My tracker said 1.3. The network said 0.9. The truth cost me a month.

My tracker said 1.3. The network said 0.9. The truth cost me a month.

The setup. CPA campaign where my self-hosted tracker and the network dashboard never agreed. I'd been trusting mine. Illustrative numbers.

The move. I scaled on my tracker's 1.3 ROI for a month, assuming the network was just slow to report.

The numbers. The gap was a 22% conversion discrepancy. My tracker fired a conversion on the postback; the network only PAID on ones that passed fraud and dedup. About 1 in 5 of my "conversions" were duplicates or fraud-flagged and never paid. Real ROI was 0.9, not 1.3. One month of scaling = ~$6,000 spent against money that didn't exist on the invoice.

The lesson. Your tracker measures what fired. The network measures what they'll pay. Those are different numbers and the difference is always in their favor. The invoice is the only ground truth.

What I'd do differently. Reconcile tracker-to-network weekly and treat any gap over 5% as a red flag, not noise. Scale on the LOWER of the two numbers, always. I now keep a running "discrepancy rate" per network — it's also how I caught a network that was systematically under-reporting and quietly dropped them.