We take the subscriber not to the counter in the channel, but to payment. Traffic goes to our own pre-lander, every combo gets its own invite link, subscriptions and CRM money come together in one report.
Meta cannot measure a conversion inside the messenger: the “subscribed” event never reaches the ad account, there is nothing to optimize for. We lead to our own pre-lander with a pixel — the algorithm gets a signal, and the person sees what they are subscribing to.
Inside Telegram you cannot see which campaign the person came from. Without tagging all combos are compared by cost per subscription — and the winner is the one that brings the cheapest and least paying people.
In our two-year project a subscription cost $2.48 to $4.07 depending on the platform, and that was not what decided: a 1.6% subscriber-to-payment conversion at a $1,281 ticket — that is what the price ceiling is calculated from.
Several versions on our own servers: what the channel is, what the person gets, a join button. We build it ourselves — it is a separate agency service, not outsourced.
A separate invite link for every “campaign → creative” pair. The only way to tie a subscription to an ad; without it end-to-end analytics cannot be built at all.
The pixel natively in the <head>, events duplicated server-side with deduplication. Clean signals — the algorithm learns from subscriptions, not clicks.
We subtract organic growth and unsubscribes. The Meta ad account’s own subscription counters are not the source of truth — we reconcile with external tracking.
Platform spend in dollars at the daily rate, subscriptions by invite link, CRM deals — in a daily dashboard. That is how you see which combo brings paying people.
A combo lives 7–14 days. We keep a flow of hypotheses: we test messages and scale a theme, not a single ad.
| $2.85 | per subscription from Meta Ads | Took Telegram subscribers all the way to payments — 569% ROMI over two years |
| $3.09 | per subscription on average across three platforms | |
| 1.6% | subscriber-to-payment conversion | |
| $1,281 | average ticket — 1,581 deals | |
| 15% | marketing share of the ticket — headroom for scaling |
Technically yes, but the algorithm has nothing to optimize for: the subscription event inside the messenger never returns to the ad account. We lead to our own pre-lander with a pixel and a link into the channel — that gives both a signal for optimization and source tagging.
By Telegram invite links: every “campaign → creative” combo gets its own link. We subtract organic growth and unsubscribes — the report shows net.
In our largest project — from $2.48 (Google Ads) to $2.85 (Meta). The price depends on geo, language and how narrow the channel’s topic is.
No: Meta does not sell impressions in Russia. For a Russian-speaking audience we work through geos outside Russia — Europe, the CIS, Israel, the UAE, the USA.
From $3,000/month for traffic. Below that the algorithm does not get enough subscription volume to exit the learning phase, and tests stretch over months.
On Meta — with ads that lead to a pre-lander with a pixel and a link into the channel. Every “campaign → creative” combo gets its own invite link, so the report shows net subscribers for every source.
Egor Balandin
online