Case: running a pixel AND a postback double-counted conversions by 31%

Case: running a pixel AND a postback double-counted conversions by 31%

The red flag: tracker conversions exceeded the network's payout count by roughly a third, every single day. ROI looked great until reconciliation against the statement.

The cause, traced step by step:
1. The offer page fired a client-side conversion pixel
2. The network also fired a server-side S2S postback for the same event
3. The tracker counted both — one real conversion logged twice
4. The pixel also fired on page refreshes and back-button hits, adding phantom conversions on top

The cleanup:
— Picked one source of truth: the S2S postback (server-side, can't be faked by a refresh)
— Removed the client-side conversion pixel entirely from the flow
— Kept a fired-pixel only for the source's own optimization, but stopped feeding it into the tracker's conversion count
— Re-imported 14 days from the network's export to set a clean baseline

Outcome: tracker conversions dropped 31% and matched the payout statement within 1%. Two campaigns that looked like 1.8 ROI were actually 1.25 — still profitable, but scaled-back to reflect reality before they ate a bigger budget.

One conversion, one source of truth. Postback or pixel, never both into the same count.

Save this SOP. Run this before every launch.