Organizational network analysis
Definition
Organizational network analysis
Organizational network analysis (ONA) maps who actually works with whom inside a business, using email, calendar, chat or survey data to show the real routes work takes. Its value is that the map rarely matches the reporting chart. That gap is the finding.
ONA is not the organisation chart and it is not a culture survey. The chart says who reports to whom; the analysis says who people actually ask when they need an answer today.
The findings are usually uncomfortable. A few people turn out to carry most of the traffic, whole teams turn out to be talking mainly to themselves — and neither shows up in a structure review.
Most of what you map is informal communication: the quick message, the standing favour, the colleague everyone asks first because they always answer.
Key takeaways
- ONA maps real working ties, not reporting lines.
- Brokers and bottlenecks are often the same people, carrying the same risk.
- Consent, aggregation and a published purpose decide whether the exercise survives.
- An isolated team is a design finding, not a motivation problem.
How it works
Organizational network analysis works in three moves: collect tie data from surveys or system logs, build a graph of people and the links between them, then read the shape for brokers, clusters and isolated groups.
| Measure | What it shows | Why it matters |
|---|---|---|
| Degree | How many people someone works with | Spots overloaded hubs |
| Betweenness | Who sits on the path between groups | Finds a single point of failure |
| Cluster | Groups that mostly talk among themselves | Shows where handovers stall |
| Reciprocity | Whether ties run both ways | Separates broadcasting from working |
| Reach | Steps needed to cross the organisation | Tests whether the layers are helping |
Survey based analysis asks a short question set: who do you go to for advice, who do you need in order to finish your work, who do you rarely hear from. The answers build the same graph with consent attached.
Passive data is faster and riskier. Metadata already sits inside your collaboration platform strategy, which is exactly why consent and aggregation rules have to be settled before the first extract runs.
Reading a business this way is a recognised trade. Among the duties the US Bureau of Labor Statistics lists for management analysts is interviewing personnel and conducting onsite observations, and it counted 1,077,100 of them in 2025.
A hub is not automatically a hero. The person on every path may be helping, or may be a gate nobody has dared route around — and the measures cannot tell those two apart without interviews.
Most of the value sits in cross-functional collaboration. The analysis shows whether two teams meant to work together exchange anything at all, or whether every exchange routes through one manager who is quietly rationing it.
The output belongs in your internal communications strategy, not in a performance file. If people believe the graph will rank them, the next survey gets answered strategically and the map is worthless.
Somebody has to own the result. In most firms that is the chief people officer, because the findings touch workload, pay bands and promotion, and all three need one accountable owner.
Government sets a comparable expectation in regulation. Under 5 U.S.C. 1103(c), the US Office of Personnel Management must design systems, including appropriate metrics, for assessing human capital management across federal agencies.
Examples
The exercise earns its cost when it finds something a chart cannot show. Four patterns come up often enough in outsourcing and shared service work to be worth naming before you commission one.
A bank integrating an acquired business maps ties six months after close. The two sides still barely exchange anything, and the few ties that do exist all run through three people who joined long before the deal.
An outsourcing provider maps a 400 seat delivery account. Client contacts talk to one operations manager and nobody else, so the account has no cover at all if that person resigns on a Friday afternoon.
A shared service centre maps ties between its Manila floor and a client team in Sydney. Half the client’s requests arrive outside the overlap hours, which is a scheduling finding dressed up as a service complaint.
A manufacturer maps engineering and quality after a recall. The two functions meet only in a monthly forum — which explains why a known defect took weeks to reach the people who could have fixed it.
Related terms
- High performing team framework: the standard a team is held to, once the map shows who the team really is.
- Cross functional team charter: the written remit that closes a gap the analysis exposed.
- Cross-functional collaboration: the behaviour the map measures rather than assumes.
- Informal communication: the unofficial channels carrying most of the ties you will find.
- Chief people officer: the usual owner of both the findings and the consent rules.
FAQ
What is organizational network analysis in simple terms?
It is a map of who works with whom, drawn from real interaction data rather than the reporting chart. The distance between the two maps is usually the finding worth acting on.
Does it need email and chat data?
No. Survey based analysis works well at team level and avoids most privacy objections, although it costs more time to run and has to be repeated to show movement. Many firms start there and move to system data only once the governance is agreed.
Is this legal?
It depends where you operate and how the data is handled. Aggregate the output, publish the purpose before you collect anything, and take advice before touching message content rather than metadata.
How is it different from an organisational chart?
The chart shows authority, the analysis shows traffic — and a manager with three direct reports can carry far more working ties than a director with forty.
How often should it be repeated?
Run it before and after a structural change rather than on a routine cycle. One map is a snapshot, and comparing two tells you whether the change did what it was meant to.
Find delivery partners whose team structures you can actually inspect through the research on Outsource Accelerator.
Independent
