Telegram reach falls as posting frequency rises — but not for Ukrainian channels

Post more often on Telegram and each post reaches fewer people. That is the pattern across ~5000 channels: reach per post falls at every step up in publishing frequency, a decline clean enough to look like a property of the platform. It is not. Split the same channels by language and one group breaks the curve outright — it publishes several times a day, where reach is supposed to crater, and holds it high anyway

The numbers are not self-reported. TGList indexes 2.4 million Telegram channels, and for a subset we store the timestamps and view counts of individual posts, which pins down two things a channel owner would otherwise be free to inflate: how often it publishes, and how many people actually open each post

What we measured

For each channel we take the posts we crawled since 2026-06-01, compute the publishing rate from the span between the first and last of them, and take the median view count across those posts. Divide median views by subscriber count and you get the share of the subscriber base that a typical post reaches

That ratio is reach rate, for example: channel with 100,000 subscribers whose posts get 10,000 views has a reach rate of 10%

Reach falls with every step up in frequency, from 16.6% for channels posting less than once every other day down to 4.5% in the 8-16 band — a factor of 3.7. The decline is steepest early: most of the damage is done by the time a channel reaches two posts a day, and after that the curve flattens into a floor

The obvious objection is size. Large channels have lower reach rates for structural reasons, and if high-frequency channels were simply bigger, the curve would be an artefact. They are not. Median subscriber count is essentially flat across the bands:

Median subscribers move between 76K and 105.4K across bands that differ in reach by a factor of 3.7. Whatever is driving the decay, it is not channel size

Then language breaks it

Split the same cohort by language and the tidy story falls apart. Ukrainian channels publish 4.9 times a day at the median — well into the range where the global curve says reach should have collapsed — and hold a reach rate of 18.2%. English channels publish 1.2 times a day and reach 4.2%

That is 4.2× the posting volume and 4.4× the reach, in the same direction. The fatigue curve is not a law of the platform

The tag layer explains what is going on. Tags in our index are independent of both category and language — a channel can be tagged war whether it is Ukrainian news or an English-language defence blog — which makes them the right instrument for cutting across the language split

The war-adjacent tags occupy the top of the table and they are also among the most frequent publishers in the entire dataset. This is the one context where the fatigue effect inverts: when the information is consequential enough — when readers are checking for things that affect their physical safety — frequency stops being a tax and starts being the product

It shows up at the far end of the curve too. The 16+ band is the only place where reach ticks back up, and it is not a random assortment of channels: 40% of it is Ukrainian, against 6% of the cohort overall. The uptick is the wartime news cluster reappearing at the extreme

Categories

Personal blogs and music channels post rarely and reach a quarter of their subscribers; jobs and betting channels post constantly and reach two percent

Most of that table follows the frequency story. Crypto does not. Crypto channels post 0.78 times a day — among the calmest in the dataset — and still reach only 1.9% of their subscribers. Under the fatigue model they should be near the top

The most likely explanation is that the denominator is fiction. Reach rate is views divided by subscribers, and it collapses when the subscriber count is inflated with accounts that will never load a post

Explore the data

The full cohort is below. Pick a category, language or tag and the curve redraws for that slice against the overall trend, with the largest channels in the slice listed underneath. Thin cells are dropped rather than plotted

Limits

• The cohort is the 5,164 channels for which we hold crawled posts with view counts, not a random sample of Telegram. It skews toward channels large and public enough to be indexed, and toward Russian-language channels, which are 60% of it
• This is a cross-section, not a longitudinal study. We are comparing channels that post often to channels that post rarely, not watching a channel change its cadence. The causal claim — that a given channel would lose reach by posting more — is consistent with the data but not established by it
• Publishing rate is inferred from a capped crawl window, so it measures the rate during the crawled span rather than a long-run average