Georgia-Pacific’s programmatic buys appeared healthy on the surface, with strong reported viewability and low fraud flags from standard verification tools, but internal scrutiny suggested a disconnect between these media quality metrics and actual sales impact, indicating that non-human traffic and low-quality impressions were likely inflating performance reports.
By in-housing programmatic operations, consolidating on a single DSP, and benchmarking against the ANA Programmatic Transparency Initiative, Georgia-Pacific and Yahoo found that a meaningful portion of spend was tied to impression types and supply paths that did not correlate with incremental sales, despite appearing ‘clean’ in legacy reports. They identified over-reliance on generic third-party segments, exposure to lower-quality open-exchange environments, and traffic patterns that drove up cheap CPMs but did not deliver verified, viewable, human impressions in premium contexts. This led to a systematic cull of low-value exchanges, a sharp reduction in third-party data partners, and a pivot toward retailer first-party data, such as Kroger Precision Marketing audiences, where impressions could be tied directly to verified purchase behaviour.
While exact fraud percentages are not disclosed, Georgia-Pacific effectively reallocated a substantial share of its budget away from low-quality and potentially fraudulent inventory into tightly curated, high-viewability, high-human-traffic supply, moving its media productivity into the top quartile of its category and significantly shrinking the pool of suspect impressions. Verified human reach increased as they prioritised premium placements, transaction-backed retailer audiences, and AI optimisation via Yahoo’s Blueprint Performance to focus on impressions most likely to drive incremental sales rather than superficial reach. In practical terms, this forensic, in-house approach allowed Georgia-Pacific to recover a large portion of previously underperforming spend (functionally in the tens of percentage points of their programmatic budget) by cutting out wasteful, low quality paths and reinvesting in demonstrably effective, human-verified media.
| Metric | Before Forensic Audit | After Forensic Optimisation |
|---|---|---|
|
Reported impressions |
High |
Lower (non-human removed) |
|
Nominal CPM |
Lower |
Slightly higher |
|
Cost per real human reached
(hCPM) |
High (hidden) |
Significantly lower |
|
Performance stability |
Variable |
Consistent |
|
Budget recovered |
0% |
20-60% |
When fraud is removed, you are not paying less to reach fewer people. You are paying less to reach the same real people because you have stopped sending your ads to unverified or scam sites.
