How to check a Telegram channel for bots
Bought subscribers and bots show on the chart and in the views. What to look for, and how the score accounts for it — on two live examples.
Subscribers are the cheapest number in Telegram: they are bought by the thousand, and on the page they look the same as real ones. So the score does not look at the subscriber count itself, but at how it changes day by day and how it relates to views.
On September 10, 2026, 2,811 channels had a full, non-preliminary score. On the «growth» signal they split like this: 986 got all 20 points, 1,551 got 12, 242 had spikes, and 32 had a history shorter than seven days. What stands behind each number is below.
How it is done
Inflation is sold openly, and the public offers show what it is made of. Not «a thousand subscribers» in one line, but a set of separate actions, each leaving its own trace.
The accounts are warmed up first: for a week they simply exist — they log in, they read, they do nothing. The warm-up is there so Telegram does not remove them on day one, which is why inflation rarely arrives in a single day.
Then the actions start. Reactions land on the last few posts thirty to sixty seconds after publication. Comments are written under new posts — not under every one, but with a probability, so it looks uneven. Posts are «read» in the background to lift the view count. Subscribing and inviting people into a chat are sold as separate services.
The pool runs from five accounts to a thousand, each behind its own proxy. Five are given away free; a hundred a month costs around three thousand roubles. This is not an underground technology but a service with a price list, and a cheap one.
We name none of these services and link to none of them: taking the mechanics apart is useful, advertising the provider is not.
What shows up in the numbers
Every one of those actions leaves a trace in our daily snapshots — we record subscribers, views, reactions and forwards for every channel in the catalogue every day.
A daily jump: one day that brought in noticeably more than every other day of the month. A steady gain on a channel that barely posts: readers keep arriving, with nothing to read. Views above the subscriber count with zero forwards: the post was supposedly seen by everyone and then some, yet nobody passed it on. Impossible engagement: that many views do not happen, and the score zeroes the row by itself.
The trace «reactions in a batch within the first minutes» is missing from the table below, and that is more honest than showing it empty: we do not record the timing of reactions at all — only their share of views. Until that measurement exists, there is nothing to count it from.
Below is how many channels of the English segment carry each trace today. As shares, without names: a trace is a question about a channel, not a verdict on it, and a single trace is explained by an ad or a giveaway.
| Trace | Channels carrying it |
|---|---|
| Impossible engagement: more views than can happen | 2 of 298 — 0.7 % |
| A daily jump standing out from the rest of the month | 80 of 298 — 26.8 % |
| Growth on a channel that barely posts | 1 of 298 — 0.3 % |
| More views than subscribers, with zero forwards | 0 of 257 — 0.0 % |
A smooth curve — 20 of 20
Honest growth is many small steps. A channel that is written and read gains a little every day: some readers brought by forwards, some by a mention in another channel. On the chart that is a gentle line, and 986 channels of 2,811 look exactly like that.
A spike is a question, not a verdict
The score treats as a spike a day whose change stands out from the rest by more than three standard deviations. One spike a month is ordinary: an ad went out, the channel was mentioned, a giveaway ran.
First example — «Осторожно, новости», 1,612,296 subscribers. In 25 days of measurement one day gave +1,450, the rest were hundreds in either direction. The spike share of 1 in 25 is multiplied by five: minus 20% of the signal, 16 points of 20. The score does not know where those fifteen hundred came from and does not pretend to; it lowers the weight of growth by exactly the share of days it cannot trust.
Three spikes a month is already minus 60%; five, and the signal is zero. Bought subscribers arrive in batches, and batches show.
Losing on most days — 12 of 20
The second sign is not spikes but direction. If the channel lost subscribers on more than half of the days in the month, the signal is multiplied by 0.6: without spikes that is exactly 12 points. That is what 1,551 channels of 2,811 look like — more than half of the catalog with a full score.
This is not an accusation. Everyone loses subscribers, and gains often do not come every day. But a channel that is in the red on 20 days out of 30 is not growing by itself — it is being grown, and the score notes that by withholding the full growth point.
Both signs at once — 8 of 20
Second example — «ЧИЖИК», a retail chain's channel, 487,802 subscribers, category «Life and health», score 29. In 15 days of measurement 12 were losses — 80% of the days. One day gave +1,826 against the usual tens and hundreds — a spike. The spike share of 1 in 15 times five leaves 67% of the signal; losing on most days multiplies by 0.6 and leaves 40%. 8 points of 20.
Neither sign alone means bought growth: for a shop, the spike could be a giveaway and the outflow the end of a promotion. That is what the score says: not «faked», but «we trust the growth at 40%».
Look beyond subscribers
Subscribers are faked first because they are seen first. So the three numbers on the page are read together. «Осторожно, новости» has 195,382 views per post with 1,612,296 subscribers — 12%; «ЧИЖИК» has 27,600 views per post with 487,802 subscribers — 5.7% against a reference of 14% for its size, and engagement gives it 6.1 points of 30. A hundred thousand subscribers at 1% views is a thousand readers, and they cannot be sold as a hundred thousand.
Engagement above 150%, or four times the reference for the channel's size, the score treats as impossible and zeroes out: nobody gets that many views without buying them.
What an owner should do
Nothing special. Growth from an ad gives one spike and minus 20% for a month, after which the day leaves the window and the signal recovers by itself: the window is 30 days. A giveaway gives the same plus an outflow after it — and also passes within a month. Growth from buying gives spikes every week and losses on the other days, and that is visible for a year.
How the signals add up to the number on the page is the Place Score guide.
How to check a channel for bots in a minute
First, the views. Open several posts in a row and compare views with the subscriber count. A living channel runs at tens of percent; one percent against a hundred thousand subscribers means the subscriptions exist and the readers do not. What to compare it with is in the table of category norms.
Second, the shape of the chart. Every channel card carries a subscriber chart for the month. Real growth is a slope; bought growth is a step inside one day.
Third, reactions and forwards. Bots subscribe but do not react: a large audience with zero reactions is a common sign.
All three numbers sit in the «What the score is made of» table on the card, and each is explained in the statistics guide.