A Telegram funnel is measurable in a way most operators never set up, and unmeasurable in a way many of them do not notice. The click is visible. The join is visible if the destination is yours. Everything between the join and the money happens inside conversations, and the conversation is exactly the part no advertising platform can see.
So the measurement problem is not analytics. It is getting a real conversion, with a real value, to travel back out of Telegram carrying the identity it arrived with.
Four numbers, and only one of them decides anything
Scalegram fires four events across the gap: click, join, registration and purchase with a value. Each inherits identity from the one before it, which is why a funnel that fires only the last event performs worse than operators expect. The chain is described in the four conversion events.
As reporting, they produce cost per join, cost per registration and cost per paying customer, per link and per creative. The first two are diagnostics. They tell you where a funnel is losing people. The third is the decision number, and it is the only one that should ever move a budget.
Take the position early: a campaign with an excellent cost per join and no depositors is a bad campaign, not a promising one. Cheap joins are the easiest thing in this market to buy and the least correlated with revenue.
Where the value has to come from
A purchase event without a value trains volume bidding. A purchase event with a value trains value bidding, which hunts for people resembling your highest payers and acts on whatever number you send with complete confidence.
Three surfaces produce a value and they are not equally trustworthy. A lead typing their deposit amount into a verification page completes well and is only as accurate as they choose to be. A synced checkout returns the amount nobody rounded up. A broker deposit pulled through an affiliate connection is the truest source, with the honest limit that some of those APIs report that a deposit happened without reporting how much it was.
The practical rule: if the only number you have was typed by the lead and you are bidding on value against cold traffic, send a fixed value instead and optimise for depositors rather than for typists. More in purchase values.
Counting each sale once
Every event goes out twice on purpose: from the browser where a page exists, and from the server with the captured click identifiers attached. Both carry the same event identifier so the platform folds them into one conversion. Miss that and every sale in the report doubles, which is how a funnel can look profitable for a fortnight before anybody checks.
Synced sales add a second exposure that has nothing to do with browsers. A sync job runs on a schedule, past sales get imported when an integration is first connected, and without a record of what has already been reported the same purchase fires every time the job runs. Scalegram claims each purchase once, however often the sync repeats, which is what makes a scheduled sync safe.
Before you trust a single return figure, confirm your reported conversion count against the number of real sales you know about. A duplicate problem never announces itself. It looks like a good week.
The lag nobody budgets for
In a website funnel a lead converts in a session or not at all. In a Telegram funnel the gap between the click and the money is often measured in weeks of conversation, and no platform attribution window is patient enough to wait that long on its own.
This is why the campaign values are written onto the contact record and stay there. Weeks later, when the person funds an account, their record still says which source, medium and campaign produced them, which means your own reporting can attribute a purchase that the advertising platform has long since forgotten. The mechanics are in how UTMs are handled.
The operational consequence: do not judge a Telegram campaign on a seven-day window. Read cost per join early, because that has volume, and hold judgement on cost per paying customer until enough of the conversation cycle has passed for the depositors to appear.
The costs that belong in the denominator
Advertising spend is the obvious one and rarely the whole one. A serious return figure includes the people answering conversations, whatever your assistant's underlying AI provider charges for the replies, the software subscriptions, and any affiliate or revenue share you pay on the other side.
Most operators in this niche compute return on advertising spend and call it return on investment. They are different numbers and the gap between them is usually the salary line. Naming it does not make a campaign worse, it makes the decision to scale a campaign correct more often.
Reading the list instead of the dashboard
The most underused report is the client list filtered by campaign. Sort by stage and the shape of the traffic is obvious in seconds. Two hundred contacts from one angle, all sitting in the first stage, is a targeting problem wearing a follow-up problem's clothes, and the tags on the rows usually name the objection doing the damage.
That is a diagnosis a dashboard cannot give you, because the dashboard only knows the transitions that happened and not the reasons they did not. It is available because the campaign value is written on the row.
Two limits to close on. Attribution measures advertising and says nothing about whether any given client will do well. Trading carries a high risk of loss, no result is promised anywhere in this stack, and a better return figure describes your marketing rather than anyone's outcome.
"Return on ad spend is not return on investment. The difference is the people answering the messages, and it is usually the line that decides whether scaling is a good idea."
— Roman Onta, Executive Director, SINGUARD
Key Takeaways
- Cost per paying customer is the decision number; cost per join is a diagnostic.
- Value-based bidding is only as honest as the source of the value you send it.
- Check for duplicate conversions before trusting any return figure.
- Telegram conversions lag by weeks, so a seven-day window will understate the funnel.
Frequently Asked Questions
What window should I judge a Telegram campaign on?
Long enough for the conversation cycle to finish, which in this niche is usually weeks rather than days. Read cost per join early because it has volume, and hold the verdict on cost per paying customer until the depositors from that cohort have had time to appear.
Why does my platform report fewer conversions than my client list?
Because the platform stops attributing after its own window closes, while the campaign values stay written on the contact record indefinitely. Your own reporting will credit a purchase weeks later that the advertising platform no longer connects to anything.
Should I send a value on every purchase event?
Only where the value came from a system that confirmed it. If the number was typed by the lead and you are running value bidding on cold traffic, a clamped or fixed value keeps the optimisation honest and costs you granularity you were not going to trust anyway.
About the Author
Roman Onta is an Executive Director at SINGUARD. He builds the Prop Firm CRM, the Broker CRM, Scalegram and CopySignals side by side with his brother Alex Onta, and he helped on the design of eTrader, the division Alex built and leads. His ground is worldwide payment processing, AML compliance and the corporate structures brokers are built on, work the two of them carry together, shaped by executive roles in the UAE and international corporates. He lives and works in Dubai for most of the year. Meet the executive duo leading Singuard's five divisions.