Skip to content

Analytics Engine

Kiket provides analytics as code through the Domain-Specific Analytics Language (DSAL) and configurable dashboards.

DSAL Overview

  • DSAL files (usually analytics/*.kql) describe metrics, alerts, and visualisations.
  • Queries run against the analytics warehouse populated from issue, workflow, and extension events.
  • Syntax mirrors SQL but adds productivity shortcuts like time buckets and state duration helpers.

Example metric: 

DEFINE METRIC cycle_time AS
  SELECT
    issue_key,
    duration_in_state('in_progress') AS active_time,
    duration_between('backlog', 'done') AS total_time
  FROM issues
  WHERE project = $project

Dashboards

  • YAML definitions map metrics to charts, tables, and goals.
  • Dashboards inherit organization defaults. Projects can overlay additional widgets via repository-level files.
  • Permissioning follows project membership; sensitive widgets can require explicit roles.

Alerts & Insights

  • Metrics can declare thresholds or anomaly detection, sending notifications through email, Slack, or extensions.
  • AI Insights (see Analytics → AI Insights) summarise trends and surface recommended actions.

Extending Analytics

  • Extensions can push custom events into the analytics pipeline through secured endpoints.
  • DSAL supports IMPORT statements to join external data sources in a controlled manner.

Analytics as code keeps reporting consistent and reviewable just like workflows and boards.

Predictive Analytics

Kiket's AI-powered analytics go beyond historical reporting to forecast trends and surface actionable insights:

  • Cycle time forecasting: Predict delivery dates based on historical velocity and current workload
  • Bottleneck detection: Identify workflow states where issues stall before they become blockers
  • Capacity planning: Model team workload to prevent overallocation
  • Anomaly detection: Flag unusual patterns in metrics that may indicate process issues

Predictive models train on your organization's data and improve over time. See AI Insights for configuration options.

Risk Intelligence Widgets

  • The Risk Scanner agent continuously evaluates stalled work, incident recurrence, deployment regressions, quota usage, and quality bursts.
  • Executive dashboards now include a “Risk intelligence” panel summarising alert volume by severity and the top contributing categories, with quick links to remediation targets.
  • Project overview pages surface a scoped risk card so teams can investigate blockers without leaving the project context.
  • The Risk Alerts log (/risk_alerts) offers filters by project, severity, category, and status plus direct links to the underlying issue or workflow action for fast drill-down.
  • Dismissals capture a structured reason (noise, duplicate, mitigation, won’t-fix, etc.) so teams can report on alert quality and tune thresholds over time.
  • Leads can acknowledge, dismiss, or reopen alerts directly from the widget or log; every action captures who responded and when, keeping follow-up transparent.