Triage by consequence
Identify the reports, measures, deadlines, consumers, and decisions that matter most. Record current failure behavior and the cost of delay, disagreement, or manual reconstruction.
This produces a bounded rescue queue. It also prevents a redesign of low-value reports from displacing the few fixes that protect consequential review.
Trace the complete path
For each priority output, follow source acquisition, transformations, model grain, relationships, measures, security, refresh, deployment, and user interaction. Capture ownership and evidence at every transition.
Do not assume the visible layer is the cause. A slow visual may be a model or source-query issue; a disputed KPI may be a source-authority or definition issue.
- Source freshness and contract
- Transformation tests
- Semantic-model structure
- Measure and filter behavior
- Refresh dependencies
- Capacity and query evidence
- Access and distribution
- Release and rollback
Choose keep, repair, consolidate, replace, or retire
Every artifact should receive an explicit disposition with rationale, dependency, owner, and timing. Replacement is one option, not the default.
A repaired estate needs an approved baseline for refresh, performance, reconciliation, access, deployment, rollback, and support. Without that baseline, the next change can recreate the same fragility.
Leave an operating record
The rescue closes with documented ownership, known defects, deferred work, monitoring, release evidence, and a prioritized improvement backlog. The objective is not merely a working report today. It is a system the responsible team can keep working.
Printable worksheet
Reporting estate triage record
Use the browser print command to save or print this ungated worksheet. Record observed evidence and open questions rather than assigning a generic score.
| Area | Questions | Evidence to collect |
|---|---|---|
| Business consequence | Which decisions, deadlines, consumers, and measures are affected? What happens when the output is late, disputed, or unavailable? | Review calendar, priority outputs, accountable sponsor, materiality notes |
| Source and acquisition | What is authoritative? What is the natural grain? How do late, corrected, missing, or duplicated records behave? | Source contract, extracts, row counts, business keys, freshness history |
| Transformation | Where do business rules change the source? Which steps are repeated, manual, untested, or dependent on one person? | Queries, scripts, pipelines, mapping rules, reconciliation checks |
| Semantic model | Are relationships, measures, time behavior, security, and reusable definitions explicit and appropriate for the grain? | Model structure, DAX or equivalent logic, metric definitions, access roles |
| Report and interaction | Which pages, filters, visuals, exports, or user actions are slow, confusing, duplicated, or no longer used? | Representative user actions, usage evidence, performance traces, owner interviews |
| Refresh and operation | What must run, in what order, by what deadline, and who responds when it fails? | Schedules, dependencies, capacity history, alerts, incident records, support ownership |
| Access and distribution | Who may see which information? Are exports, subscriptions, sharing, and role changes controlled and reviewable? | Access matrix, role tests, distribution list, review cadence |
| Release and recovery | How are changes reviewed, tested, deployed, rolled back, documented, and accepted? | Release record, test results, reconciliation, rollback evidence, known defects |
| Disposition | Should each asset be kept, repaired, consolidated, replaced, or retired? What dependency and owner control that decision? | Disposition register, rationale, priority, dependency, owner, target release |
Primary sources
Related services and experience
- Power BI health check (Independent assessment)
- Power BI performance optimization (Targeted repair)
- Reporting and BI Rescue (Broader implementation)
Need help with this system?
Reporting and BI Rescue turns the diagnostic into a bounded repair, rationalization, and operating baseline.
Review Reporting and BI Rescue