Skip to content

article

Tracking API Usage and Performance

How to read the Usage view's 30-day metrics — active keys, webhooks, request volume, latency, errors, and breakdowns by status and scope.

Overview

The Usage view in the API console is your observability dashboard for the StretchDev API platform. It summarizes a rolling 30-day window of activity so you can see how heavily your integrations are used, how fast they respond, and how often they fail — all scoped to your organization.

What the Usage view shows

1. Open the API console and go to Usage. You'll see:
- Active API keys — how many keys are currently active.
- Active webhooks — how many webhook subscriptions are currently active.
- Webhook deliveries (30d) — total delivery attempts in the last 30 days.
- Requests (30d) — total API requests in the last 30 days.
- Average latency — the mean request duration, in milliseconds, over the window.
- Errors (30d) — how many requests returned a 4xx or 5xx status.
- Requests by status — a breakdown of request volume by HTTP status code.
- Requests by scope — a breakdown of request volume by the scope used (for example projects:read), so you can see which capabilities your integrations lean on.

How to read it

  • Volume: Requests (30d) and Webhook deliveries (30d) tell you how busy the platform is. A sudden jump can mean a new integration or a runaway loop.
  • Health: compare Errors (30d) against total requests to get your error rate. A rising error rate is an early warning that a key or endpoint is misconfigured.
  • Performance: Average latency trends up when your endpoints or queries slow down; watch it alongside the request-log durations.
  • Shape: Requests by status shows the mix of successes and failures at a glance, and Requests by scope shows what your integrations actually do — useful when deciding which scopes a key truly needs.

Example

At month-end you open Usage. It reports 4 active keys, 2 active webhooks, 1,120 requests (30d), 31 errors (30d), and 142 ms average latency. The by-status breakdown shows most traffic as 200 with a small band of 401s; the by-scope breakdown shows nearly all traffic on projects:read. The 401s line up with a rotated key from last week (confirmed in Request logs), and the scope breakdown tells you a write-scoped key you issued is barely used — a candidate to pause. You've turned raw numbers into two concrete actions.

Tips

  • Track error rate (Errors ÷ Requests) over time, not just the raw error count — volume changes month to month.
  • Use Requests by scope to right-size keys: scopes that see little or no traffic can often be removed.
  • If Average latency climbs, cross-reference Request logs durations to find the slow paths.
  • Combine Usage with Deliveries and Request logs: Usage tells you that something changed; the logs tell you what.

FAQ

What time window does Usage cover? A rolling 30 days from now — older activity ages out of the counts automatically.

Is Usage scoped to my organization? Yes. Every metric counts only your organization's keys, webhooks, deliveries, and requests.

Why is a metric zero? Either there was no matching activity in the last 30 days, or (for active keys/webhooks) none are currently in the active status.

Was this helpful?

Help us improve this article

Use these controls to share whether this answer solved the issue. Feedback helps prioritize updates to StretchSuite Support.