Reports
By Pranav, Shivam Mishra, and Sony
By Pranav, Shivam Mishra, and Sony
Use reports and analytics to measure team performance, conversation volume, response times, and customer support quality.
How to read Overview Reports (realtime)?
The Overview page is a real-time update of what is happening in your account––how many conversations are open, which agents are available, etc. To view the Overview Report, go to Reports → Overview. This is what it looks like: Metrics in the Overview Report Open conversations This shows the number of conversations on your account that are currently: - Open: Ongoing customer interactions - Unattended: Unattended customer conversations - Unassigned: Conversations waiting to be assigned an agent. Agent status This shows the availability status of agents, categorized by: - Online - Busy - Offline Conversation traffic This heatmap shows what times of the day are the busiest for the last 7 days. Each cell represents the number of conversations received as a color shade; hovering over a cell will show the number of conversations received for that hour of the day. Conversations by agents This shows the number of conversations currently being handled by each agent. It can also be used to monitor their performance and workload.
How to read CSAT Reports?
If you have enabled CSAT surveys in your inboxes, you can analyze the results using the CSAT report. To view the CSAT Report, go to Reports → CSAT. This is what it looks like: Customizing CSAT Reports CSAT reports can be filtered by the following metrics. Duration You can select the duration you want to see the report for. By default, this is set for the last 7 days. Agents You can analyze the CSAT reports for particular agents only. Select one or more agent names from the dropdown. Metrics in the CSAT Report Quick Overview The upper half of your report will show metrics like this screenshot. The following list explains what each metric means. Total responses This is the total number of responses received on your CSAT surveys, during the selected period and for the selected agent or agents. Satisfaction score This is the mean satisfaction score the survey takers gave, calculated using the following formula. (Total number of positive ratings/Total number of responses) * 100 Response rate This indicates the proportion of customers that responded to the CSAT surveys sent. It is calculated using the following formula. (Total number of responses/Total number of CSAT survey requests sent) * 100 Emoji scale Whenever a CSAT survey is requested from a customer, they are asked to rate their experience on an emoji scale like this: The emoji scale metrics that you see on your CSAT report show how many people (in percentage) used which emoji to rate their experience. Details of responses The lower half of your report will list all the responses received. For eg., The screenshot below lists the CSAT survey responses received for agent Pranav Raj. Tip: You can click on a conversation ID to jump to the associated conversation.
How to read Conversations Reports?
The conversation report is an overview about all your conversations across a time-period. With the Conversations Report, you can learn significant metrics about the health of your conversations, like the volume of messages, First Response Time, Resolution Time, Resolution Count, etc. Metrics The following are the metrics that are shown in the report Conversations Counts the number of conversations daily/weekly etc, it only counts conversations created in that time-period. So if a conversation is re-opened from a previous time-period it’s not included. The heatmap in the overview report can provide you with deeper insights into conversations created. Messages received The number of messages received across all channels in a given time period. Messages sent The number of messages sent from the account in a given time period. The count includes messages sent by both bots and agents. First Response Time First Response Time or FRT is the average time a human agent takes to reply to a message over conversations created during a particular time period. The first response is calculated as a the time difference between the creation of the first human message and the last non human activity, this is usually when the conversation is opened. or when it's handed off to a human agent by a bot. Resolution Time The resolution time is the average time it takes for a conversation to be resolved, over conversations created during that time period. The resolution time is calculated as a difference between the time when the conversation was first opened and the time it was resolved. Please note that this does not consider reopens as separate conversations, if a conversation was re-opened and resolved later the resolution time will increase. Resolution Count Resolution count is the the number of conversations resolved on a particular day in a time period. Customer waiting time The amount of time customers had to wait for a message from your account. This is similar to FRT, but calculated over all outgoing messages. The higher the number, the slower your team responds to individual messages More Reports Often it is useful to read these numbers filtered over particular team or an agent, you can open one of the following reports from sidebar and view them - Agents: This report is great to monitor the productivity of individual agents and find areas of improvement. - Labels: Labels is extremely useful in categorising the kind of conversations you are receiving. Once setup correctly, this reports provides you insights of teams productivity on specific kinds of conversations. This helps you find gaps in your agents knowledge and areas of improvements in your help-center as well as bots. - Inbox: This report is helps you monitor the metrics of a particular inbox and find trends in the distribution of conversations, how effective is the response and resolution of these conversations. - Team: This report is great to monitor the productivity of a particular team and find trends as well as areas of improvement in your responses for each teams.
