Right after an ad goes live, you want to refresh the subscriber count every five minutes. To decide on the next placement, two pre-scheduled checkpoints are more useful. The first shows whether everything is working. The second helps assess which people and inquiries came in.
A day and seven days are a convenient working example. For an expensive service, the path to purchase may be longer, so the reporting timeframes should be agreed with the real sales cycle.
Before publication, leave a baseline
Record the start time, the subscriber count, the ad link, the amount and the text. Save which action the ad promoted: subscription, reading the material, an inquiry or a purchase. If the invitation is highlighted only for this platform, its statistics help separate the joins. Telegram description.
Note parallel events: another ad post, a newsletter, a new product launch. Without these notes, overall growth is easily attributed to a single source. Also list ordered activities separately.
After 24 hours, check the technical side
First open the placement itself. Do the text, link and agreed time match? Does the link not lead to outdated material? If there is an error, gather the facts and discuss a correction with the platform before analyzing conversion.
Then take the first metrics: joins via the highlighted link, available website visits, questions and started applications. Read the inquiries. Do they relate to the offer, or were people expecting something entirely different?
The Telegram statistics section, if available to the channel, helps you look at the overall dynamics of audience, views and interactions. Telegram documentation. However, an increase in overall views does not give an accurate portrait of each new subscriber.
Do not declare the test a failure just because there are no payments within a day. Check how a decision is usually made in your service. While there are only visits, the report does not yet contain confirmed revenue.
After seven days, assess the substantive result
Update the same columns while keeping the original snapshot. Divide applications into suitable and unsuitable according to a pre-chosen rule. For example, a request for a service from your offer is suitable; a partnership proposal is not counted as a buyer application.
Check confirmed payments and refunds. For each paid application, indicate the source only when it is actually known. If a person saw several placements, do not record one payment entirely in the result of each of them.
The channel's net growth over a week does not equal the retention of the ad group. Different people join and leave during this time. If there is no reliable way to observe specifically those who joined via the ad, write it as it is: "Overall channel dynamics." This is more accurate than inventing a retention percentage.
Decide based on the reason, not on mood
If there were visits but people asked questions about a different service, rewrite the advertising promise. If the inquiries are suitable but the response came too late, sort out the handling. If the link was not working for part of the time, the test must be evaluated with this limitation.
With a good result, repeat the conditions closely enough to check reproducibility. With a weak one, choose one change. Three elements changed at once produce a new test that is hard to compare with the previous one.
In a short report, leave four lines: what was launched, what was confirmed, what we do not yet know, what we will check next. Keep links to the data next to the conclusion.
Post views can be examined separately in the form below. The service relates to the selected publication; its execution should not be recorded as an organic result of the advertising placement or as proof of buyer interest.
Keep the plan and two actual checkpoints
In the training plan, the placement costs 3 000 ₽; 6 000 views and 30 confirmed subscriptions are expected. After a day, 4 000 views and 20 subscriptions were measured; after a week — 6 500 and 26. The planned cost per subscription is 100 ₽, the actual cost at these two points is 150 ₽ and about 115,38 ₽. The counters of each checkpoint are cumulative: 20 and 26 cannot be added together.
Keep three separate calculations in the media planner: the original plan, the day and the week. Note exactly how the subscriptions relate to the placement; overall growth without attribution is not suitable for this column. Fill in confirmed revenue using a single refund rule.
To compare regular posts before and after advertising, prepare two CSV files and open the statistics analyzer. Use a comparable format and the same post age. The mean and median describe the sample of publications but do not measure the retention of the ad group and do not prove a causal effect of the advertising.