Skip to main content

Insights

Insights is where Flowcall shows all the data it records. Everything is scoped to the account you're currently in — there is no cross-account view today; switch accounts to see another workspace's data.

Each page below maps to an entry under Insights in the sidebar.

Overview​

A holistic, high-level view of the whole system for the current account — activity and trends at a glance, including agent performance and broadcast performance (templates sent, succeeded, failed).

The Customer mood card on the Ticket tab shows the moods the AI flagged in conversations: asking for an agent, frustration, aggression, potential escalation, unable to resolve, and disputed outcome. It opens on By task: one bar per task, plus No Task for tickets without one, split by mood. Next to each bar is the share of that task's tickets that were flagged, so a small task with many unhappy customers stands out against a busy one. A ticket can carry more than one mood, so a bar can count more flags than tickets. Switch to Overall for the totals per mood across all tasks. Click a task to open its flagged tickets on the Tickets page.

The CSAT trend chart on the CSAT tab shows how CSAT moved across the selected period, with one line each for chat agents, email agents and AI. Switch between Average rating and % positive (ratings of 4 or 5), and between Daily and Weekly. Ranges up to a month open daily and longer ones open weekly. Days and weeks follow India time, and weeks start on Monday. Hover a point to see the average, the positive share and the number of responses, so you can tell a real shift from a day with only a few ratings.

On the Agent Dashboard tab, use Channel to select one or more channels: WhatsApp, Instagram, Live Chat, Email, Voice Call, or Manual. The selection applies to the team overview, agent performance, SLA breaches, pending tickets, CSAT, response and wait-time charts. Clear it to include all channels. Links to tickets, agent reports, Agent Activity and SLA records carry the selected channels; the CSAT page keeps the channel only when exactly one is selected.

Workforce Planning​

The Workforce Planning tab on Overview compares new tickets with the agents you had, hour by hour, over one week of business hours. Pick the week with Week starting, filter with Channel, Team Type and Teams, and use Show to switch between all teams and a single team.

  • New tickets count in the hour each ticket started waiting for an agent, not the hour an agent took it, so an hour with too few agents still shows its full demand. Tickets that arrive while you're closed count at opening. A reassigned ticket isn't counted again.
  • Waited for an agent counts the new tickets that took more than 5 minutes of open time to reach an agent, and Median wait is how long those tickets typically waited. Closed hours don't count towards the wait.
  • At capacity is the share of the hour when every Available agent had reached their ticket limit, so new tickets had to wait. With a Channel selected, it uses that channel's limit; see Agents at capacity on the Agent Timeline. Hours before Flowcall started checking capacity show a dash.
  • Suggested agents divides new tickets per hour by how many tickets one agent completed per available hour that week, and Gap compares it with the agents you had. A high At capacity share points to too few agents or ticket limits set too low.
  • Median first response starts when an agent takes the ticket; time spent waiting before that is in Median wait.

The cards at the top of the Workforce Planning tab list the busiest, most understaffed, most overstaffed and most at-capacity hours. By day and hour shows each measure for every day and hour, and Export downloads them as a CSV.

Tickets​

A very important analysis page. It lists the tickets that were created, with filters to drill down and analyze ticket volume, status, source, assignment, and resolution patterns. Pair it with the operational Tickets view.

Sales​

Flowcall lets customers purchase products directly from the platform. Direct WhatsApp purchases aren't heavily used (customers often prefer not to pay over WhatsApp), but when a purchase happens, this is where you see it.

CSAT​

CSAT is the customer satisfaction score. When a ticket is resolved, Flowcall sends the customer a message asking them to rate their experience — a standard support practice.

CSAT is captured for both resolution types:

  • AI resolution — the AI resolved the ticket, and the rating the customer gave.
  • Agent resolution — a human agent resolved the ticket, and the rating given.

Filters let you slice the results by resolution type and other dimensions.

Export CSV downloads every CSAT response that matches the current date range and filters, with the customer, agent, task, issue, channel, rating and comment. The export is prepared in the background the same way as a ticket export: progress shows above the button and under Exports in progress in the notifications panel, the file downloads by itself when it's ready, and a CSAT export ready notification keeps it for 24 hours. An export covers at most 92 days.