How to read Bot Reports?
Bot reports is an overview of bot activity on your account, these bots include CSML and Dialogflow both. This report shows data from all the inboxes with an active bot. It has the following metrics and graphs Metrics - No. of Conversations: Total number of conversations the bot has attended to - Total Responses: Total number of messages sent by the bot - Resolution Rate: The proportion of conversations attended by a bot that leads in a resolution without an agent involvement. It is calculated as a percentage of total number of conversations resolved by the bot in contrast to total number of conversations handled by the bot. A higher number indicates high effectiveness of your bot. - Handoff Rate: Percentage of handoffs from a bot to a human agent. It is calculated as a total number of conversations handed off to agents divided by the total number of conversations handled by the bot. A higher number indicates that your bot is not extremely effective and it might be worth looking at these conversations to find gaps in your bot responses. Graphs - Resolution Count: This shows the conversations resolved by a bot distributed over a time period. - Handoff Count: This shows the number of handoffs done by the bot distributed over a time period.
How to read SLA Reports?
Service Level Agreements (SLAs) are contractual arrangements between you and your customers that define the level of service you commit to deliver. Your SLAs specify the expected performance metrics, such as response times, availability, and resolution times, that you have agreed to provide. You can read more about the SLA feature here. You can filter this report based on date, agent, team, SLA policy, label and inbox as well. This will allow you to drill down into specific insights like the "SLA policy performance of the customer success team for the label 'churn-risk'" Metrics The SLA report shows the following metrics Hit Rate The Hit Rate shows the proportion of SLAs that were successfully met out of all SLAs that were applied. It is calculated by subtracting the number of SLA misses from the total number of applied SLAs, then dividing the result by the total number of applied SLAs. A higher Hit Rate indicates better performance in meeting SLAs. Lower percentages indicate room for improvement. Number of Misses This shows the number of SLAs missed out of all the SLAs that were applied. Number of Conversations This shows the number of conversations for which an SLA was applied. Ideally it's good to have a larger coverage of all the incoming conversations. SLA Report Along with the metrics, you can retrieve a log of all the SLA misses on this page You can click the "View Details" button to see a grouped list of all the SLA events that were missed along with a timestamp of when it happened.
Reading Conversations, Agents, Labels, Inbox, and Team Reports.
In Chatwoot, your Conversations, Agents, Labels, Inbox, and Team reports can all be interpreted using the same metrics. A typical report looks like the following screenshot, with a graph for each of the metrics. What each report means and what these metrics are is explained in the following paragraphs. Types of Reports Conversations Report With the Conversations Report, you can learn significant metrics about the health of your conversations, like the volume of messages, First Response Time, Resolution Time, Resolution Count, etc. Agents Report With the Agents Report, you can learn significant metrics about your agents, like who handles what volume of messages, what is their First Response Time, Resolution Time, Resolution Count, etc. Labels Report Label reports are like Conversations reports but grouped by labels. Inbox Report With the Inbox Report, you can learn significant metrics about your inboxes, like which inbox receives how many messages, what is their First Response Time, Resolution Time, Resolution Count, etc. Team Report With the Team Report, you can learn significant metrics about your teams, like which one receives how many messages, what is their First Response Time, Resolution Time, Resolution Count, etc. How to open a report? Look for the “Reports” button on your dashboard's leftmost sidebar. You’ll see the list of available reports when you click on it. Click on any one to view it. Customizing Reports All types of reports listed above can be categorized by the same metrics. Each option is explained in the following paragraphs. Duration You can select the duration you want to see the report for. By default, this is set for the last 7 days. Grouping of data Once you select the duration, you can select how to group your data. Each date range has a different group by the filter. The date range to group by filter mapping is given in the table below. Metrics Every graph shows the trend for the specific metric. Here is what each