How to run the Sentry Explore dataset field map

Use this when scripts or third-party integrations query Sentry Explore (/events/, /events-timeseries/, /replay-count/) with transactions, discover, or events datasets—or omit dataset and inherit discover—before the 2026-12-03 API cutoff. Local only; re-check Sentry Help Center.

Step 1. Confirm the December 3 2026 cutoff on the Help Center article. Note UI and SDKs are unaffected; direct API requests are in scope.
Step 2. Inventory scripts hitting the listed endpoints, including calls that omit dataset (default discover) and replay-count data_source values.
Step 3. Open the home map: pick legacy dataset class and whether queries mix errors with transactions.
Step 4. Choose dataset=spans for performance/transaction-shaped work or dataset=errors for errors. Split former discover mixes into two requests.
Step 5. Add is_transaction:true when you need transaction-shaped spans rather than all spans.
Step 6. Rewrite fields and functions using the exported map (duration, exclusive_time, percentile→p95, count→count(span.duration), apdex threshold, …).
Step 7. Remove unsupported spans functions (any, count_miserable, linear_regression, corr, total.count usages). Fix replay data_source.
Step 8. Export Markdown into the ticket; compare against searchable span properties; smoke queries; expect possible count differences from sampling extrapolation.

Related: FAQ · Use cases · Home tool. Always prefer official vendor documentation for authoritative dates.