Response rateResponses divided by triggers, times 100. Ten trigger messages with seven replies inside the hour is 70%. It shows coverage at a glance and catches the conversations that slipped through.
Average response timeYour mean time to first response across the selected accounts, shown alongside the conversation count it's averaged over.
What it measuresEvery incoming private message is a trigger; in groups, only @mentions and replies to your messages trigger the response clock.
Per-account or per-seat comparisonIn workspace mode, group the same metrics by Telegram account or by seat and sort the table to find weak coverage or a slow average.
Single account or rollupMeasure one inbox, a handful of accounts, or your whole team. The response figures recompute for whatever selection you pick.
Always currentAs new messages sync in the background, response time and reply rate update themselves. There's nothing to refresh or recalculate.
Response Performance cardResponse rate, average response time, the median, and how many conversations were responded to versus not. It labels replies as matched within 1h and compares against the previous period.
Here is the rule in one worked example. A trigger arrives at 10:00. If you reply at 10:15, the response counts, and 15 minutes goes into the average. If you reply at 11:05 instead, the reply still happened, but at 1 hour 5 minutes it falls outside the window: it does not count toward response metrics, the trigger lowers your response rate, and the average time is untouched. Response time is simply the moment your reply was sent minus the moment the trigger arrived, reported in minutes.
The asymmetry is deliberate. If an unanswered trigger were folded into the average as an infinitely long wait, one forgotten message would wreck the number and it would stop being useful. Keeping unanswered triggers in the rate, and out of the time, means each metric answers one question cleanly.
Segment before you compare people. Someone who lives in group chats has a structurally different trigger profile from someone doing one-to-one outreach, because in groups only mentions and replies count. Use the Chat type filter (All, Users, Groups, Channels) so you compare like with like. Also expect the rate to fall in a busy week: the same number of replies against more triggers is a lower percentage, so check the trigger count before calling it a slowdown.
The one-hour window is fixed, and it is the wrong tool for slower service promises. If your business commits to answering by the next business day, put the conversation on a ticket and use a ticket SLA, which carries its own deadline in hours. Response metrics stay the measure of live conversation, SLA the measure of a promise.
Where the numbers appear in Entergram: the Response Rate and Avg Response Time stat cards on the Analytics overview, the Response Performance card, and the Resp. rate and Avg 1st reply columns in the message breakdown, which you can pivot by Telegram account. The default time range is 30 days. You can switch to Today, 3 days, 7 days, 1 year, All time or a custom range.
A fixed report answers the questions someone anticipated in advance. The rest you ask directly. The same message events behind these numbers reach Claude, ChatGPT or any other MCP client through the Entergram connector, each event carrying an account, a chat, a timestamp and a direction, across every account on your seat. An AI client can work out from that whether first-reply time slipped only on Mondays, or whether it got worse after one teammate picked up a set of chats. Neither question has a button in the dashboard. When the answer belongs in a client report or a spreadsheet instead, the raw analytics data downloads.
