The situation
Nobody trusted the monthly report.
Brightpath Outdoor is a fictional direct-to-consumer gear brand. Its four-person marketing team rebuilt the monthly report by hand, which took about two days every month.
Leadership had stopped trusting the numbers, and Finance and Marketing disagreed on how many customers were active.
What the audit found
Four problems behind the numbers.
- No formulas anywhere. Every number was retyped by hand.
- Bad orders counted as sales. Test, refunded and duplicate orders inflated revenue.
- Stale ad spend. Meta spend was pulled on the 27th and never refreshed.
- Duplicate contacts. 45 duplicate CRM records inflated customer counts.
The numbers
September 2026: hand-built report vs corrected.
| Metric | Hand-built | Corrected | Main cause |
|---|---|---|---|
| Revenue | $40,752.62 | $35,368.76 | Tests, refunds, duplicates |
| Orders | 219 | 194 | Same three causes |
| ROAS | 1.77x | 1.48x | Revenue over, spend under |
| Ad spend | $23,012.17 | $23,874.03 | Stale Meta pull |
| New contacts | 58 | 53 | Duplicate CRM records |
| Email subscribers | 189 | 206 | August figure carried over |
| Active customers | 275 | 204 / 255 / 142 | Matched no agreed definition |
Active customers are shown under the three definitions in use: 90-day, 12-month and CRM flag. Revenue and orders are net of refunds.
What the client received
A report that checks itself.
Audit and prioritized fix list
Cleaning script with a full change log
Workbook where every KPI is a live formula
Scheduled run that stops if any check fails
Runbook for a 10-minute monthly process
Decisions handed back
Business definitions stay with the business.
Some questions aren't data problems. These went back to the client to decide and record:
- Which active-customer definition to use
- Whether refunds reduce revenue and orders
- How email subscribers should be measured
- Whether email platform cost counts in ROAS
Illustrative sample. All names, figures and data are synthetic.