Five hundred people left after a post you thought was good. Twelve hundred after an ad integration. Almost nobody after last Friday evening. These are the questions Telegram's built-in analytics can't answer — it shows you totals, not reasons.
What the native stats actually give you
The Statistics section (available once a channel hits 500 subscribers) shows subscriber growth and decline over time, broad traffic sources, post reach, and view data for recent weeks. Useful as a starting point. But it has gaps that matter.
You see that 800 people left this week. You don't see who — whether it was long-time readers or bots from a recent inflation wave. You see that 300 subscribers came via invite links but can't tell which specific link brought each person, let alone how many of them are still around thirty days later. And history beyond a few months simply doesn't exist.
Retention: the number that tells you what your audience is actually worth
Of 1,000 people who subscribe in a given week, 870 are still there after 7 days, 720 after 30, 550 after 90. That's a retention funnel, and it's a much more honest measure of channel health than total subscriber count.
A channel with 80% 30-day retention is worth significantly more to advertisers than one with the same subscriber count but 40% retention. The funnel is the real value of your audience.
Invite links as attribution, not just counters
Most channel owners create one invite link and watch how many people click it. That's using maybe 10% of what invite links can tell you.
Run ads on three channels with three separate links. After two weeks: Channel A brought 400 subscribers, 85% still there. Channel B brought 600, but 60% already left. Channel C brought 200 and 92% stuck around. Channel B looks like the winner on day one. Channel C is actually the best placement by a wide margin — you just can't see it without retention data per link.
Reading unsubscriptions as feedback
People leaving isn't the problem — it's information. A spike right after a specific post type tells you the audience disagreed with the direction. Losing more than 1% of your audience after an ad integration means the ad didn't match what people subscribed for — anything under 0.5% is within normal range.
Go silent for five or more days and watch what happens to the unsubscription rate. That number tells you more about your minimum posting rhythm than any general advice about "ideal frequency."
How TGuard tracks the full history
Telegram's interface shows 200 members. TGuard solves this from a different angle: from the moment it's connected, the bot logs every join, leave, and ban event as it happens through the Bot API. The history never resets and goes back as far as the connection date.
Each event includes timestamp, user name and username, country, last-seen status, bot/premium flag, and the invite link they used to join. Accounts that repeatedly subscribe and unsubscribe show up clearly in the timeline — the cycling pattern associated with bot inflation is easy to spot once you have the full log.
Using it in practice
Launch a new content series and check the unsubscription rate two weeks in against your baseline. Higher than normal means it's not landing. That's cleaner feedback than comments or gut feeling.
Judge ad placements by 30-day retention, not just subscriber count on day one. And look for the people who've been in the channel more than 90 days — those are your actual core readers, and what they engage with should be shaping your content direction more than anything else.
Frequently Asked Questions
Telegram only shows aggregated data: total growth or decline for a period and broad subscription sources. It does not show who unsubscribed, when, or after which post — and it does not store history beyond a few months.
A retention funnel shows how many subscribers who joined in a given period remain in the channel after 7, 30, and 90 days. It is a key metric for measuring content quality and audience-channel fit.
TGuard accumulates a full timestamped log of every join, leave, and ban event with user details (country, last seen, bot/premium label) and the invite link used to join. You can filter by date and manually compare unsubscription periods with your posting history.