It is a good interview question because it cannot be answered with a brochure. Here is our sequence, with the durations we actually hold to. One remark first: aiming at 20 % of operations does not mean touching 20 % of tasks. It means taking the two or three processes that carry that weight on their own.

Phase 1, weeks 1 and 2: understand

Process mapping, interviews with the teams, recording volumes, times and costs, state of the tools and quality of the data. This phase produces a costed list, not yet a solution. It is described in detail in our article on the starting method.

Phase 2, week 3: choose

Prioritisation by impact, feasibility and risk. We keep only two candidates for the quarter. That is deliberately few: three projects run in parallel by a team that already has a day job all finish late.

By the end of that week, each candidate has a costed objective, a chosen operating mode, a human boundary in place, and a written stopping criterion. The stopping criterion is what separates a project from an expense.

Phase 3, weeks 4 to 7: prove

Building a first version that handles real flow on a restricted scope. The aim is not the beauty of the system, it is proof: does the quality hold on real cases, including the awkward ones nobody mentioned in the meeting room.

At this stage everything leaving the company goes through human validation, without exception. It is also the phase where the real business rules surface, the ones nobody had written down.

A useful proof is not a successful demo. It is a full week of real flow with no hidden incident.

Phase 4, weeks 8 to 11: industrialise

This is the phase failed projects skip. What has been proven is taken back and made operable: permissions, error recovery, queues, logging, alerts, control of the cost per run, tests on the error cases met during the previous phase.

It is also the moment to decide which steps can move to automatic, based on observed correction rates, and which stay under validation.

Phase 5, weeks 12 and 13: roll out and hand over

Switch to the full scope, training for the teams involved, documentation handed over, and above all measurement: the indicators taken in week 1 are taken again identically. The end of quarter review does not discuss impressions, it looks at two columns.

Three outcomes are possible, and all three are acceptable: extend, adjust, or stop this process and go back to number three on the list.

What it requires on your side

  • An available decision maker. One hour a week, able to settle a business rule without convening a committee.
  • A business referent. The person who actually knows the process, not the one who wrote it down.
  • Technical access. Obtained in week 1, failing which the whole schedule slips by as much.

A provider promising this schedule without demanding those three conditions is promising something they do not control. That is the point on which we prefer to be unpleasant at the first meeting rather than on day sixty. Our Scale offer is built on exactly this rhythm.