A five-phase method built for evidence, not optimism
We do not open with a slide deck about best practice. We open with your data. Each phase has an exit test the sponsor signs off on before the next begins, which keeps scope honest and stops the work drifting into open-ended advisory.
Our method has five phases. Each ends with an exit test, agreed before the phase starts, that the sponsor signs off in writing. That single discipline does most of the work of keeping an engagement honest: it prevents scope creep, forces early honesty about data quality, and gives everyone a shared definition of the word done.
1. Discovery
Two weeks. We interview the people doing the work, observe the flow, and list the candidate friction points with a first estimate of impact.
2. Baselining
Two to three weeks. We build the quantitative baseline: capacity, workload, cycle time and rework. This is the phase most clients value most.
3. Design
Three to four weeks. Roles, decision rights and rhythm are redesigned together with the people who will live inside the new model.
4. Pilot
Two to four weeks. The design is run in one unit before wider rollout. Pilots are meant to fail early and cheaply where they are wrong.
5. Handover
One to two weeks. Documentation, a re-measurement method and a manager toolkit, so the client can repeat the measurement without us.
Governance that does not waste anyone's time
Each engagement has a weekly one-hour working session and a fortnightly sponsor review of thirty minutes. There is a single written status note per week, never longer than one page. Steering committees are formed only where the client's own governance requires them; we do not create committees to create the appearance of control.
Risks are logged in the open from week one, including the ones that reflect badly on us. If a phase is going to miss its exit test we say so at least a week before the review, not in the review itself.
Why phases rather than a single sprint
A single continuous sprint tends to produce a final report that nobody has time to react to. Phases produce decisions at intervals, which is how large organisations actually change. The exit tests also give a sponsor natural points to pause or stop, which we consider a feature rather than a risk: clients who know they can stop early are more willing to start.
We would rather be stopped at phase two with an honest baseline than carried to phase five on a mandate nobody still believes in.
What data we need, and what we do with it
The baselining phase normally needs read access to payroll or rostering records, workflow or ticketing data, and any existing operational reports the client already trusts. We prefer to work with extracts placed in the client's own environment where possible, and we never move personal data into our own systems without a data processing agreement in place. Individual-level data is used for aggregate measurement and is not used to rank, discipline or assess any employee.
Where data quality is poor, we treat that as a finding in its own right and say so early. A baseline built on unreliable inputs is worse than no baseline, because it creates false confidence. In those cases we recommend either a short data remediation step or a narrower scope where the inputs are trustworthy.
Every metric in the baseline is traceable to a query or a source document, and the method note explains the assumptions behind each one. If a reviewer disagrees with a number, we can show exactly where it came from and adjust it in the open rather than defending a total that no longer holds.