[FEATURE] Per-user cross-surface analytics (user_cost_report) available below Enterprise

Resolved 💬 2 comments Opened Jun 18, 2026 by 0xDEnYO Closed Jun 21, 2026

Problem Statement

Mid-sized organizations (we're ~123 people on the Team plan) need to understand who is spending what across Claude surfaces — chat, Claude Code, Cowork — to manage budgets and justify spend internally. Today the only API that provides per-user, cross-surface cost/engagement data is the Enterprise Analytics API (/v1/organizations/analytics/user_cost_report, provisioned via an analytics key at claude.ai/analytics/api-keys). That key-provisioning page does not exist on Team or individual accounts, so the entire surface is Enterprise-gated.

What's available to us on Team via the Admin API is valuable but insufficient for cost attribution:

  • Usage & Cost API — spend/tokens by workspace, not by person.
  • Claude Code Analytics API — per-user, but Claude Code only, and (per #27780) excludes subscription-seat users, which is how our whole team signs in.

So a 100+-seat Team customer has no first-class way to answer "what is each person/team costing us across Claude?" — the exact question that drives renewal and expansion decisions.

Proposed Solution

Any one of:

  1. Make the per-user Analytics API (user_cost_report) available on the Team plan (read-only, admin-gated), or
  2. Offer a lighter "Team Analytics" key with per-user cost across chat + Code + Cowork, or
  3. Add a group_by[]=user (or actor) dimension to the existing Admin Usage & Cost API so Team orgs can attribute spend per person without Enterprise.

Why It Matters

Per-user attribution is the data that lets a growing org self-justify and expand Claude adoption. Gating it entirely behind Enterprise leaves mid-market customers — precisely the ones deciding whether to scale up — flying blind on cost.

Related: #27780 (subscription/OAuth users missing from the Claude Code Analytics API), #33978 (built-in usage analytics command).

View original on GitHub ↗

This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