Analytics
Where customers come from, which reactions they use, and how many pay. How to read audience and chat metrics.
इस पृष्ठ पर
Open Analytics to see how many people started using the bot, where they came from, and how many paid. For example, compare two advertising links to see which brought more buyers. Platform request usage is shown separately in Stats.
Period and tabs
At the top, choose a period (7 days / 30 days) and switch between two tabs: Audience and Chats.
Audience
- Users: total count and growth for the period.
- Payments, ₽: the value of successful Telegram payments for the period (see Payments and balance).
- Top 5 reactions: the most frequently triggered reactions (title and trigger count).
Acquisition sources
The Sources table shows where new subscribers came from. A source is captured when a subscriber first starts the bot, using utm_source, utm_medium, and utm_campaign in a link such as t.me/<bot>?start=utm_source-vk__utm_medium-cpc, or the VK ref parameter. For example, you can compare visitors from a VK ad with another channel.
| Column | Meaning |
|---|---|
| Starts | New subscribers from the source in the selected period |
| Paid | How many of them made at least one successful payment during the period |
| Conversion | Share who paid: Paid / Starts |
| Payments | Number of successful payments |
| Revenue | Value of successful payments, shown separately for each currency |
If the bot accepts payments in multiple currencies, the amounts are never combined into one number. Subscribers without a source tag appear in the Untagged row. That row remains visible even if there are more tagged sources than fit in the table. Without the analytics.revenue permission, revenue is hidden, while Payments and Paid remain visible.
For an integration, the same data is available in the sources field of GET /api/bots/{botID}/analytics or the MCP tool get_stats.
Chats
The Activity by chat table shows each chat's message and reaction counts, activity indicator, and type (Private or Group). It helps you see where the bot is used most.
Site analytics
Open Site analytics for the selected project to see popup and widget activity. The selected dates are inclusive; the period and daily breakdown use UTC.
The Popups tab counts recorded impression, viewable-impression, interaction and conversion events. These are events, not people. An impression is viewable when at least half the popup is visible for two consecutive seconds in a visible tab; this does not prove the content was read. A conversion is not necessarily a purchase. The table calculates conversion as conversion events divided by impressions, not as a share of unique people.
The Widget tab shows received site pageview events, distinct non-empty identifiers in those events, active web dialogs and incoming messages. An active dialog has at least one inbound message in the selected period. Pageview events can also come from server-to-server submissions; identifiers are not unique people or sessions, and pageviews do not mean the widget was opened. Data can be incomplete when tracking is disabled.
Financial data is separate, in the report for an individual popup, and remains subject to its existing permissions. Do not add amounts across popups: the same event may be associated with more than one. See Site widget.
Saved funnels and experiments
A funnel shows how many people completed a sequence of steps, such as opening a catalog, choosing a product, and paying. Conversion is the share who reached a step; drop-off is the number who did not continue. Median time means half the participants completed the path faster and half took longer.
In a saved funnel, you can set alternative events for a step, conditions on their properties, a UTC date range, a current audience segment, and a time limit for completion. The report also shows trends and comparisons between groups, or cohorts.
Use experiments to compare versions of a message or workflow. A published version is fixed so results refer to the same settings. The report shows the currency, the rule for assigning a result to a source (attribution model), and the period counted (attribution window). A data-quality warning alone does not establish whether the difference between versions is statistically reliable or due to chance.
The older reaction-derived funnel remains available for its original summary use case; it does not replace the saved event funnel builder.
What's next
- Stats and log: spend and request log.
- People (CRM): detailed subscriber database.
- Reactions: basics: what is measured in tops and funnels.
- Saved event funnels: ordered event journeys and cohorts.
- Experiments and attribution: variants, facts, and revenue.