The only way to know what your ads really did
- Topic
- Marketing
- Source
- Martech.org
- Published
This summary was written by ShopDigest AI from Martech.org’s report, by Benjamin Wenner. Read the full story ↗
Platforms claim too much credit and backends often give too little. Holdout tests can show how much revenue advertising truly generated. The post The only way to know what your ads really did appeared first on MarTech.
Overview
The article explains why marketing platforms and backend analytics often report conflicting ROAS figures, with platforms over-crediting and backends under-crediting due to last-click attribution. It argues that holdout tests are the only reliable way to measure true advertising impact, especially for impression-based channels like social and display.
Key Details
- Marketing platforms inflate ROAS by counting view-through and modeled conversions, while backends undercount by crediting only the last click.
- Impression-based channels (social, display, video, CTV) are most undervalued by last-click attribution, often appearing invisible in backend reports.
- Running multiple platforms (e.g., Google and Meta) can lead to double-counting the same sale, making combined reported revenue exceed actual revenue.
- Holdout tests, which compare a test group against a control group, provide a more accurate measure of incremental revenue from ads.
Why It Matters
Relying on either platform-reported or backend numbers alone can lead to misallocating ad spend, potentially cutting effective upper-funnel campaigns. Marketers need to use holdout tests to get a true picture of performance and avoid wasting budget on channels that appear to underperform but actually drive conversions.
Frequently asked questions
Why do marketing platforms report higher ROAS than backend analytics?
Marketing platforms count conversions generously by design, including view-through conversions, modeled conversions where consent was never granted, and clicks from up to three weeks earlier. This inflates the reported ROAS compared to backend data.
Why does backend revenue reporting under-credit ads?
Most backend reporting uses last-click attribution, crediting the final touchpoint (often a brand search or direct visit) with the sale. This ignores the earlier paid clicks or impressions that initiated the customer journey, so upper-funnel channels appear less effective than they are.
What is the best way to measure true ad performance?
The article recommends holdout tests, which compare a group exposed to ads against a control group that isn't, to measure the actual incremental revenue generated. This avoids the biases of both platform over-reporting and backend under-reporting.
How does running both Google and Meta ads cause double-counting?
When a customer sees a Meta ad, later searches your brand, clicks a Google ad, and buys, both platforms may claim the conversion—Google on the click and Meta on a view-through basis. This makes the combined reported revenue exceed actual sales.