Why customers are unhappy​

When Unhappy Customer Reasons is turned on for the account, Conversation Audits has an Unhappy customers tab. The CSAT tab of Overview shows a summary card with the number of unhappy tickets, the share that have a reason, and the top reason; click it to open the tab for the same period. Flowcall turns it on per account in Super Admin Tools, and it needs conversation audits.

A ticket counts as unhappy when the customer:

  • gave a low rating: the CSAT ratings your audit triggers look for, 1 or 2 stars by default;
  • was frustrated or aggressive, made an escalation threat (to senior management, legal or consumer authorities, or in public), or disputed an outcome, as detected during the conversation.

Most customers never answer the CSAT survey, so the moods catch many unhappy customers that ratings miss. Each unhappy ticket whose conversation was audited gets one main reason, who can act on it, and how the contact ended. They are judged from the customer's comment, the conversation, the audit's findings and measured times such as the wait for the first agent reply. A ticket that has both a low rating and a mood is judged once.

Who can act on a reason:

  • Support setup: fixable in Flowcall, through flows, AI answers, FAQs, routing and settings.
  • Agent team: agent replies, speed and follow-through.
  • Business operations: delivery, product and policy, outside support.
  • None: the signal doesn't reflect real unhappiness (for example a positive comment, or a calm message read as frustrated), or there isn't enough evidence to tell.

The main reasons are Slow response, Slow resolution, Handoff or transfer, AI handling, Agent handling, Left unresolved, Policy outcome, Delivery or operations, Product quality, Signal mismatch and Unclear. A ticket can also name a second reason that made things worse.

How it ended is Recovered (the issue was handled and the customer ended calm or satisfied), Not recovered, or Unclear. A customer who rated low never counts as recovered.

The tab shows:

  • unhappy tickets by signal; click a signal to show only those tickets everywhere in the tab;
  • the share of explained tickets for each owner, and how they ended; click an owner or an ending to filter the list;
  • the main reasons, coloured by owner; click one to filter;
  • the tasks with the most unhappy customers, with their most common reason;
  • how often ratings are low by the wait for the first agent reply and by the time to resolve, across all ratings in the period;
  • the unhappy tickets, newest first, each with its signals, reason, ending, a one-line summary, the customer's comment, quotes, and links to the conversation and to the audit findings behind the reason.

A ticket with a low rating counts in the period the rating was given; a ticket with only a mood counts in the period it was opened. A ticket gets its reason on an hourly audit run the day after its conversation, once that conversation has been audited. Audits cover only the previous day, so unhappy tickets whose conversation was never audited have no reason; the header shows how many were explained. Pick the period with the date range above the tab; it can be up to 180 days.

To turn the reasons into changes, ask the Copilot why customers are unhappy: see Find out why customers are unhappy.

Issues Reported​

Very useful for the internal/development team. While reviewing chats, a customer (your team) may spot something wrong — for example, the AI didn't respond correctly. They can report it as an issue directly inside the platform, adding a comment on the expected behavior, plus priority and tags.

When an issue is reported, Flowcall captures background/debug information from the system alongside it (visible to the internal team) to help diagnose the problem. Working through reported issues for each major account — debugging and fixing them — is a regular activity. Much of the debugging can increasingly be assisted or automated by AI, with suggested fixes for review.

Automation Gaps​

Places where the AI or a workflow couldn't complete the job. Use this to find journeys that need better instructions, data, or handoff rules. (Visible when support automation gap analysis is enabled.)

Click Export CSV to download every gap that matches the current filters (not just the visible page), with the message, gap type, confidence, inference, suggested action, candidates, and review status.

Remediations​

Recommended fixes for quality and automation issues surfaced by analysis. (Visible when the relevant analysis features are enabled.)

Remediations retain an issue's occurrence history across conversations and task versions, including evidence attached after it was resolved. Copilot's Needs attention panel and each flow's detail card highlight a few top concerns; click one to inspect the dates, affected tickets, versions, and evidence, or ask Copilot to prepare a fix. See Observed issues in Copilot.

