Dead channels have high ER
Subscribers leave, the loyal core stays and keeps reacting: the denominator falls faster than the numerator. A rising ER with a falling subscriber count is a warning, not a win.
Engagement rate
ER for a Telegram post is reactions plus forwards plus comments, divided by views, times 100. ERR is average views per post divided by subscribers — the share of the audience a post actually reaches. These are the formulas TGStat uses, and AwrAiter uses the same ones, so your numbers match what an advertiser checks. Below: a worked example, medians from real channels, and the trap of a tiny channel with a 40% ER.
01 — The formulas
(reactions + forwards + comments) ÷ views × 100. A post with 8 000 views, 120 reactions, 30 forwards and 10 comments: 160 ÷ 8 000 × 100 = 2.0%. This is the engagement of the people who actually saw the post.
The same numerator ÷ subscribers × 100. With 20 000 subscribers: 160 ÷ 20 000 × 100 = 0.8%. Use it to compare channels of a similar size; it punishes channels with low reach twice.
Average views per post ÷ subscribers × 100. 8 000 ÷ 20 000 × 100 = 40%. This is what an advertiser buys: how many subscribers a post reaches.
Channel-level ER is the median (or average) over the posts of the last 7, 30 or 90 days. AwrAiter shows the channel's ER for the period and scores every post against the channel's 30-day median, so one viral post does not move the baseline.
02 — Reading the number
Subscribers leave, the loyal core stays and keeps reacting: the denominator falls faster than the numerator. A rising ER with a falling subscriber count is a warning, not a win.
With 300 subscribers a single forward is a third of a percent. Compare inside a size bucket, never across.
ER answers “did the readers care?”, ERR answers “did the post reach anyone?”. A channel can win one and lose the other; report both.
Questions and polls collect reactions; links and long reads collect forwards. Look at the mix before judging a post.
The same post at 09:00 and at 23:00 gets a different ER. AwrAiter finds each channel's best day and window from its own history.
Every connected channel gets ER, ER by subscribers and ERR per post and per period, in channel analytics, with no spreadsheet.
03 — Benchmarks
Per-channel medians across channels run in AwrAiter, 90 days to 2026-09-03: 19 channels, 7442 posts; a channel counts with at least 10 posts and 100 subscribers. Small sample, real numbers — a sanity check, not a target. Buckets with fewer than five channels are not shown.
| Channel size | Channels | Posts | ER (median) | ER by subscribers | ERR (reach) |
|---|---|---|---|---|---|
| < 1 000 | 6 | 1276 | 1.7% | 0.36% | 4.6% |
| 1 000 – 10 000 | 13 | 6166 | 1.6% | 0.30% | 21.3% |
The two buckets agree on ER (about 1.6–1.7% of viewers react) and differ four times on ERR: small channels reach a small share of subscribers. That is the number to work on first.
04 — Questions
By views, 1–3% is ordinary for a channel between 1 000 and 100 000 subscribers; 5% and above is strong; above 10% on a channel with real reach is rare. By subscribers, halve those numbers. Always quote which formula you used.
Because the denominator shrinks. Reactions come from the active core, which stays; the people who leave never reacted. Watch ERR and subscriber growth together with ER.
Yes, when the channel has a linked discussion group; comments are read from it and added to the numerator.
Same formulas, so yes, up to the moment of capture. TGStat looks at any public channel; AwrAiter looks at the channels you run and shows the posts behind the numbers.
Post in the channel's best window, end more posts with a question, cut the formats that keep scoring below the channel's median. AwrAiter's insights show each channel's best window, whether questions lift engagement, the best length, and the top and flop posts.
Connect it; up to a year of history is scored, starting the first night. One channel is free, forever.