Two placements each brought in one hundred entries. The first placement cost less, so it looks like the winner. A week later it turns out: one brings suitable questions, the other brings almost no inquiries. Entries are needed in the report, but that column alone is not enough for the next ad purchase.
Give each placement its own link
Create a log with clear names: placement, date, ad topic, link, amount, publication time, and placement terms. If you use a Telegram invite link, set aside a separate one for each placement. The service lets you see entries by additional links. Telegram description.
If the ad first leads to your website, use a tagged link to that page and a verified counter. Tags on the web address help distinguish sources of web visits. By themselves they do not connect each visitor to a subsequent Telegram entry. That requires a separate agreed-upon tracking method.
Do not reuse one ad link in personal messages and other placements. Otherwise its name will no longer match the actual source.
Fix the same measurement window
Take measurements, for example, one day and seven days after each ad goes live. These are working checkpoints that you can change to fit your deal cycle. Comparing "placement A over a week" and "placement B over an evening" is unequal from the start.
Note reschedules, deletion of the post, additional reposts, and parallel promotions. If the ad changed, keep both versions. Otherwise a successful text might accidentally be credited only to a lucky choice of placement.
With a small volume, a single order strongly changes the result. Don't make a lifetime ranking of channels based on the first attempt; write down how little data you currently have.
Calculate the cost of a clear result
Cost per entry = placement spend / entries via its link. When there are no entries, the report should say "no result," not a zero cost.
A hypothetical training example: a placement for 6 000 rubles brought 100 entries — at 60 rubles each. If ten suitable inquiries are measured, the cost of such an inquiry is 600 rubles. These are different metrics, and they need different names in the table.
Only a confirmed payment with a chosen refund accounting rule can be considered a purchase. A created order or an open payment page does not confirm revenue.
Don't call all remaining subscribers "engaged" if you don't know what they did. Define the observable action in advance: submitted a request, wrote a substantive question, went to a specific piece of content. General Telegram statistics do not provide a universal list of who read each post. Statistics documentation.
Decide what to repeat
Break the result down into three questions. Did the placement bring the right people? Did the ad create the right expectation? Did the channel give a clear next step? If there are target inquiries but no payments, check the offer and how requests are handled. Simply increasing the number of clicks can increase the load while keeping the same problem.
Before repeating, keep one element unchanged so you can understand the reason for the improvement. For example, keep the text and the offer, and compare two similar thematic placements. This is still an observation in real conditions, where seasonality and external events cannot be fully ruled out.
Do not include purchased views in evidence of advertising effectiveness. If you simultaneously use services for Telegram, record the volume and date separately. The form below relates to views of a specific post and does not replace buying advertising or attracting interested customers.
When cheap views produce expensive inquiries
A training example in the same seven-day window: placement A costs 3 000 ₽, gets 10 000 views, 100 clicks, and 2 suitable inquiries. Placement B costs 4 000 ₽, gets 8 000 views, 80 clicks, and 8 suitable inquiries. For A, CPM equals 300 ₽ and CPC is 30 ₽; for B, 500 ₽ and 50 ₽. But an inquiry costs 1 500 ₽ for A and 500 ₽ for B. A cheap click here does not mean a cheap suitable lead.
In the media planner select the actual result and enter spend, views, and clicks. Keep suitable inquiries in a separate column of your log: the tool's sales field is intended for confirmed sales, not for renamed leads. Check the source and refunds before choosing the next placement.