Network Analysis

Every company publishes an org chart. Almost none of them are accurate.

The org chart shows who reports to whom. It says nothing about who actually reads whose messages first, who gets pulled into every cross-functional decision, or which engineer every product manager instinctively messages before writing a spec. The formal hierarchy is the map. The informal network is the territory.

What the graph reveals

BASAL builds a network from communications: who emails whom, who appears in whose meetings, who co-authors documents, who gets cc'd on escalations. Each edge carries weight by frequency, recency, directionality, and interaction type (information request, decision loop, social acknowledgment).

The network that emerges from this data is almost never the network on paper.

Take a 200-person engineering organization. The VP of Engineering appears at the center of the org chart. But the network graph might show that a senior engineer in infrastructure sits at the actual center of the communication topology: 40% of cross-team email threads route through her, she co-authors documents with six different product managers, and three team leads list her as a meeting participant more than their own managers. She is the real bridge between the backend platform team and the product organization — and she has no direct reports.

That is a bridge node. BASAL finds them automatically.

Influence vs authority

Centrality scoring measures five dimensions:

• Degree centrality: raw connection count
• Betweenness centrality: how often a person sits on the shortest path between two others
• Eigenvector centrality: whether your connections are themselves well-connected
• Information flow centrality: how much information passes through this person
• Temporal centrality: whether influence is rising, stable, or declining

Authority and influence diverge more than most leaders expect. People with high betweenness centrality but low authority are often your most critical single points of failure. If they leave, two departments stop talking to each other.

Community detection and information bottlenecks

BASAL applies graph clustering algorithms to detect communities: groups of people who communicate more with each other than with the rest of the organization. These clusters often map to informal teams, shared contexts, or coalitions that no org chart captures.

More useful: the gaps between clusters. When two communities barely overlap in the communication graph, information moves slowly between them. A product decision made in one cluster can sit unknown to a dependent team for weeks. BASAL surfaces these bottlenecks as edge-weight deficits: pairs of teams with low cross-cluster communication relative to their dependency footprint.

Network evolution

Run the analysis once and you get a snapshot. Run it monthly and you get drift detection: communities forming, bridge nodes burning out, new hires integrating (or not), relationships that went quiet after a reorg.

One team lead asked: "Why did the infrastructure migration take twice as long as planned?" The network evolution graph showed the answer. The two teams that needed to coordinate most closely had a betweenness score that collapsed three weeks into the project — the one engineer who bridged them moved to a new initiative. Nobody noticed until BASAL did.

The org chart doesn't update when that happens. The communication graph does.