Skipping Data Validation: The Mistake That Makes Your Dashboard Lie to You
Declan Farquhar had been running his small consultancy for six years when he finally set up an automated financial dashboard. It pulled figures from three sources: his accounting software, a bank feed, and a spreadsheet his bookkeeper updated weekly. For two months, the dashboard showed healthy cash flow. Then his accountant found a £14,000 discrepancy. The spreadsheet had a broken formula. The dashboard never flagged it.
Where the validation gap actually sits
Most people assume automation means accuracy. It does not. Automation means consistency - it will consistently repeat whatever errors exist in the source data. The common mistake is treating the dashboard as the verification step, when it is actually the output step. Validation needs to happen before data reaches the dashboard, not after.
Specific errors that appear most often
- Duplicate transaction entries from overlapping bank feed imports
- Currency conversion applied twice when multi-currency accounts sync
- Null values treated as zero in aggregation formulas
- Date field mismatches between fiscal and calendar year settings
Declan added a simple reconciliation rule: each Monday, the dashboard compares its total against a manually confirmed bank balance. If the variance exceeds 1% of monthly turnover, it flags the row in red and pauses the automated report. It took about three hours to configure. Since then, two errors have been caught before anyone saw the numbers.
What a basic validation layer looks like
A validation layer does not need to be complex. It can be a conditional formula that checks whether the sum of category totals equals the reported grand total, or a timestamp check confirming the data feed updated within the last 24 hours. The point is having a defined rule that fails loudly rather than passing silently.