Feature guide

  • Insights — Event counts, unique users, sums, averages, and conversion funnels over customizable date ranges.
  • Funnels — Ordered-event conversion with per-step drop-off; click a step to see who reached it or dropped off.
  • Retention — Weekly cohort × week-offset grid — of the people who did something in a week, how many came back in later weeks. The core "is it sticky?" view.
  • Cohorts — Saved groups of users defined by the events they did (or didn’t) do, reusable as filters across the product.
  • Users — Identified people and their chronological timelines, device, and location.
  • Web analytics — Visitors, sessions, bounce rate, top paths, sources, devices, geography, active hours, and frustrating pages.
  • Session Replay — Consent-aware, sanitized rrweb session reconstruction with project sampling, separate retention, and isolated playback.
  • Errors — Grouped exceptions with occurrences, affected users, stack traces, and browser/OS breakdown.
  • API performance — Per-endpoint latency (p95), error rate, and traffic from $api_request events — slowest, most error-prone, and busiest routes.
  • Alerts — Watch a metric on a rolling window and fire when it crosses a threshold or drops/spikes vs. the previous period; firings can POST to a webhook and/or email recipients or notification groups, with per-recipient delivery status.
  • Feature flags — Deterministic percentage rollouts, per-user overrides, property-targeting segments, and JSON payloads (remote config) evaluated by the SDK.
  • MCP server — Connect Claude Desktop, Cursor, or another MCP client to TraceIQ as an analytics copilot — read funnels, retention, errors, users, and API health, and toggle flags with a personal access token.
  • Webhooks — Forward matching events to an external URL in real time, optionally HMAC-signed.
  • Export — Download raw events as CSV or JSON for a date range, or via the export API.
  • Identity stitching — When an anonymous visitor later identifies, their earlier anonymous events are merged onto the user automatically — one person everywhere.
  • Segmentation — Most views accept filters (os, device, country, region, city, and event properties) so you can slice any metric to a segment.