metric means: Conversations This graph shows the total conversations received during the selected period. If the data has been grouped by day, the graph displayed on this tab will show the total number of conversations received on each day within the selected period. By hovering the mouse over a specific point on the graph, you can view the exact number of conversations that occurred on that day. In the following example screenshot, you can see that on 25 Aug, 30 conversations occurred. This information can be used to understand the volume of customer interactions and identify patterns or trends in customer behavior. First Response Time This graph shows the average time taken to give a first response to a customer conversation. By hovering the mouse over a specific point on the graph, you can view the exact average time taken for first response and the number of conversations used for the calculation on that day. In the following example screenshot, on 4 Sep, the average First response time was 5 Days & 19 Hours, based on 43 conversations. This information can be used to understand the efficiency of the customer service team in responding to customer inquiries and identify areas for improvement. Customer waiting time This graph shows the duration of time a customer waited for a response from an agent. If the data has been grouped by day, the graph displayed on this tab will show the total number of conversations received on each day within the selected period. By hovering the mouse over a specific point on the graph, you can view the average waiting time on that day based on the number of conversations received that day. In the following example screenshot, you can see that on Aug 29, the waiting time was 11 hours & 22 minutes, based on 11 conversations. Resolution Time This graph shows the average time taken to resolve a particular conversation with a customer. By hovering the mouse over a specific point on the graph, you can view the exact average time taken for resolving conversations and the number of conversations used for the calculation of resolution time on that day. In the following example screenshot, on 26 Aug, average Resolution time was 7 Days & 10 Hours, based on 7 conversations. Resolution Count This graph shows the number of conversations resolved on the selected days. Messages received This graph shows the number of messages received during the selected period. By hovering the mouse over a specific point on the graph, you can view the number of incoming messages that occurred on that day. In the following example screenshot, you can see that on 12 Aug, 91 messages were received. Messages sent This graph is similar to the Messages received graph, except that it specifically shows the messages that were sent from the account instead of the incoming messages. Trend The trend shows the percentage of increase or decrease in the metric data. This is given with every graph. It compares the current time period with the previous time period to provide valuable insights. For example, when analyzing the number of conversations for a specific week, the Trend metric would compare this week's numbers with the numbers from the previous week. It is displayed on the right side of each metric in the report. The formula used for calculating the trend is given below. trend = ((current - previous)/previous)*100 Adjusting reports for Business Hours Toggle on the switch at the top-right corner of your screen to adjust the report’s metrics for business hours. Business hours are used to configure the team's availability in each channel. If Business hours are enabled, report data would be calculated based on the business hours. Refer here to set up business hours.
How to read the SLA Reports
SLA reports provide an overview of your teams compliance on your agreements, the Hit Rate is the most important metric to track, it measures the percentage of SLAs that we're complied successfully. The higher this hit rate is, the better your compliance is. Filtering You can filter the report based on the following attributes - SLA Policy - Inbox - Agent - Team - Label This allows you to get a comprehensive look of what kind of conversations have the best or the worst hit rate. This can also be useful in tracking the performance of the team and agents. You can also filter the report based on a date, this makes it easy to find the hit-rate of a particular team every quarter, and track historically how this metric has changed. Conversation list The conversation list shows the most recent SLA misses, click on the View Details button gives you a summary of the timeline of SLA missed events, it can be used by managers to triage the conversations which have consistently missed their SLAs and being able to take action to ensure that your customers get the desired service in the stipulated time.
