A government R&D organisation knew its innovation output was a fraction of what it could be — but had no structural explanation for why. Network analysis answered that question. In doing so it revealed two further things nobody had thought to ask about, one of which was about to cause significant structural damage.
The organisation's sole purpose was generating and applying new knowledge. By headcount and resourcing it was equipped for the task — but leadership believed innovation output was significantly below what 1,100 researchers should be producing, with no structural explanation for the gap.
At the same time, a restructure was being planned: consolidate the research support specialists, who were distributed across research teams, into a single central unit. The rationale was operational — reduce duplication, simplify reporting lines.
HumanDynamics was engaged to map the actual collaboration structure to answer the innovation question. The analysis produced three findings. The first answered the brief. The second and third were things the network revealed that nobody had asked about.
"The formal structure had nothing to do with how innovation actually happened here. The network told a completely different story."
HumanDynamics structural analysis findingAcross 1,100 researchers, a small cluster of teams occupied the bridging positions through which the majority of cross-organisational collaboration flowed. Their structural importance was visible only from the actual pattern of communication across the organisation — not from the hierarchy, formal titles, or management's understanding of how the organisation worked.
Most research teams collaborated effectively within their own group. Cross-team collaboration — the mechanism through which the most significant innovative outputs were generated — was far more concentrated than leadership assumed. A small number of teams were acting as connective tissue between otherwise separate research groups: those through whom ideas crossed boundaries, methods were shared, and collaborative projects were initiated. From the org chart they were unremarkable. Their structural centrality was invisible until the network was mapped.
The constraint was architectural, not cultural. Innovation was concentrated in a small number of structural positions, and most of the organisation had limited access to it. Increasing innovation output became, structurally, a question of increasing the reach and resilience of those positions — not of motivating the broader research population.
For the first time, leadership had a structural map of where innovation was actually happening. Decisions about team design, resource allocation, and organisational development could be grounded in where collaboration was concentrated, where it was absent, and which structural positions were sustaining it — rather than in org chart assumptions or management perception.
The evidence also clarified what not to do: any redesign that disrupted the team positions identified as central risked dismantling the mechanism the organisation existed to operate.
In a portfolio company this finding surfaces consistently: the teams actually driving execution are rarely those identified as doing so on the org chart. Structural mapping before a 100-day plan or VCP workstream is designed identifies where real capability is concentrated — ensuring decisions are built around evidence, not assumed importance.
In answering the innovation question, the network also surfaced the structural role of research support specialists embedded across research units. They were functioning as informal bridges between research groups with no other natural collaboration pathway. When HumanDynamics learned that centralisation of this function was planned, the implication was immediate: the restructure would sever those bridges — reducing the cross-team connectivity that was already the organisation's primary innovation constraint.
Research support specialists were embedded within research teams as an operational convenience, not a strategic design choice. But embedding had a structural consequence: each specialist connected their research team to others through informal relationships built in daily collaborative work. A researcher engaging a specialist who also supported another team gained indirect access to methods, problems, and people they would not otherwise encounter. The function was undervalued — treated as a support cost. The network showed it was the informal connective tissue between research groups. Its distributed presence was not inefficiency. It was infrastructure.
Consolidating the research support specialists would have preserved the group's internal cohesion while eliminating their external connectivity. Each specialist would have moved from being embedded in the daily fabric of a research team to sitting in a separate unit — no longer part of the informal network of any research group. The cross-team bridges would have quietly disappeared. At the precise moment the organisation was trying to increase innovation output, a planned operational efficiency measure would have reduced the structural conditions making it possible.
Post-acquisition integrations routinely consolidate embedded support functions on the basis of reducing duplication. Where those functions are distributed across business units, their role as cross-unit connectors is almost never assessed. The damage surfaces in execution performance months later, with the cause no longer traceable to the restructure that created it.
The organisation's managers were not embedded in the collaboration flows through which innovation was being generated. They were peripheral to those networks — receiving limited signal from the teams doing the structural work, and poorly connected to one another. The result: a management layer that could not see where innovation was concentrated, could not coordinate across functions, and had no shared view of how the organisation was actually operating.
The research teams identified in Finding 01 as structurally central were not the teams with the most management attention or resource. Managers were connected to the parts of the organisation least involved in the cross-team collaboration driving output — and had limited visibility of the teams most involved. Management decisions about resource, priority, and team design were being made without structural sight of where innovation capacity actually sat. Innovation was happening despite management, not because of it — not through any failure of intent, but because the structural information needed to direct it was not reaching those with authority to act.
The managers were also poorly connected to one another. The network showed limited collaboration and information flow across the management layer — meaning that even where individual managers had visibility of specific dynamics, that knowledge was not being shared or acted on collectively. The combined effect was a management group with neither the structural information nor the internal connectivity to drive innovation in a coordinated way.
Management isolation is a consistent finding in post-acquisition and transformation contexts: the leadership layer is structurally peripheral to the execution activity they are responsible for directing. Operating partners typically discover this when VCP workstreams stall and management cannot explain why. Structural mapping identifies the disconnect before the programme is designed around the assumption that management has sight of the organisation they are leading.
Three findings from a single analysis. Each changed a different set of decisions.
| Decision area | Without structural analysis | With HD structural analysis |
|---|---|---|
| Finding 01 — Innovation concentration | ||
| Innovation strategy | Output gap attributed to broad underperformance. Interventions designed for the whole organisation equally, without structural targeting. | Structural concentration identified. A small number of team positions were enabling the majority of cross-organisational innovation. Interventions targeted to structural position, not headcount. |
| Redesign decisions | No mechanism to identify which positions were structurally critical. Decisions based on formal seniority and stated priority. | The positions sustaining cross-team innovation identified before any redesign. Structural dependency quantified — ensuring the redesign preserved rather than disrupted them. |
| Finding 02 — Centralisation risk | ||
| Restructure design | Research support specialists consolidated on operational grounds. Embedded positions treated as duplication to be eliminated. | Embedded specialist positions identified as structural bridges between research clusters. Centralisation redesigned to preserve those positions where cross-team pathways were at risk. |
| Structural risk | Impact of centralisation on cross-team collaboration not assessed. Risk invisible until performance declined post-restructure. | Structural consequence identified before the restructure was executed — and before it compounded the innovation constraint already present in Finding 01. |
| Finding 03 — Management isolation | ||
| Management effectiveness | Management assumed to have operational visibility. Output gaps attributed to researcher performance rather than management information failure. | Structural isolation of managers from innovation activity identified. Development directed at the structural disconnect rather than individual performance. |
| Coordination design | Poor connectivity across the management layer not visible. Cross-functional coordination assumed rather than evidenced. | Absence of cross-management collaboration identified structurally. Redesign built the management connectivity that was absent — not assumed present. |
The structural evidence changes the decisions. The conversation starts here.
Request a Consultation