Skip to main content

Statistics

The Statistics dashboard is Polygent's answer to "what's the team actually shipping?" — a platform-wide analytics view of sessions, tickets, costs, and user activity. Built for managers and stakeholders who need numbers, not status meetings.

Layout

The dashboard is organized into 6 tabs. The workspace and date-range filters always sit above the tabs and apply to whichever tab is open. Each tab fetches its data the first time you open it, so switching tabs is instant after the first load.

TabWhat's inside
OverviewHeadline summary cards with prior-period comparisons
TicketsTicket cycle metrics + breakdown by type / source / priority
SessionsSessions, Bots, Automations, Session Performance
User ActivityActive users, top contributors, per-user averages
More MetricsRound Tables, Insights, Workflows, Hook Tasks
Cost & TokensSpend, token usage, cost by provider and model

Summary cards

The top of the dashboard shows headline metrics for the selected filter range, each with a prior-period comparison:

  • Total sessions
  • Completed tickets
  • Active users
  • Automation runs
  • Total insights
  • Workflow runs
  • Bot conversations
  • Total cost (USD)
  • Automation success rate

Session statistics

PanelWhat it shows
Count by StatusWaiting, In Progress, Done, Canceled
Source BreakdownManual Chat / Manual Develop / Ticket / per-Bot / per-Automation
Provider DistributionSessions by agent provider (Claude Code, Gemini, etc.)
Daily TrendSessions started per day
Duration StatsAvg / Median / P95 minutes for completed sessions
Messages StatsAvg / Median / P95 messages per session
PR RateCompleted sessions that produced a PR
State CountsActive (in-flight), Idle (>7d in non-terminal state), Errored count
Outcome RatesSuccess vs. cancellation ratios
Top Sessions by CostTop 5 sessions in the window ranked by aggregate USD cost
Top Sessions by MessagesTop 5 sessions in the window ranked by message count

Ticket metrics

PanelWhat it shows
Stage DistributionHow tickets are spread across the pipeline right now
Daily TrendTickets created/completed per day
Completion RateWhat percentage finishes vs. fails
Cycle Time (Avg / P95)From Pending to first Completed; P95 highlights long-tail outliers
Time to First ActionAvg time from ticket creation until it exits the Pending stage
Avg Implementation TimeAvg time spent in the Implementation stage
Avg QA TimeAvg time spent in the QA Approval stage
Avg PR Review TimeAvg time spent in the Pull Request stage
ReopenedTickets that entered Completed more than once (rework signal)
QA Rejection RateHow often QA sends tickets back
Lifecycle SnapshotOpen / In Progress / Blocked counts (current state, ignores filter)
Avg Blocked AgeAvg age (days) of currently blocked tickets — flags long-stuck work
Breakdown switcherChoose Type, Source, Priority or Assignment for the breakdown pie. Source values are labelled Tickets Page / Import / Created from Plan / Created from Insight.
Open Ticket AgeAging buckets (< 1d / 1-3d / 3-7d / 1-2w / > 2w) for backlog

Above the breakdown pie, two filters narrow the breakdowns (Type / Source / Priority / Assignment):

  • Status — All Tickets / Open / Closed.
  • Tags — a multi-select of the tags in use across your workspaces. Pick one or more to scope the breakdowns to tickets carrying any of the selected tags; leave it empty to include every ticket. The Tags filter is hidden when no tags exist.

Both filters affect only the breakdown pies — the cycle/throughput stats and the lifecycle/aging snapshots ignore them.

Automation metrics

PanelWhat it shows
Total RunsAll automation executions in the range
Success RateCompleted / (Completed + Failed)
Duration (Avg / P95)Mean and P95 run time across completed executions
Total CostAggregate USD across automation-linked sessions
Avg Cost / RunTotal cost divided by total executions in the window
Cost Over TimeDaily automation cost trend for the window
Enabled / DisabledCurrent count of automation definitions
Never RunAutomations defined but never executed (snapshot)
Stale Past-DueEnabled automations whose next run is over an hour overdue
Schedule TypesOne-Time, Recurring, Loop, Manual, HTTP Trigger
Session TypesDevelop, Chat, Bot
Provider BreakdownExecutions by agent provider
Daily TrendStarted / Completed / Failed runs per day
Most UsedTop automations by execution count, with success rate
Top FailingAutomations with the highest failure counts

Bot usage

  • Most Used Bots — top bots by conversation count
  • Daily Trend — bot conversations per day

Workflow metrics

  • Most Used Workflows
  • Success Rates by workflow
  • Daily Trend

Insight distribution

  • By Type — Documentation, Skill, Process Improvement, Performance, User Experience, Nice to Have, Other
  • Daily Trend

Cost and tokens

PanelWhat it shows
Total CostAggregate USD across all providers
Input / Output / Cache TokensToken breakdown
Cost per TicketAverage cost to complete a ticket
Daily TrendCost over time
Cost by ProviderClaude Code vs. Gemini vs. others
Top Cost ModelsWhich models cost the most

Use this to spot cost regressions early or to forecast monthly spend.

User activity

  • Active User Count (with prior-period comparison) — exact set-union of session creators, ticket creators, and round-table participants
  • Workspace Members / Inactive / Activation Rate — total members in scope, count with no activity in the window, and the active/total ratio
  • Per-active-user averages — sessions, tickets created, and tickets completed per active user in the window
  • User search — filter the active rank list by name
  • Rank switcher — pick a single dimension or the combined view:
    • Combined Activity — sum of sessions, tickets created, tickets completed, automation runs (per-user)
    • Sessions
    • Tickets Created
    • Tickets Completed — credits the assigned developer for tickets that entered the Completed stage in the window. Tickets with no assigned developer still count toward total completions but are not attributed to anyone here.
    • Automation Authors — users whose automations drove the most executions in the window
  • Daily Trend

Session performance

PanelWhat it shows
Average / Median DurationHow long sessions take
Average MessagesConversation length
API LatencyProvider response times
PR Creation RateSessions that produce PRs
Duration by ModeChat vs. Develop vs. Bot
Provider DistributionHow sessions split across providers
Success by ProviderWhich providers complete most reliably
Daily TrendPerformance over time

Hook task metrics

Per-hook execution statistics — how often hooks fire, fail, retry, and how long they take. Use this to find broken or slow hooks dragging down your team.

Filters

Two filters apply to every panel:

  • Workspace — All workspaces or a specific one
  • Date Range — Last 7 / 30 / 90 days, This Month, or a custom range

Performance

Statistics are cached for 30 seconds server-side, keeping the dashboard responsive even with large datasets. Results refresh automatically within that window.

Examples

Forecast next month's AI spend Open Cost & Tokens, set the range to Last 30 Days, and read total cost plus cost-by-provider. Combine it with Cost per Ticket to estimate spend for a planned batch of tickets — then set a budget accordingly.

Find what's slowing the team down On Tickets, check QA Rejection Rate and Avg Implementation Time. A high rejection rate points at unclear specs (tighten the Planner step); a long implementation time on one provider suggests trying a different model.

Catch broken hooks The Hook Task metrics show which hooks fail or retry most. A hook with a high failure rate is dragging every session in that workspace — fix or disable it.

Report shipping velocity Use Overview with prior-period comparison to answer "are we shipping more than last month?" — completed tickets, PR rate, and active users at a glance, no status meeting needed.

Permissions

PermissionCapability
statistics.viewView the dashboard

The dashboard reflects what's actually happening in real time — no separate analytics setup required.