Transformations are won or lost in the informal network: a case study

Cophi has spent several years measuring that system, using Organisational Network Analysis (ONA). This note sets out what it looks like from the top, why it is invisible from there, and what changes once it is made visible, drawing on a recent engagement with a European energy infrastructure company.

The problem as leaders experience it

The strategy is agreed. The early quarters show movement. Then, at some point, effort keeps rising while results flatten, and there is no single failure to point to. Each function reports that its own part is working. The dashboards show nothing specific. The leadership team accumulates explanations that each cover part of the slowdown, and none that cover the whole.

BCG's 2025 analysis of around 1,700 transformations found only about a third accelerate growth in a value-accretive way. An earlier BCG study of more than 850 digital transformations put detail on the rest: 26 per cent created little or nothing, and 44 per cent created some value while missing their targets. That middle band is the one that matters. The typical transformation does not fail; it under-delivers, slowly enough that no moment forces the question of why, and by the time the question is asked the market the strategy was written for has often moved.

Why the top cannot see it

Information is edited on its way up. Issues arrive at the executive level already framed, softened and paired with a proposed response. Leaders hear the organisation's account of itself, not a direct observation of it.

The org chart describes accountability, not work, which is why a redesign that looks right on paper can still disappoint. Redrawing the boxes does not redraw the habits. The old routes, and the reasons people use them, mostly survive the restructure intact. Work moves through an informal network: who asks whom, who responds, who is routed around, and which team has quietly become the point through which a disproportionate share of volume passes. That network determines how quickly anything gets done. It is measurable, and almost no leadership team has measured it.

A strategy can stall for procedural reasons: targets that pull against each other, or a planning cycle that cannot keep up. But much of the friction that decides whether a strategy lands is relational rather than procedural. An engineering function built around safety and a commercial function built around speed will each read the other's caution or urgency as obstruction, and neither will say so, because both positions are defensible and the conflict is one of concern rather than competence. A team let down by a central function years ago will still route around it long after the people and the reasons have changed, and will describe the relationship as fine, because the workaround has become normal.

The combined effect is that transformations fail invisibly, in the informal network, and usually well before the failure is reported to anyone in a position to act on it.

What the gap looks like when measured

We measure collaboration at the interfaces between teams and divisions: who works with whom, and how well each side says the interface works, scored separately in both directions. Each direction gets its own score, and together they make up the Net Collaboration Score (NCS). The directional structure is the important feature. Scoring each side separately exposes asymmetry, the cases where one party's experience of a relationship differs materially from the other's, which is exactly the information a conventional engagement survey averages away.

Case: a European energy infrastructure business

The sponsor of a growth strategy built around several newer business lines had noticed persistent friction with a central project-control function. But the friction looked different depending on where in the organisation you stood, and a felt problem is not enough to secure the budget or attention to fix it.

We measured and visualised the collaboration network between every team in the organisation, scoring each interface in both directions. That did three things.

It confirmed the sponsor was right, and showed who agreed: the business lines rated the relationship well, the central function rated the same relationship poorly. That gap turned a private hunch into a shared, evidenced case.

It changed the diagnosis. The friction was not a breakdown; it was a one-sided load. The central function was receiving far more requests than it could handle. Each business line could see only its own requests, so none of them realised how much was arriving in total. What looked like slowness from outside was a queue nobody could see. What looked from the business lines like an unresponsive function was, from inside it, a capacity problem.

It surfaced a second issue underneath the first: the business lines' exploratory way of working was colliding with the central function's structured one. Two legitimate operating models, each experiencing the other as friction.

It is not that nobody knew. People usually sense these things. But everyone held a slightly different version of it, and opinions that differ cannot be acted on. None of it would have shown in a survey that averages both sides of a relationship together either. What the measurement changed was not whether people agreed the problem existed, but what they understood the problem to be. The interface is now being redesigned with both sides in the room, working from the same picture.

What measurement changes

The shift is from treating collaboration as a sentiment to be surveyed to treating it as a system to be mapped. Three things follow.

A shared picture is what mobilises people to act. A leader can hold a private view of a problem for years and it will not carry a business case on its own. A small group holding an evidenced view, agreed by the people on both sides of the friction, is what actually secures a workshop, a budget or a redesign.

The intervention is targeted, not organisation-wide. When collaboration is measured properly, the interfaces that need attention are a small fraction of the organisation. The effort goes to the handful of relationships carrying the strain, and most of the business is not asked to change anything at all.

The fix matches the fault. Seeing the specific behaviour behind a friction point, rather than a general sense that something is off, means capacity problems get capacity, structural problems get redesign, and behavioural problems get coaching. Getting that match wrong is where most transformation effort is wasted.

Read across a whole organisation, the same data answers three questions: where collaboration flows or breaks down, which recurring behavioural patterns limit performance, and where an intervention has the greatest leverage on the system as a whole. None of this requires new systems or a long programme. The measurement is a short survey in which teams rate the interfaces they work across; the picture it produces is available in weeks; and the work that follows is scoped to the interfaces that need it. We structure it as Diagnose, Design, Develop and Sustain, but the sequence matters more than the labels: see the system accurately, agree what needs to change, do the specific work required, and confirm it held.

Conclusion

When a strategic programme stalls, the usual conclusions are that the strategy was flawed or that the business cannot deliver it. In the engagements we have run, both have been premature. The strategy was sound and the people were capable; what was missing was visibility of the system through which execution has to pass. The cost of that missing visibility is measured in the months it takes for a good strategy to become an outdated one.

If a programme of yours is stalling and you cannot name why, that is usually the symptom this measures. Start by making collaboration visible. Get in touch and we will talk it through.