Analytics

The dashboard’s analytics tab — revenue, members, classes, and workouts metrics for an org’s owner and admins. Implemented as a wide, deliberately denormalized read surface over the operational tables.

What

GET /organizations/:orgId/analytics/... returns per-area summaries and trend series:

Surface	Endpoint	Service method
Revenue summary	`GET /analytics/revenue/summary`	`getRevenueSummary`
Revenue trend	`GET /analytics/revenue/trend?months=12`	`getRevenueTrend`
Plan distribution	`GET /analytics/revenue/plan-distribution`	`getPlanDistribution`
Members summary	`GET /analytics/members/summary`	`getMembersSummary`
Members growth	`GET /analytics/members/growth?months=6`	`getMembersGrowth`
Members activation	`GET /analytics/members/activation`	`getMembersActivation`
At-risk members	`GET /analytics/members/at-risk?sample=5`	`getAtRiskMembers`
Popular classes	`GET /analytics/classes/popular?days=30&limit=10`	`getPopularClasses`
Class utilization	`GET /analytics/classes/utilization?days=30`	`getClassUtilization`
Workouts summary	`GET /analytics/workouts/summary`	`getWorkoutsSummary`
Coach overview	`GET /analytics/coaches/overview`	`getCoachOverview`
CSV export	`GET /analytics/export?tab=<tab>`	per-tab payload → `toCsv`

Why

Owners track the gym as a business. Without revenue, MRR, debt, churn, and class utilization in one place, they fall back on spreadsheets.
The same metrics power the Spotter agent’s analytics.* tool family — agent answers like “how’s the gym doing?” pull from this service.
Keeping every query close to the operational tables means metrics never lag — there is no nightly aggregation job that could drift.

Who

Owner / admin — full access (role-gated server-side).
Coach / member — no access (403).
Spotter agent — staff-only via the agent tools.

Persona impact

Persona	Impact
Owner	Single-pane view on revenue, growth, retention, programming engagement. CSV export for each tab to share with accountant/business advisor.
Admin	Same as owner.
Coach	Coach Overview (separate endpoint) gives them their own students-and-engagement view; the broader analytics tab is locked.

Capabilities

Revenue

MRR computation normalized by plan interval (INTERVAL_MONTHS mapping in analytics.service.ts).
Month-over-month delta percentages on MRR and collected revenue.
Outstanding receivables (subscriptions with debt > 0).
12-month trend by month-start bucket; configurable horizon.
Plan distribution: active subs per plan with monthly revenue contribution.

Members

Total / active / activated-active counts (the activatedActive discriminator distinguishes Clerk-linked from CSV-imported members).
New + churned this month with MoM deltas.
Activation funnel (imported → invited → activated).
At-risk sample (past-due subscriptions) with ?sample=N cap.
Growth trend over configurable months.

Classes

Most-booked class types over a window.
Capacity fill rate per class type plus a roll-up.

Workouts

Assignments scheduled / completed MTD, PRs MTD.

Coach overview

Per-coach panel: assigned athletes, recent activity, programming load.

CSV exports

Per-tab synchronous CSV via GET /analytics/export?tab=... — handler builds the same rows the dashboard renders, then writes them as CSV with Content-Disposition: attachment.

insights — small ranked widget; the agent’s org_insights tool delegates to it.
payments / subscriptions — source of all revenue metrics.
memberships / users — source of growth + activation metrics.
class-sessions / bookings / class-types — source of class metrics.
workouts / workout-assignments / workout-results — source of workout metrics.
spotter-agent — analytics.* tools mirror these endpoints 1:1 for the agent.

Status

Shipped. New endpoints land as the dashboard grows. CSV export covers the most-requested tabs.

Gaps

All metrics are computed on-read with no caching layer. A 5000-member org with 5 years of payment history will see latency creep. Pre-aggregated rollups remain future work.
No cohort retention, LTV, or churn curve — only point-in-time counts and MoM deltas.
Class utilization is over a fixed past window — no “next 30 days at current bookings pace” forecast.
CSV exports are synchronous; very large orgs may need the async queue path the members export uses.
No PostHog/Mixpanel-style event funnel — pure SQL aggregations on operational tables.