BayesIQ Data Reliability Audit
Audit Period: December 2025
Fair — significant data quality issues detected
BayesIQ audits a SaaS company's core reporting pipeline covering active users, churn rate, and MRR calculations. The audit surfaces 4 critical issues — including a 434% active user undercount caused by deduplication failures and a churn rate overstated by 84% due to denominator drift. Leadership had been making expansion and retention decisions based on fundamentally wrong numbers. After remediation, the reliability score improves from 31 to 64.
Key metrics: reported vs audited
| Metric | Period | Reported | Audited | Delta |
|---|---|---|---|---|
| active_users | 2025-12 | 1,794 | 9,583 | understated 434.2% |
| churn_rate | 2025-12 | 30.6% | 4.8% | overstated 84.4% |
Top findings
- High102 rows affected
Near-duplicate rows detected
102 duplicate user records inflating MRR calculations and distorting cohort analysis
- High40 rows affected
Null values in required column: user_id
40 records missing user_id — these accounts are invisible to retention tracking
Recommended actions
- 1
Investigate duplicate records. Add dedup logic keyed on non-key fields.
Owner: Data Engineering · Effort: Medium
- 2
Fix null values in required column 'user_id' at the source.
Owner: Data Engineering · Effort: Medium
- 3
Investigate root cause of active_users discrepancy for 2025-12. Check for duplicate events, missing data, or filter logic differences.
Owner: Data Engineering · Effort: Medium
- 4
Investigate root cause of churn_rate discrepancy for 2025-12. Check for duplicate events, missing data, or filter logic differences.
Owner: Data Engineering · Effort: Medium
Prepared by BayesIQ · 2025-12