Stats and Log
See how many requests the bot uses, which actions cost more and where to check each charge.
На цій сторінці
Stats shows how many requests the bot used and which actions caused the spend. If usage rises, choose a period and check the reactions with the highest cost. You can inspect individual charges in the Request log and export them as CSV for a spreadsheet.
What counts as a request
A request is the billing unit. What is counted is the work a reaction did: not the incoming messages from users:
| What the reaction did | Charged |
|---|---|
| Delivered text message | 1 |
| Delivered media (photo, video, document, voice) | 2 |
| Call to an external service (web request, webhook, integration) | 1 |
| AI call (model reply, speech recognition) | 5 |
| In-platform actions (conditions, variables, delays) | 0 |
| Message that was not delivered | 0 |
One reaction run costs at least 1 request: even if all it did was save a variable. In a chain of reactions every link counts separately, with its own sum and its own floor. Every charge lands in the log, and that is how all spend is tracked.
Summary
At the top: period (7 / 30 / 90 days) and cards: Charged for period, Requests, Current balance, Forecast, days (an estimate of how many days the balance will last at the current spend rate). The Requests / Spend toggle controls the chart.
Breakdowns
- By action type: which actions drive spend.
- By chat: private chats vs. groups.
- Top reactions by spend: the most "expensive" reactions.
Request log
The "Request log" card shows the latest entries; Open full log leads to a separate screen. The log offers filters by period, reaction, and scope (All / Private / Groups), sorting by time and amount, paginated output, and CSV export for downloading into a spreadsheet or external analytics tool.
Test users (see Test mode) are marked test. Their cost is displayed but not charged. In the CSV export, test is the last column.
What's next
- Analytics: audience, chats, and funnels.
- Payments and balance: top-ups and plans.
- Actions: exactly what consumes balance.
Reaction health
The health tab compares reaction duration and failures within the selected period. P50 means half of measured runs completed within that time; P95 means 95% did. Error rate shows how often a run fails, while the number of current runs shows how many are running now. Open a slow or failing reaction from the list to inspect its settings.