Metric Definition
Mapping how people interact in meetings
Track from
Participant network analysis
Participant network analysis is the study of who interacts with whom across meetings, treating each participant as a node and each exchange as a connection. It surfaces collaboration patterns, isolated members, and the people who hold conversations together. It turns raw attendance and speaking data into a picture of how information actually flows through a team.
7 min read
What is participant network analysis?
Participant network analysis is the practice of mapping the interactions between people in meetings as a network, where each person is a node and each direct exchange is a connection. It answers a question that attendance lists cannot: not who was in the room, but who actually engaged with whom. A meeting of eight people where two dominate and six stay silent looks very different on paper to one where conversation flows across the whole group.
The analysis draws on signals like who responded to whom, who was addressed by name, who asked and who answered, and who shared follow-up work. From these signals it builds a graph. Some people sit at the centre with many connections. Others sit at the edge with one or two. A few act as bridges, linking otherwise separate clusters of the team.
The value is in the pattern. A healthy collaboration network spreads connections widely and has few isolated nodes. A fragile one concentrates connections on one or two people, which creates a single point of failure when those people are away or overloaded.
Read it carefully
Participant network analysis describes interaction, not contribution quality. A person with many connections is well connected, not necessarily more valuable. Read the network alongside outcomes, not as a ranking of people. Treating connection counts as a performance score creates pressure to talk for the sake of appearing engaged.
How to calculate participant network analysis
The headline measure is network density: the share of all possible connections that actually exist. With eight participants there are 28 possible pairs. If only 10 pairs interacted directly, density is 10 divided by 28, or roughly 0.36. A density near 1 means everyone engaged with everyone. A density near 0 means interaction ran through a tiny number of people.
Density alone is not enough. The same density can describe a balanced web or a star with one person at the centre. To tell them apart you need node-level measures, which describe each person rather than the whole network.
- 1
Participants (n)
The count of distinct people in the network for the period under review. Decide early whether observers who never speak are included, as this changes density.
- 2
Actual connections
Each unique pair of people who directly interacted, counted once. A direct interaction is a response, a question and answer, a hand-off, or being addressed by name.
- 3
Degree per node
The number of distinct people each participant connected to. High-degree nodes are central. Zero-degree nodes are isolated and worth following up on.
- 4
Betweenness per node
How often a person sits on the shortest path between two others. High betweenness marks the bridges who hold separate groups together and become bottlenecks when absent.
Participant network analysis in a metric tree
A single density number tells you the network is thin or rich, but not why or what to do. A metric tree breaks the network into the drivers that move it, so a flat or concentrated network points to a specific action rather than a vague concern about collaboration.
Metric tree insight
When a network is too concentrated, the tree shows whether the cause is a few people dominating, separate clusters that never mix, or new joiners failing to connect. KPI Tree assigns RACI ownership to each branch, so the person accountable for cross-team mixing is not the same as the person accountable for onboarding new joiners, and each sees the node they can actually move.
Participant network analysis benchmarks
There is no universal target, because the right shape depends on the meeting. A daily stand-up is meant to be lean and will look sparse. A cross-functional planning session should look dense and mixed. The ranges below are a starting reference for recurring working meetings of five to fifteen people, not a rule for every format.
| Network signal | Concentrated | Balanced | Reading |
|---|---|---|---|
| Network density | Below 0.25 | 0.35 to 0.6 | Higher density means conversation spreads across the group rather than through a few people. |
| Top two share of connections | Above 60 percent | 30 to 45 percent | A high share means two people hold the meeting together and the team depends on them. |
| Silent attendees | Above 40 percent | 10 to 20 percent | A large silent share suggests the meeting is too big or the format excludes people. |
| Cross-team connections | Below 10 percent | 25 to 40 percent | Low cross-team mixing means departments are talking past each other, not with each other. |
How to improve participant network analysis
Improving the network is not about forcing everyone to speak equally. It is about removing the structural reasons that connections fail to form: meetings that are too large, formats that reward the loudest voice, and silos that keep teams apart. The actions below target the drivers in the tree.
Right-size the invite list
Large meetings drive density down and silent attendance up. Invite the people who need to interact and route the rest to notes. A smaller room makes it easier for every node to connect.
Rotate facilitation
When the same person always runs the meeting they accumulate most connections by default. Rotating the facilitator spreads the central role and reduces the single-point-of-failure risk in the network.
Draw in the edges
Identify zero-degree and low-degree participants and address them directly with a specific question. A named prompt is the most reliable way to turn a silent attendee into a connected one.
Mix the clusters
If the network splits into separate department clusters, seed cross-team pairings in the agenda. Assign joint follow-up work so connections form across the lines rather than within them.
Common mistakes when tracking participant network analysis
- 1
Treating connections as a scoreboard
Ranking people by connection count rewards talking over listening. It pushes people to interject for visibility and degrades the meeting. Use the network to fix structure, not to grade individuals.
- 2
Counting attendance as interaction
Being in the room is not a connection. If the analysis treats every attendee as connected to every other, density is meaningless. Only count direct exchanges.
- 3
Ignoring meeting type
Comparing a stand-up to a planning workshop on the same density target produces false alarms. Benchmark each meeting format against its own history, not against a single universal number.
- 4
Reading one meeting in isolation
Networks are noisy in any single session. A quiet member one week may carry the next. Track the trend over several meetings before drawing conclusions about isolation or dominance.
Related metrics
Action item distribution balance
Workload evenness across owners
Operations MetricsMetric Definition
Distribution Balance = 1 - (Items on Busiest Owner / Total Open Items)
Action item distribution balance is a measure of how evenly open action items are spread across the people accountable for them. It exposes hidden bottlenecks where one owner carries far more than their share. A balanced distribution keeps work moving and protects against single points of failure.
Employee turnover rate
Staff attrition
HR & People MetricsMetric Definition
Turnover Rate = (Separations / Average Headcount) × 100
Employee turnover rate measures the percentage of employees who leave an organisation during a given period. It is one of the most closely watched HR metrics because high turnover disrupts productivity, erodes institutional knowledge, and drives up recruitment and training costs.
Sprint velocity
Agile planning metric
Operations MetricsMetric Definition
Sprint Velocity = Sum of Story Points Completed in a Sprint
Sprint velocity measures the amount of work a team completes during a sprint, typically expressed in story points, ideal days, or another unit of estimation. It is a planning tool that helps agile teams forecast how much work they can commit to in future sprints based on their historical completion rate. Velocity is one of the most widely used and most frequently misunderstood metrics in agile software development.
Metric decomposition
Metric Definition
Learn how to break participant network analysis down into the driver metrics that explain how meeting interaction patterns shift over time.
Metric trees for operations teams
Metric Definition
See where a meeting interaction metric like participant network analysis sits within the wider operations metric tree and which outcomes it influences.
Turn meeting interactions into an accountable network
Build participant network analysis as a metric tree in KPI Tree, with density, centralisation, and inclusion as branches and a RACI owner on each. When the network concentrates or isolates people, the accountable owner is notified and can act on the specific driver behind it.