The challenge of proving incrementality in retail media: Why brands are demanding a new standard of measurement
Retail media has evolved from a niche marketing experiment into a foundational pillar of modern advertising strategy, commanding budgets that now rival long-standing search and social media channels. As spending in this sector surges—with global retail media ad spend projected to exceed $150 billion by 2026—brand leaders and corporate finance departments are shifting from an initial focus on rapid adoption to a rigorous interrogation of return on investment. The core question now facing marketing executives is not merely whether a campaign generated sales, but how many of those transactions were truly incremental, rather than purchases that would have occurred regardless of ad exposure.
Juan Baron, head of international and global accounts at Decentriq, has highlighted this tension as the defining hurdle for the next stage of the industry’s maturation. The industry is currently trapped in a cycle of "attributed success" that often obscures the true causal impact of advertising dollars.
The inherent bias in current attribution models
The current retail media landscape relies heavily on closed-loop reporting, a system where retailers possess both the advertising exposure data and the final transaction data. Because retail media often targets consumers at the point of purchase—where intent is already significantly elevated—the resulting data frequently presents an inflated view of ad performance.
For instance, when a shopper searches for a specific brand on a retailer’s website and clicks a sponsored listing before purchasing, the current attribution model credits the ad with that sale. However, empirical evidence suggests that a significant percentage of these shoppers were already loyal to the brand or had already decided on the purchase prior to seeing the sponsored result. By counting these "inevitable" sales, retailers provide reports that appear to show exceptional return on ad spend (ROAS), while masking the true incremental lift—the actual change in consumer behavior directly attributable to the ad.
A timeline of the retail media measurement evolution
The trajectory of retail media measurement can be categorized into three distinct phases.
The first phase, spanning approximately 2018 to 2020, was the "Gold Rush" era. During this period, retailers began opening their digital real estate to advertisers, and the primary objective was reach. Measurement was basic, focusing on clicks, impressions, and simple conversion tracking.
The second phase, from 2021 to 2023, saw the professionalization of the channel. As budgets grew, brands demanded more sophisticated reporting, leading to the adoption of "walled garden" attribution models. These models provided the transparency brands requested but remained largely self-serving for the retailers, who controlled the methodologies.
We are currently in the third phase: the "Era of Accountability." Since late 2023, there has been a global push from major CPG (Consumer Packaged Goods) firms to standardize measurement. Brands are no longer satisfied with reports prepared by the party selling the inventory. They are increasingly demanding external validation and rigorous incrementality testing to justify their budget allocations.
The methodologies of incrementality testing
To move beyond simple attribution, marketers are turning to incrementality testing, which relies on the "counterfactual"—the statistical estimation of what would have happened in the absence of the ad.
The "gold standard" remains the randomized holdout experiment. In this setup, a segment of the target audience is intentionally excluded from seeing an advertisement, allowing analysts to compare the purchasing behavior of the "exposed" group against the "control" group. If the conversion rate in the exposed group is statistically higher than in the control group, the difference represents the incremental lift.
However, randomization is not always feasible. In scenarios where it is technically or commercially difficult to withhold ads, marketers utilize alternative methods:
- Geo-experiments: Spending is increased or decreased in specific geographic regions to observe the relative impact on sales compared to baseline regions.
- Matched-market testing: Similar geographic or demographic regions are paired based on historical sales performance, with one region receiving the ad campaign and the other serving as a control.
- Synthetic controls: Advanced statistical modeling creates a "virtual" control group by synthesizing data from various sources to mirror the behavior of the target audience, approximating a randomized trial.
These methods are not without challenges. Timing is a critical factor; for example, a short-term promotion might pull future sales forward, creating a temporary spike that looks like success in week one but leads to a performance slump in week six. Furthermore, defining "incrementality" itself is a point of contention. A brand may define an incremental sale as a net-new customer acquisition, while a retailer might view that same sale as merely a "switch" from one brand to another within their store. Such discrepancies highlight the need for alignment before any campaign goes live.
The data impasse and the role of privacy-preserving tech
A significant barrier to accurate measurement is the siloed nature of data. Retailers hold the exposure and transaction records, while brands hold the customer lifetime value, pricing history, and cross-channel sales data. Neither party is typically willing to share this raw data due to stringent privacy regulations such as GDPR and CCPA, as well as the need to protect sensitive commercial information.
This leads to the current "self-marking" problem, where retailers provide aggregated reports that brands cannot fully audit. To resolve this, the industry is increasingly turning toward privacy-preserving collaboration tools, most notably data clean rooms.
Data clean rooms allow retailers and brands to match their respective datasets in a secure, encrypted environment. The underlying data remains inaccessible to both parties; instead, the system runs pre-approved queries to compute the results. This allows the brand to measure the true lift of a campaign across multiple channels without exposing individual shopper identities or the retailer’s proprietary business intelligence.
Implications for the future of retail media
The shift toward rigorous incrementality testing will have profound implications for the retail media ecosystem.
- Budget Reallocation: As brands gain better visibility into which retail media investments actually drive incremental growth, they are expected to shift budgets away from underperforming placements. This will likely lead to a consolidation of spend among retailers who can prove their effectiveness.
- Standardization of Metrics: We can anticipate an industry-wide move toward standardizing how incrementality is defined and measured. This will be driven by industry bodies and large-scale, multi-retailer collaborations.
- The End of "Black Box" Reporting: Retailers that continue to rely solely on proprietary, opaque attribution models will face increasing pressure from brands to open their measurement processes to independent validation.
- Governance as a Priority: The success of data clean room adoption depends less on the technology itself and more on the governance framework surrounding it. Retailers and brands must agree on the fundamental questions, the logic of the queries, and the protocols for handling suppressed data.
Conclusion: Proof as the engine of growth
The growth of retail media is not in jeopardy, but its next phase of expansion will be tethered to proof. The retailers that thrive will be those that view transparency not as a liability, but as a competitive advantage. By enabling brands to rigorously test the incrementality of their media spend, retailers can build the trust necessary to secure larger, more stable, and long-term marketing commitments.
As Juan Baron suggests, the technical infrastructure to bridge the data gap already exists. The remaining obstacle is a cultural one: the industry’s collective willingness to adopt a shared, objective definition of success—even if that definition results in lower, but more accurate, figures than current attribution models provide. The companies that embrace this reality today will likely define the market standards of tomorrow.
