SQL, data modeling, data quality, EDA, visualization · insight-0065
Reconcile a new analytical table to an independent trusted total before anyone uses it
- Evidence type
- practitioner heuristic
- Difficulty
- easy
- Roles
- Data analyst, Data scientist
- Industries
- aviation, banking, retail
Summary: situation, action, reason, boundary
Situation
You have built a new revenue or passengers table for analysis or model training, and it will feed decisions.
Action
Pick an independent reference (finance revenue, ticketing passenger counts, bank statement totals), reconcile by period and major segment, and document the tolerance and any explained differences such as timing, currency or refunds. Rerun the reconciliation after each logic change.
Reason
Internal consistency checks cannot catch errors that are consistent; an external total can.
Boundary
KPI tables, training datasets, executive dashboards and regulatory reports.
Symptom or risk
Join, filter and deduplication mistakes produce totals that differ from finance or operations by a few percent, and the difference surfaces only after a decision.
Exceptions
Metrics with no independent source; compare two methods of computing them instead.
Trade-offs
Reconciliation needs access to finance data and time to explain small differences.
Pitfalls
Reconciling only the grand total, where segment errors can offset each other.