How to Measure Ad Incrementality Beyond ROAS

TL;DR

Measure advertising by combining return on ad spend with new-to-brand customer rates, because a high platform-reported ROAS may include purchases that would have happened without the ads. Audit spending across every funnel stage, use imperfect but timely signals, and validate true lift through regional holdout tests that compare an exposed market with a similar unexposed market.

Transcript

On the show today, we have an incredible guest. I want to welcome Benino Veter. He is the chief media officer and digital commerce officer at Liquid Death and he is the best person I have ever met when it comes to talking about advertising and how you use advertising to generate real revenue. He's built a fullfunnel media system. He has unique crea... Read More

Key Insights

• A full-funnel media audit is the starting point for understanding advertising performance. Liquid Death categorized where money was spent, how it was spent, and which funnel stage it supported, then used that map to identify stages receiving too much or too little investment.
• Perfect measurement is not required for timely media decisions. Benoit Vatere prioritizes the strength of available signals because waiting for an ideal metric, confidence level, or lengthy model can delay action until the opportunity has already passed.
• Platform-reported ROAS is not proof of incremental revenue. Meta or another platform may attribute three dollars of revenue to one dollar of spending even when some or all of those purchases would have happened without the advertisement.
• New-to-brand customer share is a practical signal of incrementality. When both ROAS and the percentage of customers new to the brand are high, the advertising is more likely to be creating additional demand instead of claiming existing demand.
• Incremental ROAS measures sales lift attributable to advertising. The approach goes beyond reported conversions by estimating whether sales increased because media ran, which requires comparing observed outcomes with a credible representation of what would have happened without that spending.
• Geo holdout testing works by comparing similar regions with different media exposure. If spending occurs in New York but not Los Angeles, and sales rise only in New York, that difference provides evidence that the advertising produced incremental results.
• Social commerce functions primarily as a discovery tool in Vatere's current assessment. Its broader value should therefore be evaluated through its effect on scalable retailer sales, including purchases made through Amazon, Walmart, and Target rather than only transactions credited inside a social platform.
• Liquid Death does not sell directly to customers through its website. Its digital commerce measurement consequently focuses on retailer websites, making cross-channel influence and retail sales lift especially important when judging whether awareness or discovery activity generates real revenue.
• More videos from this channel:
  • How to Build AI Agents with GPT and N8N Workflows
  • How to Build an AI RevOps Team with Perplexity
  • How to turn daily tasks into reusable AI skills
• More from this channel: Marketing Against the Grain

Questions & Answers

Q: Why can a high ROAS give marketers a false signal?

A high return on ad spend can be misleading because an advertising platform may claim credit whenever someone views or clicks an ad and later buys. Some of those customers would have purchased without the advertising. In that case, the platform reports revenue but the campaign did not create all of it. Marketers therefore need to evaluate incrementality, not only attributed return.

Q: How can marketers quickly estimate whether advertising is incremental?

Marketers can pair ROAS with the percentage of customers who are new to the brand. High ROAS shows that the platform associates spending with revenue, while a high new-to-brand share suggests that the campaign is reaching people beyond existing demand. When both signals are strong, the spending is probably more incremental, although a controlled test can provide stronger validation.

Q: How should a company begin a full-funnel media audit?

A company should first document where its media money is spent, how each investment is deployed, and which funnel stage it supports. Categorizing the entire mix makes it possible to see whether one stage receives too much funding while another receives too little. The company can then add performance and incrementality signals to triage spending and rebalance the system.

Q: What is a geo holdout test for advertising?

A geo holdout test compares two regions with similar shoppers while changing media exposure in only one of them. For example, a brand can spend in New York and withhold comparable spending in Los Angeles. If sales rise in New York but do not move in Los Angeles, the difference provides evidence that the advertising caused incremental sales rather than merely receiving attribution.

Q: What does incremental ROAS measure?

Incremental ROAS focuses on the revenue lift caused by advertising, rather than all revenue an advertising platform associates with a campaign. It asks whether sales would have occurred without the spending. Measuring that difference may require models, platforms, or controlled comparisons, but it produces a more meaningful view of effectiveness than relying exclusively on reported clicks, views, conversions, and attributed revenue.

Q: Why does Liquid Death use signals instead of waiting for perfect measurement?

Liquid Death uses signals because perfect measurement can take too long to obtain. Waiting for an ideal metric, confidence level, or a lengthy model may postpone a decision until the useful moment has passed. Platform data is imperfect, but it still supplies an initial signal. Combining multiple signals, such as ROAS and new-to-brand rate, supports faster and more informed allocation decisions.

Q: How should social commerce be evaluated for a retail brand?

Social commerce should be evaluated not only by sales reported inside the social platform, but also by its effect on larger retailer channels. Vatere describes social commerce primarily as a discovery tool and examines whether that discovery influences scalable sales through Amazon, Walmart, and Target. This approach tests broader commercial impact instead of treating platform-attributed transactions as the complete result.

Q: Why is retailer sales measurement central to Liquid Death's media strategy?

Retailer sales measurement is central because Liquid Death does not sell directly to customers on its own website. Its products are sold through retailer websites, including Amazon, Walmart, and Target. As a result, the media team must determine whether creative, social discovery, and other advertising investments produce incremental purchases across those retailers, rather than optimizing around a direct-to-consumer checkout funnel.

Summary & Key Takeaways

• Liquid Death built a full-funnel media system that connects its distinctive creative work to sales through retailer websites. Benoit Vatere began by categorizing existing media spending by funnel stage, identifying imbalances, and reviewing available platform measurements. His objective was to understand where money went, how it performed, and whether results represented incremental demand.

• ROAS alone can misrepresent effectiveness because advertising platforms may claim revenue from customers who would have purchased without seeing an advertisement. Liquid Death therefore paired ROAS with the percentage of customers who were new to the brand. High ROAS combined with a high new-to-brand rate provided a faster signal that spending was probably incremental.

• Geo holdout testing offers a clearer measurement of incremental ROAS. A brand can select two regions with similar shoppers, advertise in one region, withhold spending in the other, and compare changes in sales. Liquid Death also examines whether discovery channels such as social commerce influence scalable retailer sales on Amazon, Walmart, and Target.