Data Quality¶
Overview¶
Poor data quality destroys trust in value reporting. This page describes how to validate inputs, maintain source credibility, and trace data lineage so every number is defensible.
Who this is for¶
- End User
- Admin
- Developer
Prerequisites¶
- Access to Administration > Integrations and Data Explorer.
- Understanding of source systems (CRM, ERP, spreadsheet).
- Defined data owners for each integration.
Step-by-step instructions¶
1. Validate at the point of entry¶
- Enable required field validation on custom fields.
- Set data-type rules (numeric, date, enum) to prevent free-text pollution.
- Use dropdowns instead of open text where possible.
2. Assign data owners¶
- For each integration, name a business owner and a technical owner.
- Document the owner in the connector settings description field.
- Review ownership quarterly during portfolio reviews.
3. Monitor freshness¶
- Set a freshness SLA for each source: daily, weekly, or real-time.
- Configure alerts in Administration > Integrations > Health when a source exceeds its SLA.
- Treat stale data as a P2 incident if it blocks month-end reporting.
4. Maintain lineage¶
- Ensure every actual includes a source system reference.
- For manual entries, require an evidence link or attestation.
- Use the Lineage tab in the initiative detail view to trace a number back to its origin.
5. Run periodic quality audits¶
- Quarterly, export a sample of 20 initiatives.
- Check for null required fields, duplicate records, and orphaned actuals.
- Score each initiative red, amber, or green.
- Publish the scorecard to initiative owners.
Permissions required¶
| Role | Permission | Scope |
|---|---|---|
| Admin | Configure validation rules | Organization |
| Admin | View data health | Organization |
| User | Edit data in assigned initiatives | Assigned initiatives |
Limits and guardrails¶
- Limit Custom validation rules per entity: 20.
- Limit Freshness SLA minimum: 1 hour.
- Limit Lineage depth displayed: 5 hops.
Troubleshooting¶
Issue: Duplicate records after a sync
Cause: The source system changed a unique identifier, or the deduplication key was too narrow. Resolution: 1. Identify the duplicate set in Data Explorer. 2. Update the deduplication key to include a composite of email + source_id. 3. Merge duplicates and re-run the sync.
Issue: Manual actuals lack evidence links
Cause: Users skipped the field because it was optional. Resolution: 1. Make the evidence link required for manual entries. 2. Train users that an attestation is acceptable when a system link is unavailable.
Related pages¶
Escalation path¶
| Severity | Condition | Contact |
|---|---|---|
| General | Data quality coaching | Customer Success Manager |
| Urgent | Source system corruption affecting many records | support@valuepact.ai |