Unwinding the
Feature Factory

This is the same client as Case Study 01. During COVID-19, the organisation suspended all external engagements as a cost-control measure. When the Payroll product organisation needed structural help, the leadership team made a specific exception to recommission this work.

No pitch. No competitive process. The decision was made on the basis of results already observed in the same organisation. That kind of endorsement — revealed preference under constraint — is a different quality of signal from any reference or testimonial. I include this case study because that decision is itself part of what this engagement demonstrates.

The full diagnostic data, time allocation methodology, Five Dysfunctions assessment detail, and intervention documentation are on orchiture.com.

Two Payroll product teams. Engineers working as isolated individuals, assigned to whichever feature needed resource at any given moment. No defined product area. No team ownership of outcomes. Ticket throughput was the measure — not value delivered.

The feature-centric model had a quantified cost. Context switching — the tax on individual assignment — consumed 8% of total engineering time. Waiting, driven by an unresolved inter-team dependency, consumed a further 6.4%. Together, these two structural failure modes were costing the equivalent of nearly three-quarters of a working day per engineer per week — close to 38 days per engineer each year, or the equivalent of almost eight working weeks of engineering capacity lost to structural inefficiency.

All changes were proposed to the teams as experiments — not imposed as policy. The prevailing management culture had favoured control over education and experimentation. The approach here was deliberately the inverse.

Time allocation data was captured at baseline (~2 months in) and at close (~16 months in). The comparison provides a direct measure of how structural and cultural changes affected the distribution of engineering time.

Time allocation is a more honest diagnostic than velocity. Velocity can be gamed; time allocation cannot. When 70% of your engineering time is not on business value work, no amount of sprint process improvement will fix it. The structural model has to change.

The inter-team dependency — the single most impactful structural problem — was solved by the teams themselves, not imposed by the practitioner or by leadership. That the teams self-organised to resolve it is a more durable outcome than any mandated fix would have been. It's also a signal about where the real work of an embedded engagement happens: not in the recommendations, but in the coaching that makes the teams capable of solving their own problems.

Cultural change takes longer than structural change. Trust, conflict tolerance, and results-orientation all moved quickly once psychological safety improved. Accountability — teams holding each other to standards — takes longer. It requires the other four dimensions to be stable before it can take root. It should be the last thing expected, not the first thing measured.

A score of 3.40 reflects a team that had moved from avoiding difficult conversations entirely to beginning to hold each other to shared standards. The journey, not just the destination, is the signal.

The engineers in these teams were the same people at close as at the start. What changed was the conditions in which they were working: a defined product area, team-owned accountability, and the habit of resolving their own structural problems. A team capable of those things does not need the practitioner to keep working. That is what the engagement built toward.