Measure one scope
Do not mix platforms, regions, versions, and traffic classes without a documented weighting rule.
Project service targetsAvailability, support hours, response targets, measurement methods, and service remedies are defined for the selected endpoint scope and delivery plan.
CONTRACT CHECKLIST
Metrics to define together
| Metric | Definition to agree | Evidence |
|---|---|---|
| Availability | Included endpoints, region, time window, exclusions, and formula | Named monitor, timestamps, request IDs, and raw results |
| Latency | Percentile, timeout, sample size, endpoint mix, and region | Fixed probes and period report |
| Data acceptance | Required fields, freshness, completeness, empty and partial results | Accepted samples and field-level checks |
| Support | Service window, severity, initial response, update cadence, escalation | Ticket timestamps and agreed channel |
| Capacity | Average and peak traffic, burst, concurrency, queueing, rate limit | Gateway metrics and project traffic log |
| Remedy | Eligibility, calculation, cap, claim window, and exclusions | Monthly evidence package |
Preserve evidence
Keep synchronized timestamps, route, status, latency, request ID where available, and the monitoring region.
Review changes
Confirm how new endpoints, migrations, maintenance, and client-side errors affect the measurement.