Marking a remediation Resolved records the task's published version and configuration at that time. It does not prove the change worked. Later evidence can identify an older version still being used, an issue observed again on the resolved version, or a possible regression. Earlier resolutions may have no recorded task version.

Conversation Audits​

Review conversations against quality criteria. (Visible when conversation audits are enabled in Settings → Conversation Audits.) When Unhappy Customer Reasons is turned on, the Unhappy customers tab explains why customers were unhappy: see Why customers are unhappy.

When full coverage has started, the Coverage tab lists eligible conversations and their audit outcomes. Select a day or date range to review earlier conversations; the list and the Eligible, Audited, Skipped, Failed and Remaining counts use that range and the selected channels. Clear the date filter to return to the latest audit run's window. Viewing an earlier day does not start another audit run.

Findings in audited conversations breaks down the findings of the Audited conversations in the same range and channels: how many findings, how many conversations had at least one, and for each category its findings, conversations and tags. A finding with several tags counts under each of them. A conversation can have findings in several categories, so the category counts add up to more than the number of conversations with findings.

A conversation that received new messages after its audit shows Re-audit pending and counts under Remaining until it's audited again, so an older audit isn't reported in the period of the newer messages. Its findings are left out of the breakdown, the Findings tab and the exports until then.

Select Eligible, Audited, Skipped, Failed, or Remaining to review those conversations, and filter Skipped by skip reason. Click Export CSV to download all conversations matching the selected bucket and filters, across every page. The file includes the customer and their email, channel, audit status, skip reason, whether another audit is needed, errors, finding counts (No issues found for an audited conversation without findings), categories, tags and details, the agents named in the findings, the dates of the messages each finding cites, conversation links, and record IDs. Occurred and processed timestamps are in UTC.

The Findings tab and its Export CSV use the same date as Coverage: the date of the conversation's latest message, not the day it was audited. A conversation whose latest re-audit failed is counted as Failed, and its findings are left out of the Findings tab and export until an audit succeeds.

Billed conversation audits​

The Billing tab of Conversation Audits shows which audited conversations count toward your audit billing for a month. Pick a month under Month audited to see the total at the top and a list of every counted conversation: the ticket (click it to open the conversation in a new tab), channel, conversation date, when it was audited, the triggers it matched, and how many findings it has. The month runs from midnight on the 1st to midnight on the 1st of the next month in your business timezone. It opens on last month, the one most recently invoiced.

How audited conversations are counted for billing:

  • Each conversation counts once, in the month its first audit completed. When a conversation continues and is audited again, the re-audit is not counted, and the conversation stays in the month of its first audit. The row shows the latest re-audit time under Audited at.
  • Skipped and failed audits are not counted.
  • The month follows the audit date, not the conversation date, so a conversation from the last days of one month can be counted in the next.

Click Export CSV to download every counted conversation for the month, not only the page on screen. The file has the ticket number and ID, channel, conversation date, audited at, last re-audited at, triggers matched, findings and a link to the conversation, with times in your business timezone.

Call Recordings​

All the calls that happened, with their recordings. (Visible when voice calling is enabled.)

Agent Activity​

Tracks human-agent productivity. Whenever a ticket is assigned to an agent, Flowcall creates a tracker recording things like:

  • How much time the agent spent on the ticket.
  • When they first responded.
  • How long they were actively working on it.
  • How they ended the ticket (closed, marked awaiting response, etc.).

Use Source in the filters to show activity from one or more channels.

Export CSV downloads every activity that matches the current date range and filters, with durations, first response, active, wait and blocked times, and the end reason. Like a ticket export, it's prepared in the background: you can keep working, the file downloads by itself when it's ready, and an Agent activity export ready notification keeps it for 24 hours. An export covers at most 92 days.

Agent Timeline​

Agent Timeline shows one day of every agent's status on a single page: when each agent was Available, Busy (with the busy reason, such as Meeting), On break, or Offline. Only business hours count. Time when you're closed is greyed out, and if business hours aren't configured the whole day is shown. Pick the day with Day; today is the default and refreshes every minute.

