Skip to main content

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 driverSource
Failed runsJob costs × failure rate from Job Costs
Re-runs after data fixPairs of (failed run → fixed re-run) attributed to source DQ issue
Wasted readsQueries against data later flagged as bad (Anomalies)
Auto-remediation costCompute spent by Auto-Remediation playbooks
Manual triageTime logged on Incidents × loaded hourly rate
Downstream impactSelf-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.

DomainCOPQ% of domain computeTrend
Sales$8,40012%
Finance$3,2005%flat
Customer$14,80022%

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