Cost of Poor Data Quality (COPQ)
The COPQ page at /finops/copq quantifies the financial impact of data quality problems — wasted compute on bad data, retries from failed pipelines, manual remediation effort, and downstream business cost from incorrect outputs.
Why measure it
DQ teams often struggle to justify investment because the business impact is fuzzy. COPQ converts that fuzziness into dollars: "we burned $42k last quarter on jobs that ran on stale data and had to be re-run".
What's tracked
| Cost driver | Source |
|---|---|
| Failed runs | Job costs × failure rate from Job Costs |
| Re-runs after data fix | Pairs of (failed run → fixed re-run) attributed to source DQ issue |
| Wasted reads | Queries against data later flagged as bad (Anomalies) |
| Auto-remediation cost | Compute spent by Auto-Remediation playbooks |
| Manual triage | Time logged on Incidents × loaded hourly rate |
| Downstream impact | Self-reported business cost on incidents (optional) |
Cards
- COPQ this period — total $ across all drivers
- % of total compute — COPQ / total compute cost
- Top driver — biggest contributor
- MoM trend — improving / worsening
Driver breakdown
Stacked bar chart over time showing each driver's share. Click any segment to drill into the underlying anomalies / incidents / failed runs.
Per-domain rollup
A table breaking COPQ by business domain. Useful for showing each team their slice of the bill.
| Domain | COPQ | % of domain compute | Trend |
|---|---|---|---|
| Sales | $8,400 | 12% | ▼ |
| Finance | $3,200 | 5% | flat |
| Customer | $14,800 | 22% | ▲ |
Reducing COPQ
The page surfaces specific actions:
- High-fail rules — rules that fail often without leading to fixes; tune or remove
- Recurring playbook successes — automate the manual fix steps
- Top incident sources — invest in upstream stability
API
GET /finops/copq?from=2026-04-01&to=2026-04-30
GET /finops/copq/drivers
GET /finops/copq/by-domain
Related
- Incidents — feed time-cost into COPQ
- Auto-Remediation — playbook costs counted here
- DQ Scorecard