​Just dropped a research note on Telegram Ads.

Here’s your TL;DR ↓

– The experiment started with 7 ad creatives at 5 TON CPM and a 30 TON initial budget, aimed at finding the most efficient target clusters first.
– Telegram Ads is CPM-based, and channel owners get 50% of sponsored-message revenue – which makes high-volume delivery very attractive.
– Broad targeting looked strong on paper, but the data got weird fast: Telegram showed almost 200 new subscribers, while the channel grew by only 30.
– In the broad setup, 86% of the subscribers left within hours, so the platform still counted them as paid conversions.
– The author’s read: channel owners can watch when they’re targeted and spin up bot traffic to make the ad look better than it is.
– Even targeted-channel buys were not clean; some joiners looked fake, with “last seen” timestamps that did not fit normal human behavior.
– The practical takeaway is blunt: compare platform actions with net growth, check Joined/Left spikes, and use a separate invite link per ad.

Read the full story here: Telegram Ads: Growth experiment that revealed a money farming machine