FMCG / Anonymized case study
Separating invalid traffic from a 119.5 million impression headline
A high-volume display placement retained 68.46% of gross impressions after traffic-quality and count-once filtering.
Observed window: 4 Jun 2026 to 24 Oct 2026 (UTC)
Snapshot: 25 Oct 2026
31.31%
GIVT excluded, as a share of gross impressions.
The buying question
A large delivery count tells a media team how much activity was recorded. It does not tell them how much should survive a consistent measurement policy. For this FMCG portfolio, the useful question was how much of one high-volume display placement remained after invalid traffic and duplicate counting were removed.
What AdProof measured
Across the observed window, the placement recorded 119.53 million gross impressions. General invalid traffic (GIVT) filtering reduced that total to 82.11 million. The final count after the remaining traffic-quality and count-once filters was 81.83 million. GIVT accounted for 31.31% of the gross total; duplicates accounted for a further 0.23% of gross impressions.
Why the distinction matters
Most of the difference in this placement arose before deduplication. A buyer investigating the gap would therefore start with the traffic-quality breakdown, rather than treating duplicate callbacks as the entire explanation. The data establishes the measured difference. It does not establish the amount invoiced, a refund, or a decision to move budget.
How to use the finding
Put gross impressions, the count after GIVT filtering, and the final net count beside the partner report for the same placement and dates. Ask the partner to reconcile the definitions and exclusions at that level before combining the placement with other inventory. Keep a separate record of any resulting commercial adjustment.
Take this into your next campaign review
- Reconcile the placement before combining it into a portfolio average.
- Keep the GIVT exclusion and duplicate count as separate lines.
- Use the agreed billing definition to assess any financial impact.
More from this portfolio
When repeat counting is the larger part of the delivery gapPut the method into practice
Understanding invalid-traffic classifications