We found the leak before we spent another pound chasing it — 3.8x to 12.6x reported ROAS, at unchanged media spend.

Glamy Lab’s reported return looked weak, and the instinct was to change the campaigns. The brief was to find out why the numbers were low before spending another pound chasing them.

Diagnose before you spend. We treated it as a measurement problem first and a media problem second, and held spend constant so any change could be attributed to the fix rather than to new budget.

• Tracking diagnosis
• Server-side event repair
• Catalogue & customer-file rebuild
• Paid media management

The real issue was a pixel deduplication failure: browser and server events were arriving with mismatched identifiers, so genuine purchases were being discarded rather than matched to the ads that drove them.

1. 01Diagnosed the deduplication failure at the tracking layer before touching the media plan.
2. 02Rebuilt the collaborative-ad catalogue descriptors and normalised the customer file, recovering 969 valid records from roughly 1,970 raw rows.
3. 03Left media spend unchanged while the fix took effect, to isolate the impact of the tracking correction from any new budget.