Developing
Observational data with no randomisation. The in-market test has not yet been run.

In 60 seconds
- Kroger, Vidmob and MMA Global analysed 1,934 video and image ads (1,062 on Meta, 872 on DV360) run from January 2025 to January 2026.
- Predictive scoring forecast e-commerce conversion with 81% accuracy on unseen Q1 2026 data.
- About half of historical spend on the two platforms went to lower-performing creative; creative aligned with recommendations averaged about four times the conversion rate.
- Of 231 attribute levels tested, 22 survived statistical filtering.
Impact Layer
- What changed
- A retailer has linked creative attributes to actual e-commerce sales rather than a proxy such as view-through rate.
- Why it matters
- If creative drives a large share of results, scoring ads before launch could shift budgets, but the evidence so far is observational.
- Who is affected
- Retail and CPG brands, creative teams and agencies.
- Martech stack area
- Analytics, Advertising
- Stack impact
- Run a controlled test of any scoring tool before moving budget on its advice.
- Business impact
- Moderate A vendor-involved study as reported by Marketing Dive and PPC Land. The 81% has no published classification threshold or confidence interval, and Kroger does not disclose its spend on these platforms.
- Related technologies
- What to watch
- The planned in-market test.
Kroger, Vidmob and MMA Global found that Kroger could forecast e-commerce purchase conversion from creative before a campaign launched, according to Marketing Dive’s September 30, 2026 report.
What the study did
- Analysed 1,934 video and image ads (1,062 on Meta, 872 on DV360) that ran from January 2025 to January 2026, scoring messaging, narrative structure, branding and on-screen human interaction.
- Filtered 231 attribute levels down to 105 that passed statistical testing and 22 that survived (17 for DV360, 5 for Meta), per PPC Land.
Headline results
- 81% accuracy forecasting conversion on unseen Q1 2026 data.
- Creative aligned with the recommendations averaged a four-times improvement in conversion rate; PPC Land’s summary adds about 70% lower cost per conversion for high-scoring creative.
- About half of historical spend on the two platforms went to lower-performing creative; reallocation could produce more than twice the conversions from the same budget.
Caveats
The data are observational, with no randomisation; the 81% has no stated threshold or confidence interval; Kroger’s share of its $1.18 billion 2025 ad spend on these platforms is not disclosed; and Vidmob sells creative-scoring products. The in-market test has not yet happened.
Our read
Creative is a large lever, and linking it to sales rather than a proxy is progress. Before reallocating on a score, hold out a test group.
Sources
- Marketing Dive: What Kroger's predictive scoring work means for e-commerce marketers (opens in a new tab)
Source of the headline results and the Vidmob and Kellanova context.
- PPC Land: DOJ wants AdX open to Prebid in six months, Google asks for twelve (opens in a new tab)
The page's section on the study gives the sample split, filtering counts and caveats.
