Presenting condition

A c£10M programme to transform a c£90–100M social care service. The programme had been attempted for six years using qualified practitioners and proven frameworks, and had failed repeatedly. A new attempt was beginning with the same objectives, similar resourcing — and, crucially, the same underlying conditions that had defeated every previous effort.

The historical baseline for delivering a new operational service in this environment: twelve to eighteen months from programme start. Six years of prior attempts had not produced a working service.

What was about to happen instead

The proposed approach was another structured programme using the same methodology as its predecessors — better-resourced, better-governed, with lessons-learned documentation from the previous failures incorporated. The assumption was that the previous attempts had failed due to execution problems: insufficient rigour, inadequate change management, insufficient stakeholder engagement.

Each of those diagnoses was plausible. Each was wrong.

Diagnosis

Three conditions operating simultaneously, each credibly explaining the six years of prior failure, with interventions that directly contradicted each other. Contradictory: incompatible mandates between the commissioning function, the operational teams, and the political leadership, with no mechanism for resolving conflicts between legitimate competing priorities. Complex: a social care system whose behaviour emerged from interactions between dozens of teams, agencies, and regulatory requirements that no individual or team fully understood — where the previous attempts had treated as predictable a system that was genuinely unpredictable. Insecure: a trust deficit accumulated over six years of failed programmes, where people had learned that surfacing problems led to blame rather than support.

The sequencing was the critical diagnostic insight. Applying the Complex response before the Contradictory condition was resolved would have generated learning in a system still receiving contradictory signals — the learnings would have been undermined before they could be acted on. Applying any constraint or experimentation before the Insecure condition was addressed would have added pressure to teams already operating in fear. The sequencing mattered as much as the interventions themselves.

Intervention

Contradictory condition first: forced ranking of competing mandates, explicit decision rights assigned, a named individual with authority to resolve conflicts within 24 hours. This was politically uncomfortable. It required each stakeholder group to accept that their priorities would sometimes lose. Without resolving this condition first, every subsequent intervention would have been undermined by the same structural conflict that had defeated six years of prior effort.

Complex condition second: structured learning protocols established, safe-to-fail experiments designed with explicit rollback triggers, systematic capture of what the system was teaching implemented before any further change was attempted. The shift from predict-plan-execute to probe-sense-respond — within a programme structure that had previously demanded certainty it couldn’t have.

Insecure condition third and throughout: leadership vulnerability modelling, blameless retrospectives with structural follow-through, safe channels for early problem surfacing. This was the lever that needed the longest sustained application — trust built through consistent behaviour over weeks, not announced in a single session.

Outcomes

Six years of prior failure. Three months to first delivery. The practitioners who had attempted this programme before were not less capable. The frameworks they had used were not inadequate. The single variable that changed was diagnostic precision applied to which condition was addressed first, in which sequence, with which lever. That is the case for diagnosis before intervention made as plainly as I can make it.

Presenting condition

A £100M flagship transformation programme. Sixty concurrent projects. 1,200+ dependencies. Progress had stalled across multiple workstreams despite capable teams, experienced programme management, and visible executive commitment. Three different senior leaders offered three different explanations for why delivery was underperforming — and each explanation was internally coherent.

That level of diagnostic divergence at senior level is itself a signal: the actual condition was being read through professional lenses rather than diagnosed from evidence.

What was about to happen instead

A programme-wide intervention was being planned: a single unified governance framework applied across all 60 projects, standardised reporting, and increased coordination overhead. The diagnosis was that the programme lacked coherence and that tighter alignment mechanisms would restore momentum.

Applied uniformly to three different underlying conditions, this would have improved one, had no effect on another, and actively worsened the third.

Diagnosis

Three entirely different underlying conditions presenting identical surface symptoms across different parts of the portfolio. Contradictory conditions in the governance layer — incompatible mandates with no explicit hierarchy and no decision authority at programme level, causing teams to optimise for visibility rather than completion. Complex conditions in the technical integration work — system behaviour emerging from interactions nobody fully understood, where expert analysis was producing competing theories rather than convergence. Insecure conditions in the delivery teams — a trust deficit causing problems to hide until they became escalations, and best people going quiet rather than surfacing the decisions that needed to be made.

Intervention

Condition-specific interventions applied in parallel across the portfolio, with different levers for different workstreams: Clarity in the governance layer (forced ranking, explicit decision rights, named resolution authority with 24-hour conflict resolution), Learning in the integration workstreams (safe-to-fail experiments, parallel hypothesis testing, systematic capture of what the system was actually teaching), Empathy in the delivery teams (leadership vulnerability modelling, blameless retrospectives, safe channels for early escalation).

The coordination overhead being planned was never implemented. Tight coordination across three different conditions would have added governance to a system where the governance layer was already part of the problem.

Outcomes

The critical validation: identical surface symptoms across the portfolio required opposite interventions depending on the underlying condition. This wasn’t instinct applied at scale. It was a replicable diagnostic pattern producing consistent results across radically different project types within the same organisation.

Presenting condition

A transformation programme that had been running for two years was losing momentum. ThroughFlow — the proportion of work completing cleanly versus being reworked — had dropped to 52%. Lead times were extending. Senior leaders were spending 35–40% of their week in escalation meetings, firefighting problems that should have been resolved at team level.

What was about to happen instead

The existing consultancy had recommended adding an orchestration layer — more structured governance, tighter reporting, cleaner accountability. The instinct from the programme team was that better coordination would restore momentum.

This would have added governance to a system where the real problem was that accurate information wasn’t reaching the people who needed to act on it. More structure. Same trust deficit. Escalations continuing to compound.