Track conversation topics with Monitors
Monitors help you follow recurring topics in customer conversations, such as refund requests, cancellation intent, or product errors. Describe what you want to find in plain English, review the matching conversations, and use an automation or a Slack alert to help your team act on them. This guide walks through a refund monitor, then shows five ways to use monitors across your team. All screenshots use sample data and example Slack channels. Before you start - Availability: Monitors are in early access on the Business and Enterprise Cloud plans, with up to 20 active monitors per account. They do not use Captain credits. - Permissions: Administrators create and manage monitors, review individual matches, and configure automations and Slack alerts. Teammates with access to reports can view monitor reports. - Self-hosted Enterprise: Your instance administrator needs to configure the OpenRouter connection in Super Admin → Installation Configs and enable Conversation Monitors for your account. - For automations: Create any teams and labels you want the rule to use. For Slack alerts, you will also need a connected Slack workspace. 1. Create a monitor Go to Reports → Monitors. The Your monitors tab lists your monitors and the number of matching conversations created in the last seven days. Monitors list with five example topics and their conversation counts for the last seven days Click Create monitor to start with your own description, or open Templates and choose a starting point. Templates cover topics such as login problems, refund requests, cancellation intent, payment failures, feature requests, and escalation requests. Templates tab with ten monitor templates for support, finance, product, and operations teams For this example, choose Refund requests. You can change the template before creating the monitor: 1. Give it a clear Name, such as “Refund requests”. You can also choose an icon or emoji and a color. 2. In Which conversations should this monitor find?, describe what should match and what should be excluded. The description can contain up to 500 characters. 3. Click Preview matches to check your definition against a small sample. 4. Once you are happy with the definition, click Create monitor. Create monitor dialog with the Refund requests name, an include and exclude description, Preview matches, and Create monitor controls Write a useful description Describe the customer's intent, then add boundaries that distinguish it from similar requests. For example: A customer asks for money back on a payment they already made, including duplicate charges and refunds after cancelling. Exclude questions about the refund policy and requests for a discount. With this definition, “I was charged twice. Please refund the duplicate payment” should match. “What is your refund policy?” should be excluded. A specific definition makes the results easier to review and the actions more predictable. Preview matches checks up to five recent conversations from the last seven days. It is a sample, so a preview with no matches does not mean that the full scan will find none. Conversations without usable text may be excluded from the sample. 2. Review the report and matching conversations After creation, the monitor checks conversations created in the previous seven days. It then continues checking as new public customer messages and agent replies arrive. Live checks normally use the latest five public messages with usable text, along with configured customer attributes. Historical scans and rechecks can use more conversation history, subject to context limits. Private notes are excluded. Open the monitor to see its chart. Use the date-range control and choose Group by 1 hour, 6 hours, or 1 day to explore the pattern. Refund requests monitor showing 17 matching conversations grouped by day Each conversation counts once per monitor, on the date the conversation was created. If a conversation created on Monday first matches on Thursday, Monday's bar can increase. The chart counts matching conversations, not matching messages or the time an alert was sent. The same conversation can match more than one monitor. As an administrator, click a bar to open the conversations behind that count. Select a conversation to read the full exchange and check whether it fits your definition. Drilldown for November 8 showing four sample refund conversations behind a chart bar Results update automatically. While a scan or recheck is running, counts can still change. If the report warns that a period was not fully checked, treat its count as incomplete. 3. Run an automation when a monitor matches Use an automation when you want Chatwoot to take an action on a matching conversation, such as assigning a team, adding a label, changing priority, or sending a webhook. 1. Open the monitor and click Create automation below its chart. 2. Give the rule a name, such as “Route refund requests to Billing”. The Monitor Matched event and your monitor are selected for you. 3. Add optional conditions if the rule should apply only to particular inboxes, teams, or customers. 4. Add the actions. For our example, choose Assign a Team → billing and Add a Label → refunds. 5. Save the rule and check that it appears as Active under the monitor's Automations section. Example automation with Monitor Matched event, Refund requests monitor, Assign a Team billing action, and Add a Label refunds action You can also create a rule from Settings → Automation: choose Monitor Matched as the event, select a running monitor, and configure the conditions and actions. See the automation guide for more about conditions and actions. For a customer message such as “I was charged twice for order #1042. Please refund the duplicate payment”, the example rule assigns the matching conversation to billing and adds the refunds label: Sample conversation actions panel showing the billing team assigned and the refunds label added When does the rule run? - A rule runs once per conversation when that monitor first finds a match from new live activity. Allow time for the monitor to evaluate the message and process the action. - The initial historical scan, rechecks, and scans of a paused period do not run actions for older matches. - Creating a rule after a conversation has already matched does not run the rule for that existing match. - Messages sent by another automation can contribute to reporting, but do not start a monitor rule on their own. Test a rule with a new sample conversation after enabling it. Check the match, the resulting team or label, and any additional conditions before relying on it for your workflow. 