Organizational Network Analysis (ONA) for HR

Ask a CEO how work gets done and you’ll hear a story about the org chart. Ask the org chart, and it will mislead you. It shows who reports to whom, not who actually solves problems together. Organizational network analysis is how HR and people teams see the difference: a data-driven map of the informal network that the formal structure hides.

This guide explains what organizational network analysis (ONA) is, how it works, the roles it reveals, and the concrete decisions HR can make with it, from grounding feedback in reality to catching flight risks before they resign.

What Is Organizational Network Analysis (ONA)?

Organizational network analysis (ONA) measures and maps how people actually collaborate, based on the strength, frequency, and nature of their interactions. Instead of the formal reporting structure, it reveals the informal network: who exchanges information, who works across teams, and who the organization quietly depends on.

ONA grows out of social network analysis, and its leading practitioner is researcher Rob Cross, founder of the Connected Commons, who has studied collaboration networks for more than two decades. His definition is precise: ONA “measures and graphs patterns of collaboration by examining the strength, frequency and nature of interactions between people in networks.”

The point is not to draw a prettier chart. It is to make visible the coordination that drives output, so leaders can support it instead of guessing at it.

Why the Org Chart Hides How Work Really Flows

The org chart captures formal structure, not functional reality. It shows reporting lines but not the cross-functional partnerships where most work happens: a designer helping an engineer, a rep pairing with marketing on a customer story. Those connections form and dissolve constantly, and none of them appear on the chart.

Sometimes cross-functional groups are created on purpose. More often they emerge on their own, and that improvisation is where both innovation and duplicated work live. If you plan a reorg or a launch off the chart alone, you are steering by a map that was never accurate.

The gap is measurable. Rob Cross found less than 50% overlap between the people who are actually central connectors in an organization and the names on its official top-talent list. In other words, half the people holding the network together aren’t the ones leadership is watching. We’ve written more about why the org chart is a lie and why your org is really a graph.

How ONA Works: Nodes, Edges, and Weight

ONA models an organization as a graph. People are the nodes, and their working relationships are the edges. Each edge carries a weight based on how much two people actually interact: meetings attended together, messages exchanged, pull requests reviewed, documents co-edited. Heavier edges mean stronger collaboration.

Modern ONA builds this from the digital breadcrumbs work already leaves behind, with no forms to fill out:

• Engineers reviewing each other’s code in GitHub
• Designers commenting on the same files in Figma
• PMs writing and editing shared docs
• Reps joining calls with support and success
• Teams messaging in Slack and meeting on calendars

The hard part is weighting. A 1:1 clearly signals a strong tie. A 28-person all-hands mostly signals attendance. A six-person design review sits in between, with two people deep in discussion and the rest observing. Distinguishing genuine contribution from mere presence is the core technical challenge, and richer meeting metadata keeps making it more tractable.

The Roles ONA Reveals

ONA surfaces roles the org chart cannot name. Network math turns them from anecdote into measurement, using centrality scores that quantify each person’s position in the flow of work.

• Super-connectors (high centrality): People who move information across many teams. Eigenvector centrality, essentially PageRank for people, weighs not just how many others you work with but how connected those people are.
• Brokers (high betweenness): People who sit on the shortest path between groups. If a broker leaves, information takes a detour and teams can quietly lose touch.
• Cross-team linkers: People who stitch together parts of the organization that would otherwise fragment into silos.
• Isolates: People with few, deep ties who do focused individual work. Not a problem by default, but a signal worth interpreting.

These roles concentrate more than most leaders expect. Cross’s research across hundreds of organizations shows that 3-5% of people account for 20-35% of the value-adding collaborations, while a separate 15-20% are collaboratively overloaded and at risk of burnout.

What HR Can Actually Do With ONA

For HR, ONA turns collaboration data into decisions. It grounds feedback in real working relationships, flags flight risks before they resign, shows whether new hires are integrating on schedule, and gives reorg decisions an evidence base. Each use rests on the same underlying map of who works with whom.

