Skip to content

Guides

Assess customer impact from Jira reports

Use Driver to compare related Jira reports and build a reproducible account-level impact summary. Start with a defined ticket set so another person can check the count.

Define the reports and matching rule

  1. Select the candidate tickets in Jira using the relevant project, time window, and symptom. Give Driver their full URLs, or paste a sanitized ticket extract. The agent needs access to the selected Jira projects to read URLs.
  2. Specify the exact start and end times with a timezone, environment, affected versions, and failure signature. For example: HTTP 429 followed by no webhook retry, in production, during October.
  3. Supply a stable account identifier for each report, including any aliases. Two customer contacts or tickets may belong to the same account. Include an event or request ID when available so repeated reports of one failure can be recognized.

Request the account summary

Ask in Driver:

Review the matching rule and citations before using the totals. Ask Driver to leave missing account IDs or unverified signatures unresolved rather than assigning them to an account or silently excluding them.

An illustrative result: SUP-31 and SUP-32 both report event evt_781 for account acct_acme; SUP-33 reports evt_804 for acct_northwind. If both events match the failure signature, the supplied set supports two affected accounts, two distinct events, and three tickets. It does not establish the number affected outside those reports.

Use the result to prioritize

Compare the customer consequences and workarounds, then copy the reviewed account table, count, matching rule, time window, and unreadable or unresolved tickets into the prioritization issue. Ask engineering to assess the fix against the cited code path; with prepared Advanced context, ask Driver to trace its dependents and candidate tests.

To extend the analysis to runtime records, supply a scoped extract with the same account and event identifiers. Include the record source, population, and query window with any broader count. Accounts that use the same code path but have no observed failure belong in a separate potential-exposure group.