4. Send matching conversations to Slack Slack alerts let the relevant team know when a monitor first matches a conversation. Configure them directly on the monitor; you do not need a separate automation rule to send a monitor alert. Connect your Slack workspace 1. Go to Settings → Integrations → Slack, or click Connect Slack in the monitor's Slack alerts section. 2. If Slack is not connected yet, click Connect, choose your workspace, and review the requested permissions in Slack. Your workspace may require an administrator to approve the app. 3. Complete the Slack integration setup in Chatwoot. For a private channel, add the Chatwoot app to that channel in Slack before trying to select it. The Slack integration guide covers workspace connection and adding Chatwoot to private channels. Connected Slack integration showing Two-way sync and Alerts only options, with Alerts only selected The general Slack integration also has its own conversation sync settings. Its Two-way sync mode sends replies in synced Slack threads to customers; Alerts only keeps those discussions internal. Review this setting when connecting the workspace. The general conversation channel and the destination chosen for a monitor are separate settings. Choose a channel for the monitor 1. Return to Reports → Monitors and open your monitor. 2. In Slack alerts, open the channel dropdown. 3. Select a destination, such as #billing-alerts. The selection saves automatically and the status changes to On. Enabling Slack alerts does not send alerts for conversations that have already matched. Slack alerts channel picker with billing-alerts, customer-success, product-feedback, and support-incidents example destinations Refund monitor with an active automation and Slack alerts switched On for billing-alerts An alert includes the monitor name, a link to the conversation, the contact's name, an excerpt from the latest customer message, the inbox, assignee, status, match confidence, and the monitor description. Open the conversation link to investigate and reply from Chatwoot. Confidence represents the model’s match score, shown as a percentage. It does not guarantee that the classification is correct. For example, the refund alert could show: Refund requests A conversation matched this monitor. Conversation #18 · Jamie Lee “I was charged twice for order #1042. Please refund the duplicate payment.” Inbox: Customer support · Status: Open · Confidence: 94% Alerts are sent for the first match from new live activity. Historical scans do not send alerts, and later messages in the same already-matched conversation do not send repeated match alerts. To turn off alerts for a monitor, select Don't send alerts. If a private channel is missing, add the Chatwoot app in Slack and refresh the page. If alerts stop after a channel is archived or removed, select another channel. If Chatwoot asks you to reconnect Slack, complete the connection again and check the monitor's destination. Five common use cases Start with one topic and review its matches before adding actions. These examples show how a description, an automation, and a Slack destination can work together. 1. Surface escalation requests Find: Customers explicitly asking for a human agent, manager, specialist, or higher support tier because their issue remains unresolved. Exclude routine transfers and sales requests. Act: Assign a support lead, raise the conversation priority, and notify #support-escalations so the team can review the conversation and arrange the right follow-up. 2. Follow up before a customer cancels Find: Customers who say they intend to cancel or will not renew. Exclude seat reductions and downgrades if those are handled separately. Act: Add a cancellation-risk label, assign Customer Success, and notify #customer-success. An account manager can read the conversation and offer a relevant follow-up. 3. Spot recurring product errors Find: Errors, crashes, outages, or severe slowness that prevent a customer from completing a task. Exclude routine how-to questions. Act: Assign Technical Support, raise the conversation priority, and notify #support-incidents. Review the chart for increases and inspect the conversations before deciding whether they point to the same incident. 4. Collect feature requests for product planning Find: Requests for a capability the product does not currently offer. Exclude help using an existing feature. Act: Add a product-feedback label and send matches to #product-feedback, or use a webhook action to pass the conversation to your feedback workflow. Review the chart and the underlying conversations together when prioritizing requests. 5. Catch missing order updates Find: Missing order confirmations, shipping updates, or delivery notices. Exclude delivery-status questions where the customer already received the tracking email. Act: Route matches to your operations team and notify an order-support Slack channel. Use the conversation details to distinguish a single order problem from a wider notification issue. Keep your monitors useful - Edit: Refine the description when matches are too broad or too narrow. Changing it clears existing matches and rechecks previously checked conversations. Skipped periods remain skipped; a paused monitor rechecks when resumed. - Duplicate: Try a variation while keeping the original definition and results available. - Pause: Stop checking new activity while keeping the collected results. Pausing also disables linked automations. - Resume: Choose Check the paused period to catch up, or Start from now to skip that period. Re-enable linked automations yourself in Settings → Automation when ready. - Delete: Remove a monitor and its results. Its conversations remain available, and linked automations are disabled. If a monitor shows Needs attention, review the displayed error and use Retry once the cause is resolved. On self-hosted installations, an administrator may need to check the OpenRouter connection. If the shared AI classification allowance is reached, existing results remain visible while evaluations and previews wait for the reset shown in the dashboard.