• Ground reviews and feedback in reality. Most tools ask people to nominate their own reviewers, so they pick friends. ONA suggests reviewers based on recent, actual collaboration, which makes continuous feedback and performance reviews fairer and easier to give.
• Catch flight risk early. Someone drifting to the periphery, or tied closely to an overloaded colleague, is more likely to leave. Cross found attrition up to 200% higher among people connected to an overwhelmed person, and academic work has used network position to predict voluntary turnover.
• Reduce key-person risk. Identify the brokers holding critical connections so a single resignation doesn’t sever how two teams communicate.
• Track onboarding. Measure how fast new hires build ties relative to their start date, normalized against peers, so you can spot someone integrating too slowly and intervene.
• Manage collaboration overload. Cross estimates organizations can recover 18-24% of wasted collaborative effort by rebalancing where the load falls.

ONA Without the Survey Overhead

Traditional ONA relied on surveys asking employees to name their collaborators. Those surveys are slow, biased toward friendships, and stale the moment they’re collected. Passive ONA reads the trail work already leaves, so the map stays current automatically and reflects behavior rather than self-report.

That shift matters for accuracy and for trust. Because passive ONA runs on systems people already use, it captures recent collaboration without adding a single form to anyone’s week. Deloitte frames ONA as a way to prioritize human-centric metrics and improve the worker experience, not to surveil it.

Two guardrails keep it trustworthy. First, weighting has to reflect real contribution, not just attendance. Second, the network has to be permission-aware: it should never surface a connection or a piece of content to someone who couldn’t already see it in the source tool.

How Windmill Builds ONA Into the Context Graph

Windmill runs ONA continuously as the People layer of its context graph for your people. Its AI agent, Windy, derives the real collaboration network from the tools where work already happens, then uses it to request feedback from genuine collaborators and to power reviews, 1:1s, and calibrations that reflect how the organization actually operates.

ONA is the foundation the rest of the graph sits on, which is why we treat it as core science rather than a dashboard feature. Once the network exists and stays current, new workflows open up: staffing decisions, succession planning, coaching recommendations, and organizational diagnostics all draw from the same living map.

The direction this heads is design by simulation: a network view that evolves week to week, and eventually the ability to model a reorg before you run it. Running a great company is mostly coordination, and coordination starts with actually knowing how your people work together. See how ONA fits the broader shift toward an HR context graph and year-round performance signals.

Frequently Asked Questions

What is organizational network analysis (ONA)?

Organizational network analysis (ONA) measures and maps how people actually collaborate, based on the strength, frequency, and nature of their interactions. Instead of the formal reporting structure, it reveals the informal network: who exchanges information, who works across teams, and who the organization quietly depends on to get work done.

How is ONA different from an org chart?

An org chart shows formal structure, who reports to whom. ONA shows functional reality, who works with whom. Most real work happens through cross-functional partnerships that never appear on the chart. Rob Cross's research found less than 50% overlap between an organization's central connectors and its official top-talent list.

What data does ONA use, and do you need surveys?

Traditional ONA used surveys asking employees to name collaborators. Modern passive ONA reads the digital trail work already leaves: meetings, Slack messages, code reviews, shared documents, and calendar activity. Passive ONA stays current automatically and reflects real behavior instead of self-report, which tends to favor friendships over working relationships.

Can ONA predict employee turnover?

ONA surfaces early signals of attrition risk. Employees who are disconnecting from the network, or who sit on its periphery, are more likely to leave, and research has operationalized network position to predict voluntary turnover. Rob Cross also found attrition rates up to 200% higher among people connected to an overloaded colleague.

How does Windmill use organizational network analysis?

Windmill runs ONA continuously as the People layer of its context graph. Its AI agent, Windy, derives the real collaboration network from the tools where work already happens, then uses it to request feedback from genuine collaborators and to power reviews, 1:1s, and calibrations grounded in how the organization actually operates.