What happens when the numbers across business systems don't agree? I used SQL and Power BI to investigate £5.19M in payment discrepancies, £1.08M in duplicated CRM value - and discovered why not finding a neat root cause can sometimes be the most important analytical finding.