Diagnosis

Two days of structured observation revealed layered conditions: Rapid on the surface — too much WIP, queues that never shortened, constant context-switching — but with an Insecure condition underneath. A trust deficit was causing problems to hide at team level until they became escalations. The proposed orchestration layer would have addressed the Rapid symptoms while leaving the Insecure condition untouched and worsening it through increased surveillance.

Sequencing: Empathy lever first to rebuild psychological safety, Resilience lever second to constrain WIP. Reverse the sequence and WIP limits land on a team still operating in fear — adding constraint to fragility.

Intervention

Phase 1 (weeks 1–8): Leadership vulnerability modelling — specific commitments made and kept weekly, visible to teams. Blameless retrospectives reintroduced with structural follow-through. Safe channels for surfacing problems before escalation. Making problems visible earlier, not rewarding their surfacing after they’d already become crises.

Phase 2 (weeks 9–18): Hard WIP limit set at 65% of historical finish rate. Nothing new started until something finished. Explicit priority hierarchy published and maintained. Lead time measurement introduced so the improvement would be visible, not just felt.

Outcomes

The orchestration layer was never built. What shifted things was addressing the trust deficit that was preventing accurate information from reaching the people who needed to act on it. The Rapid condition resolved as a consequence — once problems surfaced early, the escalation load dropped and throughput recovered.

Presenting condition

A volunteer-heavy maritime rescue charity absorbing a 60% funding drop while operational costs held fixed. COVID had simultaneously removed the fundraising infrastructure that sustained the organisation and increased pressure on the operational capability the charity existed to provide. The leadership team had more commitments than before, less resource than before, and no slack in which to make deliberate choices about what to deprioritise.

One team had moved from functional to effectively non-delivering over six months — zero completed work across a 12-week period prior to the engagement, despite the team being intact and technically capable.

What was about to happen instead

The proposed response was to add resource and increase oversight. Both were unavailable at the scale required, and neither would have addressed the conditions that had caused the team to stall. The instinct was to treat a structural problem as a capacity problem.

Diagnosis

Two conditions. Rapid: the funding shock and COVID disruption had dramatically increased the number of active threads while reducing the organisation’s capacity to finish anything. The team was carrying more concurrent commitments than at any point in the previous three years, with no explicit priority model and no mechanism for saying no to new requests. WIP had roughly doubled while finish rate had collapsed. Complex: the interdependencies between volunteer networks, fundraising, operational delivery, and regulatory compliance were behaving in ways the organisation’s existing models didn’t account for. Interventions that had worked reliably before were producing unpredictable results in the changed environment.

Intervention

Resilience lever for the Rapid condition: a hard WIP limit set at three concurrent items per person, down from an average of eight. Explicit priority decision made by leadership on which threads were active and which were formally paused. Nothing new started until something completed. The discomfort of visible paused items was acknowledged and held rather than managed away by adding capacity.

Learning lever for the Complex condition: two-week experiment cycles with explicit hypotheses, defined success signals, and rollback triggers if signals didn’t improve. Prior assumptions about volunteer behaviour, donation patterns, and operational interdependencies treated as hypotheses to be tested rather than facts to be acted on. Systematic capture of what the changed environment was actually teaching, replacing the assumption that pre-COVID patterns would return.

Both levers applied simultaneously, each to its matched condition. Single-lever discipline within each condition; parallel application across conditions.

Outcomes

The value of this case is the context. Financial services and technology are the natural habitats of transformation recovery work. A volunteer charity absorbing a funding shock in a pandemic is as far from that environment as it’s possible to get. The conditions were the same. The levers were the same. The sequencing discipline was the same. The results followed the same pattern. Different sector. Same diagnostic approach. Same improvement.

Presenting condition

Two product teams that had been highly productive six months earlier were now producing almost nothing. Lead times had quadrupled. Rework was running at 40% of total effort. The engineering lead attributed it to technical complexity; the product lead attributed it to unclear requirements; the CEO attributed it to the teams needing to scale up with additional headcount.

Three diagnoses from three senior people, none of them the same. That level of diagnostic divergence is itself a signal — the actual condition is being read through professional lenses rather than diagnosed from evidence.

What was about to happen instead

The CEO was preparing to hire four senior engineers. The assumption was a capacity problem: more skilled people would restore delivery throughput. The hiring process was already in motion when the diagnostic engagement began.

Diagnosis

Observation over two days revealed two conditions, layered. Rapid: the teams had been handed 23 concurrent threads following Series B — product expansion, technical debt, infrastructure scaling, regulatory compliance — with no explicit priority and no capacity model. WIP had roughly tripled with no corresponding increase in finish rate. Anxious: a failed production incident four months earlier, handled badly from a blame perspective, had produced a risk-aversion culture. Engineers were seeking approval for deployment decisions that should have been team-level calls. The approval bottleneck was adding days to delivery cycles and the fear of the next incident was creating over-engineering that generated rework.

Intervention

Week 1: WIP hard limit — each team carries a maximum of three threads concurrently. 23 threads reduced to six total across both teams, by explicit priority decision from the CEO. Six things were visibly paused. The discomfort was real and acknowledged rather than managed away.

Week 2: Explicit risk boundaries published by engineering leadership — what decisions teams could make without approval, what required a flag (not approval), and what required escalation. Not removing governance, but making its scope explicit so teams could move fast within it without triggering anxiety about whether they were overstepping.

Weeks 3–6: Two-week delivery cycles with blameless reviews — what shipped, what didn’t, what we learned. No post-mortems on incidents, only forward-looking process improvement. Making it safe to finish things and ship them, rather than safe only to keep working on them.

Outcomes

The headcount request was withdrawn after week four. The capacity was always there — it was being consumed by rework and approval overhead generated by two undiagnosed conditions. Neither additional people nor additional process would have addressed either cause.