How to interpret the Inbound Metrics report: Bots Section?
This report allows you to analyze the performance of active bots created with Flowbuilder. It shows the conversations that started in the selected period, how users behaved within the flow, and what results were obtained.
❓Frequently asked question or initial context
Can I see all the classifications applied during a conversation, both by the bot and the advisor?
Why don't the tags appear in the report?
Why are bot and advisor classifications mixed when filtering by abandonment?
How do I interpret a customer's journey within a flow?
🤓 Explanation
This report only considers bots created with Flowbuilder and allows you to analyze both the general interaction and the detail of the path users take within a bot.
Includes the following tabs:
Results: overview of bot performance.
Bot Flow: step-by-step customer journey.
Audit: complete detail of each conversation.
💬 What does it mean for a conversation to be abandoned?
A conversation is considered abandoned when the customer stops interacting:
During a flow or bot: if the customer no longer responds and remains at a point that is not the end of the flow or assignment.
During a flow or bot: if the customer no longer responds and remains at a point that is not the end of the flow or assignment.
With assigned agent: if after the assignment the client does not send any more messages.
With assigned agent: if after the assignment the client does not send any more messages.
💬 What is an assigned or reassigned conversation?
An "assigned" conversation is one that has been delivered to an agent, either automatically (by bot) or manually.
An "assigned" conversation is one that has been delivered to an agent, either automatically (by bot) or manually.
If that conversation is passed to another agent before being closed, it is considered "reassigned". Time metrics (first response and average) are eliminated and measured again from scratch for the new agent.
If that conversation is passed to another agent before being closed, it is considered "reassigned". Time metrics (first response and average) are eliminated and measured again from scratch for the new agent.
⚙️ Detail by tab
📊 Results Tab
Available filters:
Conversation start date
Channel (WA, FB, IG, Web, etc.)
Source type (inbound)
Bot name
Classification and type of classification
Main metrics:
Active bots: bots with at least one conversation in the period.
Conversations started: active conversations in the date range.
Assigned to agent: conversations routed to a human.
Sales: conversations classified as a sale.
Performance per bot
Distribution by last classification
Summary of classifications
📌Important about classifications:
The reports only consider the last classification applied to the conversation, since that is what determines the closure.
If a conversation had multiple classifications, only the last one will be shown.
This is an expected behavior of the system and is aligned with the objective of correctly classifying each management.
🔎Example:
Day 1: the bot applies a negative classification.
Day 1: the bot applies a negative classification.
Day 2: the client re-enters and an agent applies a positive classification. When running the report filtering by Day 2, only the last positive classification will be seen.
Day 2: the client re-enters and an agent applies a positive classification.
When running the report filtering by Day 2, only the last positive classification will be seen.
If you want to analyze all the intermediate classifications within the bot, it is recommended to use the Bots Flow tab.
🧭 Bots Flow Tab
Visualize the customer's journey within the flow:
First option chosen: first customer selection in an "Evaluate" type block.
Flow completion: abandonment, end of flow, or assignment to agent.
Classification path: how many times each classification configured in the flow was passed through.
Tag path: how many times each tag was activated.
Last message prior to abandonment
📌Important about tags:
Although this tab shows the path by tags, the data depends on the tags remaining active in the flow.
If a tag was deleted or modified in the bot, it will no longer appear in the report, even if it appears in old records.
🔎Frequent case:
When reviewing the "March 2025 Event" bot, it was detected that the tag "GP Interested Groups" does not appear because it was removed from the flow. Therefore, the system no longer recognizes it as active and does not reflect it in the current results.
📁 Audit Tab
Shows up to 2,000 records on screen.
Download available for up to 1,000,000 records in .csv, .xlsx or .json.
Includes all the fields mentioned in the report, such as channel, date, bot name, assignment, tags and last classification.
✅ Possible solutions
📌 How to see more than one classification per customer?
It is not possible to visualize all the classifications applied in a conversation in the same report. Only the last one is shown.
To see multiple classifications applied by the bot, it is recommended to use the Bots Flow > Classification Path tab, which will show the number of times a conversation passed through each one.
📌 Why are bot and agent classifications mixed when filtering by abandonment?
This happens because the "Abandonment" classification is based on the type of flow end, not on the classification applied. Therefore, the summary of classifications may include classifications made by an advisor, even if the conversation started with abandonment.
📌 Why does the download show less data than the KPI on screen?
When exporting the classification summary and applying filters, it may appear that there is less data than in the Atom KPIs. This may be due to:
Incorrect date filter.
Deletion of tags or classifications from the bot.
Conversations that did not complete the expected journey.
📝 Additional notes
Started conversations do not have a time limit if they are not classified. It is essential to ensure the correct classification of each conversation to maintain traceability.
Started conversations do not have a time limit if they are not classified.
It is essential to ensure the correct classification of each conversation to maintain traceability.
The metrics in this report do not represent the same value as the Your Subscription section, as the reporting is based on the flow and not on billing for effective conversations.
The metrics in this report do not represent the same value as the Your Subscription section, as the reporting is based on the flow and not on billing for effective conversations.