Article
Network analytics on corporate e-mail
Who e-mails whom – and what does that reveal about hidden influencers in the organisation?
Executive Summary
Modern organisations run on e-mail. But who actually influences information flow? Which people sit at the centre of cross-team communication, and where do we see potential bottlenecks or single points of failure?
In this case study we use a large anonymised corporate e-mail network (the EuAll SNAP dataset) to answer a simple but powerful question:
“Who e-mails whom – and what does that reveal about hidden influencers in the organisation?”
Data & Methods
The dataset is stored in Matrix Market (.mtx) format as a large, sparse adjacency matrix:
265,214 nodes – anonymised e-mail accounts inside a European institution.
~419,000 directed edges – each edge represents at least one e-mail from user i to user j.
There are no timestamps or message content; we only use the who-e-mailed-whom pattern.
The pipeline implements several complementary views:
Global degree distribution (log–log histogram).
Top in-degree and top out-degree rankings
In- vs. out-degree scatter plot (log–log).
Circular “chord-style” diagram for the top influencers.
We select the top 12 nodes by total degree and draw them on a circle. Directed links between them show who e-mails whom inside this elite core. Visually this mirrors a chord diagram: thick bundles of edges indicate strong mutual communication between key influencers; isolated spokes flag individuals who are highly connected to the wider organisation but less connected among themselves.
Together these views move from global structure to a focused lens on the informal leadership circle.
Results
Operational risk & continuity
When 0.1–1% of mailboxes carry over half of all email traffic, the organisation is structurally dependent on a small group of individuals or shared inboxes.
These addresses are single points of failure: illness, turnover, or access problems can disrupt large parts of the communication flow.
Recommendation: treat high-degree mailboxes as critical infrastructure – set up backup owners, shared access, and clear escalation paths.
Process mapping through communication patterns
The roles inferred from in- vs out-degree help map informal processes:
High-in / low-out nodes → intake & escalation points (customer support, reporting mailboxes).
High-out / low-in nodes → broadcast channels (internal newsletters, HR announcements).
High-in / high-out nodes → coordinators / project managers.
Even without knowing job titles, management gains a high-level picture of how work actually flows and whether this matches the formal org chart.
Targeted compliance and awareness programmes
The same high-degree mailboxes are also where phishing, data-loss, and compliance breaches are most likely to have impact.
Instead of a one-size-fits-all security programme, the network view allows:
Prioritised training and monitoring for key influencers and hubs.
Focused anomaly detection on these nodes (sudden spikes, unusual contacts, off-hours behaviour).
Change-management and communication planning
For major strategic announcements or change programmes, the identified hubs are the natural amplifiers:
they already have trust and visibility in the network,
and their out-degree indicates real reach.
Communication and HR teams can use the list of top hubs as a “network of champions” to involve early, brief more deeply, and equip with tailored materials.