Filter the Agent Timeline with:

  • Channel — agents who take tickets on that channel. Agents who are only in backend teams count for every channel.
  • Team Type and Teams — members of those teams. Pick No team in Teams to see agents who aren't in any team.
  • Agents — specific people.

Channel and team filters use each agent's current channels and teams, not the ones they had on that day.

The Coverage strip at the top of the Agent Timeline shows whether anyone could take a new ticket. It's red when no agent in the current filter was Available, amber when agents were Available but every one of them was at capacity, and green when at least one Available agent had room. The cards above it add these up: time with no one available, time when everyone available was full, the most agents available at once, and the average number available.

On the right of each row, the Agent Timeline totals that agent's business hours: time Available (including time at capacity), time at capacity, break time, and busy time. Switch to Compact to see the same day as a grid of 15-minute cells (30-minute cells when business hours are longer than 12 hours), which is easier to scan with many agents. The bottom row of the compact grid shows the fewest agents with a free slot in each cell.

Agents at capacity on the Agent Timeline​

An agent is at capacity when they're Available but can't receive another ticket, because their assigned tickets have reached the channel's ticket limit, including team limits and channel capacity borrowing (see Agent Settings → Ticket capacity). The Agent Timeline shows this time as green with dark stripes. With a Channel selected, capacity means that channel is full. With all channels, an agent counts as at capacity only when every channel they handle is full.

Flowcall checks each Available agent's load every 5 minutes during business hours, so a spike shorter than that may not show. Days before capacity checks started show no capacity.

Tickets that waited on the Agent Timeline​

When Track Agent Assignments is on (see Ticket Behavior → Track agent assignments), red bars under the Coverage strip count the tickets that couldn't be assigned in each 15 minutes, either because no agent was available or because every available agent was full. Hover a bar to see which. These bars follow the Channel and Teams filters; No team counts tickets that had no team.

SLA​

An SLA (service-level agreement) is a promise about how quickly work gets handled — for example, "resolve a ticket within one hour." This page tracks SLA timers and breaches.

You configure SLA timers (see Settings → SLA Rules) with multiple breach levels — for instance, level 1 at 1 hour, level 2 at 5 hours, level 3 at 24 hours. As levels breach, Flowcall can escalate by notifying team leaders or emailing the relevant owners.

Use All sources / channels at the top of the SLA dashboard to select one or more channels: WhatsApp, Email, Live Chat, Instagram, Voice Call, or Manual. The selection applies to the totals, charts, team and task reports, aging reports, and SLA config-wise breach summary. Clear the selection to include all channels. Links to SLA records and aging tickets carry the selected channels into the detail page.

The SLA dashboard counts SLA timers, not tickets. A ticket covered by two SLA rules counts twice, once per rule, and a timer that restarted (for example after the ticket was blocked) counts again. With Active selected, the dashboard shows every SLA someone can act on right now and ignores the date range, so a long-running breach never drops out of view. With Inactive or All, it counts SLAs whose timer started within the date range. Clicking a number opens the SLA page with exactly those timers. The aging reports are a snapshot of today's open tickets: every SLA on a ticket that is still open, including blocked tickets and tickets waiting on a reply, grouped by how old the ticket is. They ignore the date range and the Active / Inactive filter, so old tickets still show in the 8–15, 15–30 and 30+ day buckets; the SLA config and channel filters still apply. Clicking a bucket opens those SLAs on the SLA page. The dashboard can lag live data by up to an hour, so a drill-down may show a few more or fewer.

My Overview​

A personal view for an agent — their own activity and ticket work. It only shows data when you're logged in as an agent. Admins can see the menu entry but it will be empty for them.

How to Use Insights​

  1. Start with Overview and SLA to find operational pressure.
  2. Review CSAT and Issues Reported to understand customer experience.
  3. Use Conversation Audits and Call Recordings to inspect quality directly.
  4. Check Automation Gaps to find workflows that need better instructions, data, or handoff rules.
  5. Turn recurring findings into Remediations, workflow changes, or team coaching — then feed the changes back into Workflows and Settings.