Common failure modes

• Pilot as an exception. The pilot team gets special attention, special approvals and extra staff, so the result does not generalise.
• No decision rule. The pilot succeeds and the organisation argues about whether to proceed.
• Success measured only technically. Deployment frequency improves; the business outcome does not move.
• Scale before stabilise. Rollout begins before the pilot's practices are documented, so each new team reinvents them.
• Old path stays open. Everyone congratulates the new process and uses the old one under deadline pressure.

Related standards

• Organisational change
• Maturity model
• Workload health KPIs
• KPI tracking template