Skip to content
E-commerce · 2026

Fixing the tracking before touching the ad account

Reported orders were a third off real ones. Everything downstream was being optimised against fiction.

Cost per order
-36%
Reporting variance, was 34%
3%
Attributed revenue
+58%
Blended ROAS
4.4x

The challenge

The account looked like it was performing. It was not. Purchases were tracked on the thank-you page, which customers frequently never reached after a bKash redirect, so a large share of real orders went uncounted — and the algorithm was learning from the wrong ones.

What we did

We moved conversion tracking to the payment callback, added the Conversions API alongside the pixel, and validated a full week against the order table before trusting a single number. Only then did we restructure campaigns around what the corrected data showed.

The result

  • -36%

    Cost per order

  • 3%

    Reporting variance, was 34%

  • +58%

    Attributed revenue

  • 4.4x

    Blended ROAS

Ready to turn your business into a brand?

Book a free 90-minute growth audit. We will look at your site, your analytics and your ad accounts, then tell you exactly where the money is leaking — whether or not you hire us.

Or email us at hello@froxit.bd