Skip to content

Usage Analytics

Usage Analytics shows how much AI analysis your workspace uses (tokens, requests, and estimated cost). It is separate from Business Metrics (time saved, money saved, errors fixed) and from the Audit Logs (the per-event activity record). Together, the three pages answer different questions:

  • Audit — what happened (events, who did what, when)
  • Metrics — how much value did it deliver (time / money saved, success rate)
  • Usage — how much did the AI cost (tokens, requests, USD)

What is Usage Analytics?

Usage Analytics tracks:

  • Token usage — prompt, completion, and total tokens
  • Cost — estimated cost in USD for AI calls (where shown)
  • Models — which models were used
  • Requests — how many AI analysis calls ran

Accessing Usage Analytics

Usage page (main menu)

  1. Open Usage from the main menu (with Targets, Errors, Metrics, and related items).
  2. Review Plan usage at the top (fixes used vs limit and when the period resets — billing period for paid workspaces, monthly cycle for Personal). Only analyses that reach the AI provider with complete file context (live callback from the target or a stored detection-time snapshot when the site cannot serve a refresh) and return a usable fix suggestion count toward the limit; failed attempts, missing file content, and analyses that produce no fix do not reduce your remaining quota. If your plan includes unlimited fixes when using your own API keys with Profile → AI → Use my own AI keys turned on, the page explains that the monthly fix quota does not apply in that mode (fair-use and other limits still apply).
  3. Open Filters (collapsed by default). Set a date range (presets or custom From/To), optional Tenant / Target, and Group by (for example Day or Target). When you have more than one connector flavour, you can also narrow by Target type, Target language, and Target framework. Superadmins can toggle Include departed tenants; workspaces with their own AI keys can toggle Include BYOK cost estimates.
  4. Use the summary cards, charts, and tables to see cost and token trends.
  5. Use Export on the Usage page to download a CSV of the current view (when available).

Profile → AI (quick summary)

In Profile → AI, the AI Usage & Costs area summarizes usage and links to the full Usage page for charts, tables, and export.

Understanding the data

Token usage

  • Prompt tokens — input to the model
  • Completion tokens — model output
  • Total tokens — prompt + completion

Cost

Costs are estimates based on token counts and the USD per 1 million tokens rate assumptions Patcherly syncs from public pricing sources (per provider and model, typically refreshed weekly). They are not a substitute for your OpenAI, Anthropic, or other provider invoice — use the provider’s billing dashboard for authoritative spend.

Date range

Use presets (e.g. last 7 / 30 / 90 days) or a custom range.

Grouping

Group by Day, Hour, or Target (and other options your workspace shows) to slice usage the way you need.

BYOK and departed tenants

  • Include BYOK cost estimates — when off, charts and tables count only Patcherly platform-paid (system-key) usage. When on, your own API-key traffic is included in the headline totals (estimates only — your provider bills you directly). Workspaces with active own keys may default this on.
  • Patcherly vs your AI keys — when BYOK is included in the headline totals, Usage also shows a comparison card (tokens, cost, requests) so you can see platform-paid vs your-key spend for the same range.
  • Include departed tenants (administrators only) — when on, soft- or hard-deleted tenants still inside the AI-usage retention window stay in aggregates.

Exporting usage data

  1. Go to Usage.
  2. Set date range and grouping, then load the data.
  3. Use Export to download CSV for the current view.

Tips

  • Check usage regularly if you use your own API keys or want to control spend.
  • Use Group by Target to see which apps drive the most analysis.