Unlocking Organisational Intelligence
from Communication Data

What can your data tell you?

Most organisations hold years of communication data — emails, chats, collaboration logs — and extract almost no strategic value from it. Social Network Analysis changes that. By modelling individuals as nodes and their interactions as weighted, directed edges, we surface the hidden structure of how your organisation actually works — not just how it appears on an org chart.

Visualising the Organisation as a Network

Each node represents a person. Each edge is a communication channel. Node size reflects overall influence (PageRank), and colour indicates which natural community the algorithm assigned each person to — without any prior knowledge of the org chart.

Figure 1 — Communication Network Graph. 6 distinct communities emerge automatically from email patterns alone. Labelled nodes (largest circles) are the most influential individuals. Tightly clustered groups correspond to teams; nodes bridging clusters are critical information brokers.

When Does Your Organisation Communicate?

Timestamps on every message let us analyse not just who talks to whom, but when. This reveals work rhythms, crunch periods, after-hours activity, and potential early burnout signals.

Figure 2 — Weekly email volume, 2024-01-01 to 2026-03-13. The dashed line shows the 4-week rolling average, revealing long-term activity trends. Spikes may correspond to project deadlines, conference seasons, or grant submissions. Sustained low-volume periods can indicate seasonal slowdowns or disengagement.

Figure 3 — Email volume by hour of day and day of week (local time). Darker cells = higher communication volume. The concentration on weekday business hours (approx. 08:00–18:00) confirms typical academic office patterns. Weekend and late-night activity would surface here immediately if present.

Who Are the Real Leaders?

Beyond the org chart, SNA quantifies actual influence. We combine four independent metrics — PageRank, betweenness centrality, in-degree, and reciprocity — into a composite influence score, then assign each person a functional network role.

Figure 4 — Influence vs. Brokerage. X-axis: PageRank (how influential). Y-axis: Betweenness (how much of a broker). Bubble size: in-degree (popularity). Top-right nodes are both influential and structurally critical — removing them disrupts the entire network.

Figure 5 — Multi-dimensional centrality for top 15 nodes. Four independent metrics side by side. A node scoring high on all four is a true organisational linchpin. Diverging scores reveal specialised roles (e.g., high betweenness but low in-degree = a quiet broker).

Top 10 Most Influential Individuals

How Do Teams Collaborate?

Community detection reveals organic groupings that emerge from behaviour, not hierarchy. Comparing internal vs. cross-department communication per community shows which teams are siloed and which are highly collaborative. A modularity of 0.4303 indicates strong community structure — teams are clearly distinct with limited cross-boundary traffic.

Figure 6 — Internal vs cross-department email volume per community. Blue bars = emails within the community. Green bars = emails to/from other communities. A dominant green bar indicates high cross-team collaboration. A dominant blue bar may signal a self-sufficient team — or a silo.

What Can You Do With This?

The analyses demonstrated here apply directly to a wide range of real-world organisational challenges. With richer, labelled customer data, each use case becomes a concrete, decision-ready product delivered on your data.

How Was This Built?

This case study was produced entirely from a raw edge-list file with three columns: sender, receiver, timestamp. No labels, no names, no metadata. Everything you see was derived from communication structure alone.