Track community health

A developer community, a customer community, and a paid membership group measure success differently, so Kite agrees on the definitions with you before it counts anything. It reads your channel list and recent activity, then proposes which channels count as help channels, which are announcements, and which are social. That matters because response time only makes sense in channels where people ask questions, and an announcements channel where only staff can post would drag down every participation number.

Kite also proposes the time windows that suit your community. A busy Discord server with thousands of members might be measured weekly. A small Slack community for customers might be measured monthly, because a quiet week there means nothing. Once you agree, Kite saves the definitions in your team's knowledge base and uses them for every report, so the numbers stay comparable over time.

Then it builds a baseline. Kite reads back through the history it can reach and calculates each measurement for the months it covers. Your first report already shows a trend instead of a single number.

The clearest engagement measurement is the share of your members who actually take part. Kite counts the members who posted or replied in each period and compares that with your total membership. It also counts new conversations started, replies per conversation, and, in Discord, reactions, because a community where every post gets three replies is healthier than one with twice the posts and no conversation.

Kite looks at who is doing the talking, not just how much. If a fifth of your active members write most of the messages, Kite names that group as your core contributors and follows them separately. Losing two of them can matter more than losing fifty quiet members, and an overall activity number would hide it.

For Discord, Kite reads the channels and threads your bot can see and the member list, which includes the date each person joined. For Slack, Kite reads the channels it has been invited to, along with the replies inside each thread.

Response time is the measurement members feel most directly. Someone who asks a question and hears nothing for two days is unlikely to ask a second one.

Kite reads each message in your help channels and decides whether it's a question that needs an answer. That takes judgment, because "anyone else seeing this error?" is a question and "thanks, that fixed it" is not. For each question, it finds the first reply from someone other than the person who asked and records how long that took. From there, Kite reports the typical time to a first reply, the share of questions answered within a day, and the questions that never got a reply at all.

Kite also tracks who answers. When more questions are answered by members instead of your team, the community is starting to support itself, which is one of the strongest signs of health a community can show. When your team answers nearly everything, the community depends on a few employees' availability, and response times will slip whenever they're busy.

Because Kite reads the questions themselves, it can tell you where the delays come from. It might find that questions posted on weekends wait three times as long, or that questions about one feature keep going unanswered because nobody in the community knows it well.

Retention shows whether people who join find enough value to keep coming back. Kite groups members by the month they joined and follows each group forward, checking how many are still posting one, two, and three months later. This approach is called a cohort, and it shows whether the newest members are sticking around better or worse than earlier ones, which a single overall number can't tell you.

In Discord, the member list gives Kite each person's join date, so it can also see how many people join and never post at all. In Slack, Kite measures from each member's first message, because its Slack connection reads conversations rather than workspace join dates.

Kite pays separate attention to your core contributors. If someone who used to answer a dozen questions a month goes quiet, Kite flags it in the report, often before the drop shows up in the overall activity numbers.

Moderation activity tells you whether the community is staying a good place to be and whether your moderators can keep up. In Discord, Kite reads the server's audit log, which records bans, kicks, timeouts, and messages removed by moderators, along with your automatic moderation rules. Discord keeps the audit log for only 45 days, so Kite's monthly run saves each period's moderation record before it disappears. That gives you a history Discord itself doesn't keep.

Kite reports how many moderation actions happened, what kinds they were, and how many of your moderators took them. A rise in removed spam after a growth spurt is normal. A rise in bans tied to arguments in one channel, or one moderator handling nearly all the actions, is worth a conversation.

Kite's Slack connection reads messages rather than Slack's admin logs. If your Slack community keeps a moderation log channel where moderators note what they removed and why, Kite reads it and counts the entries.

Some problems can't wait for the monthly report. You can ask Kite to keep watch on a Slack channel, such as your help channel, and tell you when something needs attention. Kite checks the channel on a schedule, no more often than once an hour, and stays quiet until something matches, such as a question that's gone four hours without a reply or a thread that's turning heated.

For a Discord server, you can schedule a daily check that reads the help channels and sends you the list of unanswered questions, with a suggested reply for each one.

Leadership wants to know three things about a community: whether it's growing in a way that matters, why the numbers moved, and what you're doing about it. Kite's monthly report answers all three.

Kite builds the report with charts for each measurement and the comparison with the previous period. It explains the meaningful changes in plain language, such as "Questions answered by members rose from 30 percent to 45 percent after we started the weekly office hours." It publishes the report as a page with a shareable link, which you can limit to people at your company, and it can keep a dashboard at a stable address that it updates each month.

Kite leads the report with the measurements that show value, such as active members, response time, and retention, and keeps member count as context. That makes the report harder to dismiss, because it shows that people are getting help and coming back, not just signing up.

Measurement is only useful when it changes what you do next. Once Kite finds a problem, it proposes the work to fix it and does the parts it can:

• Unanswered questions. Kite drafts replies to open questions using what it knows about your product, and with your approval posts them in the thread.
• Falling activity. Kite writes a members' digest from the month's best conversations, which gives quieter members a reason to return.
• Rising moderation load. Kite writes community rules, a moderation policy, and an agreement for volunteer moderators, so more people can share the work.
• New members who never post. Kite drafts a welcome message and starter prompts for an introductions channel.

The next month's report shows whether the change worked. Kite records what it learned in your team's knowledge base, so its later suggestions build on what has helped your community before.