Skip to main content
Org charts show who reports to whom. Organizational Health maps collaboration across your company using organizational network analysis (ONA), helping you identify central people, connectors between groups, and isolated clusters. Use it to understand questions that reporting lines alone cannot answer, such as which people connect teams and where collaboration is fragmented.

​
Who can see Organizational Health

Unlike the other Analytics reports, Organizational Health is only available to people who can see Analytics for the entire organization — Workspace Admins, HR Admins, and Data Analysts. Managers and individual contributors don’t see this report at all, even though they can open the other Analytics reports for their own scope. Because the network spans the whole company, there’s no partial, manager-limited view of it. You’ll find it in the Analytics section of the Windmill Dashboard.

​
How the network is built

The network is built from collaboration signals across the work tools you’ve connected to Windmill. The more two people interact, the stronger the connection Windmill draws between them. Windmill groups those signals into the following categories:
  • Meetings
  • Messages
  • Code reviews
  • Tasks
  • Documents
These categories appear in a person’s Top score source and Source mix metrics. The available signals depend on the integrations you’ve connected. The report reflects the current state of your collaboration network. There’s no date-range picker. Windmill tunes the graph automatically to surface the strongest, most meaningful connections and keep the view readable. Relationships that don’t clear that threshold aren’t shown. Use the employee filter at the top of the page to hide people, such as contractors, from the view. You can filter by group, job title, manager, start date, tenure, or name. The filter hides those people and their relationships. It does not recalculate the network, so the overview numbers still describe the whole network. Windmill picks the threshold so that the visible graph targets a healthy average number of connections per person. In very sparse networks the actual average can fall short of that target. Scores refresh overnight. If you’ve just connected a new integration, its collaboration signal shows up after the next daily refresh — not immediately.
Organizational Health surfaces collaboration patterns and strength — who works with whom, and how much. It doesn’t show the contents of anyone’s messages, emails, or documents.

​
What you’ll see

The report has three sections: the Network graph, the Key People lists, and the People table. A Guide button in the top right opens an in-app reference for every term used on the page.

​
Network graph

An interactive map of your collaboration network:
  • Each dot is a person.
  • Each line is a working relationship. Thicker lines mean stronger connections.
  • Larger dots have more or stronger connections — their size reflects total connection strength.
Click any dot to select that person; their direct connections highlight and the rest of the graph dims so you can see their immediate network. Click empty space to clear the selection. You can also use the Search people box above the graph to jump to someone by name. The graph can include unarchived employees who do not have Windmill member accounts. The Key People lists and People table are limited to active Windmill members. Use Color by to group people by Team or by Community, which Windmill detects from relationship patterns. Select a group in the legend to highlight it.
If you search for someone and see “Employee not in graph,” it means they don’t currently have relationships strong enough to appear in this view — not that there’s no data on them.

​
Network overview

When no one is selected, the panel beside the graph summarizes the whole network, including people the employee filter hides:
  • People — How many people are in the network.
  • Relationships — How many connections are shown. A stronger connection means more relationship signal between two people.
  • Density — How tightly connected the network is overall. Higher density means people are connected to more of the others.
  • Components — The number of separate groups in the network. One group means everyone is connected through some chain of relationships; more than one means there are clusters with no link between them.
  • Communities — Groups whose relationships are more concentrated within the group than outside it. Detected with the Louvain algorithm.
  • Modularity — How strongly the network separates into communities. Higher values indicate more distinct clusters.
  • Max core layer — The highest k-core layer present in the network. Anyone at this layer sits in the most tightly interconnected subgroup — each member has at least this many connections to other members.

​
Selecting a person

When you select someone, the panel switches to their profile: their name and manager, every people metric for that person, and a Strongest connections list — their top relationships ranked by connection strength (the score is a relative strength signal, not a percentage).

​
Key People

Three ranked lists that surface the people who play notable roles in the network:
  • Most connected — Active members with the highest Centrality across the company.
  • Critical connectors — Active members who link otherwise separate groups. These are often the people whose departure would fragment collaboration.
  • Cross-team linkers — Active members whose relationships reach most broadly across org-chart groups.
Each list ranks active Windmill members from the full company network.

​
People table

A sortable, per-person table of network metrics for every active Windmill member in the graph. Click any column header to sort, or use Filter People to find someone by name. Click any value to open a side panel that explains the metric and shows a top-10 leaderboard for it. By default, the table sorts by Strength (highest first). Click Export CSV to download the table. The file has one row for each active Windmill member in the table, sorted by Strength (highest first). It has the Name, Email, and Job title of each person and each people metric. Top score source has two columns: the source, and its share of the person’s signal as a percentage. Filter People does not change the export. The employee filter at the top of the page does: the file includes only the people who match it.

​
People metrics

These metrics describe each person’s role in the network. They appear in the People table, in the side-panel profile when you select someone in the graph, and in the in-app Guide.

​
FAQs

Only Workspace Admins, HR Admins, and Data Analysts — the roles with organization-wide Analytics access. Managers and individual contributors don’t see Organizational Health at all. Because the network spans the entire company, there’s no manager-scoped version of it.
No. Organizational Health measures how much and how broadly people collaborate — it’s built from activity signals across your connected work tools (messages exchanged, meetings attended together, code reviewed, shared projects). It doesn’t surface the contents of any message, email, or document.
Two reasons are common. First, the graph only shows relationships strong enough to clear Windmill’s readability threshold — someone with only weak or sparse collaboration signal won’t appear. Second, the Key People lists and People table include active Windmill members only. Searching for a person and seeing “Employee not in graph” means the former.
You can filter by employee, but you cannot change the date range. Use the employee filter to hide people by group, job title, manager, start date, tenure, or name. The report always shows the current state of your collaboration network, tuned automatically to surface the strongest connections.
The graph is built from your connected integrations. If the available sources do not provide enough collaboration signal, there may not be enough data to draw a meaningful network. Connect additional collaboration tools from the Integrations page to enrich it.Scores refresh overnight, so recently connected data sources will start appearing after the next daily refresh — not immediately.
Centrality measures how central someone is — being connected to other well-connected people raises it. Bridge measures how much someone links otherwise-separate groups. Someone can be a strong bridge without being highly central, or highly central without being a bridge.
Centrality is designed to give useful values even in fragmented networks where not everyone is reachable from everyone else. Each separate group gets its own set of well-formed scores, based on how well-connected people are within that group, rather than dropping to zero because the groups aren’t linked to each other.