Measuring community health

Measure community health by asking whether members get useful help, whether contributions are shared fairly, and whether people return when they have a reason to. Registration counts accounts and traffic counts visits; neither establishes those outcomes. Start with a few measures tied to the community’s purpose, then read the conversations behind the figures.

Define a healthy exchange

Write down the main job members come to do. In a customer forum, that may be getting an answer and knowing what to try next; in a professional group, it may be comparing approaches with relevant peers. A quiet member who finds a useful answer can benefit without posting, while a lively thread can leave its original question unresolved.

Agree on an observation period and a definition for each measure before reviewing a dashboard. State who is included; staff, test accounts, guests and people who no longer have access may need different treatment. Keep the definitions beside each report so a change in the counting rule does not look like a change in the community.

Use a balanced scorecard

Keep counts beside percentages, especially in a small group where one conversation can move a rate sharply.

Discourse admin reports include staff activity and an accepted-solutions report; they also report anonymous pageviews. GitHub Discussions counts reactions and votes among actions in its contributor insights. These measures do not, by themselves, show whether someone received useful help. Check definitions, permissions and available features for the platform and plan in use.

Read outcomes as well as activity

Review a manageable sample of questions and discussions. Include unanswered threads and conversations that attracted little attention. For each, note the member’s task, the response, any follow-up and whether the outcome is confirmed, indicated by the exchange or unknown.

A selected answer can help readers find a reply. It reflects what someone marked as a solution at that time; it does not establish that the advice fits every reader or remains current. Where a member does not confirm the result, do not treat a helpful-looking reply as a confirmed success.

Separate member replies from staff replies. Notice whether newcomers can ask questions, experienced members have room to contribute, and a few people do most of the answering. A higher post count could come from useful exchanges, repeated greetings or one unresolved argument; the count cannot tell you which.

Compare like with like

Choose a period that suits the community’s pace. To assess return, define who entered the cohort and follow those same eligible members through a stated later window. Dividing this month’s returning visitors by this month’s new registrations does not produce a return rate for one cohort.

Consider why someone joined before interpreting absence. A person seeking a one-off answer may be well served and have no reason to revisit next week. Someone in an ongoing peer group may reasonably return as their work changes.

Record changes to access rules, reporting, prompts and staffing. A shift in the numbers then merits investigation; it does not prove the change caused it. Analytics tools can also define users and return events differently, so compare only figures built from compatible rules.

Interpret platform counts in context

GitHub Discussions insights cover more than discussion activity: the contribution total combines contributions to discussions, issues and pull requests. Page views are split between logged-in and anonymous viewers, so neither figure alone describes whether members found useful help in discussions.

Access also affects who can inspect the evidence. Repository administrators and people with maintain access can view the GitHub Discussions insights dashboard. Its period selector offers 30 days, 3 months or 1 year; hovering over a period in the graph shows the exact data for it.

In Discourse, only staff can access built-in admin dashboard reports. That makes the reports useful for an internal review, but not a view every member can inspect directly. Consider who needs access to the review findings when deciding how to share them.

Platform Differences in Community Health Metrics

Interpret answer markers carefully

For how to code and interpret question-level answer markers, see the supporting article; a platform marker alone does not confirm a useful outcome.

Turn the review into a decision

Choose the member problem that needs action. If relevant questions are hard to find, improve the route to them. If they remain unanswered, assign response ownership.

If answers arrive but members cannot use them, check clarity and scope. If useful work depends on one volunteer, address that reliance before expanding activity.

Record the decision, its owner and the evidence that would change it. Recheck the affected conversations after members have had time to use the change.

When the review depends on platform reports, check who can see them and what their reporting window represents. Keep the chosen window consistent when comparing reviews.