Every retail media vendor claims their technology drives revenue. Most of those claims never get tested by anyone outside the vendor's own sales deck. In April 2026, one did. The Drum Awards for Commerce Media judged a submitted case study on unified ranking against every other entry in the category, not against a marketing brief written to flatter it. Breaking the Legacy Stack, built on Pentaleap's optimization layer, won Ad-Tech Innovation and was highly commended for Retail Media Infrastructure/Platform of the Year.

The case study behind the award is public. So is the retailer's data, reported directly by The Drum as part of the award writeup. That makes it a useful thing to look at closely, because it answers a question a lot of retail media teams are quietly asking right now: does unifying ranking logic actually move revenue, or is "unified ranking" just this year's version of "AI-powered"?

TL;DR

A major US department store retail media network added Pentaleap's optimization layer on top of its existing ad server, without adding a single new ad slot. Per The Drum's published case study, the initial controlled A/B test showed a 78% increase in ad revenue, an 85% increase in CTR, and a 152% increase in conversion value. At scale, sustained gains settled at a 23% ad revenue increase and a 21% CTR increase across Search, Browse, and product detail pages. The mechanism was not more ads. It was ranking sponsored and organic products against the same relevance bar, and letting more demand compete for the same fixed set of positions.

The problem: growth pressure, but no room to add more ads

The retailer at the center of this case study was not starting from zero. It was already running a mature onsite retail media network with real revenue behind it, per the case study. That is a harder starting point than it sounds. A network that has already captured the easy wins usually has one lever left on the table: add more ad slots. And that lever comes with a cost most growth targets do not account for. More sponsored placements crowded into a shopping page tends to degrade the experience it's supposed to monetize, which is exactly the tension the case study names directly as the retailer's core constraint: how do you grow onsite revenue without breaking the promise of search.

Underneath that constraint was a structural one. Organic ranking on the site was optimized for shopper intent. Sponsored placements were optimized for bid. Two systems, two goals, running independently on the same page. When that happens, the products landing in the most visible positions do not necessarily reflect the best match for the shopper standing in front of them. They reflect whichever system won that particular row. Why Your Fixed Slots Are Quietly Costing You covers this same structural gap in more detail, for teams trying to work out whether it applies to their own stack.

What actually changed

The retailer did not replace its ad server. It kept the incumbent system as the workflow layer, the place where brands and internal teams already managed campaigns, budgets, and bids, and added Pentaleap as a separate optimization layer governing ranking and serving decisions underneath it. Per the case study, brand-facing workflows in the incumbent UI did not change at all. This is the same low-disruption pattern behind Your Stack Decides the Integration, Not the Vendor's Roadmap: the stack a retailer already has decides how the integration works, not the other way around.

What changed was arbitration. Instead of sponsored products competing only against each other for a fixed set of reserved positions, every product on the page, sponsored and organic, was evaluated against the same relevance standard. A highly relevant sponsored product could now legitimately outrank a mediocre organic result. A weak sponsored product could no longer coast into a top row just because it had a bid attached to it.

The non-negotiables stayed fixed throughout the rollout, which matters more than it sounds like it should: no new ad slots were added, mobile stayed capped at four sponsored placements and desktop at eight, and every promoted product had to clear the same relevance bar as the organic result it would displace.

Why revenue went up without more ads

This is the part that tends to get skipped in vendor pitches, because it is less flattering than "we added AI and revenue went up." Two separate mechanisms did the work here, and they are worth separating because they answer different objections.

Better ranking put the right products in front of the right shoppers. When sponsored and organic compete under one relevance standard, the products that actually match shopper intent get the visible positions, whether they are paid or not. Per the case study, this alone showed up in the data as fewer mismatches in high-impact rows, meaning shoppers were seeing sponsored products that better matched what they were already looking for, not products competing purely on advertiser budget.

Deeper auctions put more demand behind the same positions. With ranking validated, the retailer expanded demand access without touching ad load. Per the case study, the incumbent demand source alone went from returning 8 bids per request to 24, and once additional demand sources were connected, up to 40 bids competed for the same 8 positions. The number of ad slots never moved. What moved was how much competition existed behind each one, and competition behind a fixed set of positions is what drives price up.

Those two mechanisms, relevance-driven ranking and deeper auction competition, are also why the lift compounded rather than came from a single lever. Revenue increased because better products won the positions and because more bidders were competing to win them. Neither one alone would have produced the same result. The first mechanism is the same argument made in What Actually Boosts Sponsored Product Revenue (It's Not More Ad Slots). The second is covered in more depth in Why Real-Time Bidding Might Finally Be Ready for Primetime in Retail Media.

The numbers, and what happens to them at scale

Initial controlled testing is where vendors like to stop the story, because early numbers in a small test are usually the best numbers a program will ever post. This case study kept reporting after the initial test, which is part of why it held up under award judging.

In the initial A/B test, per The Drum's published case study: ad revenue rose 78%, revenue per request rose 66%, CTR rose 85%, and conversion value rose 152%.

After expanding unified ranking across channels and device types, the sustained figures settled lower, which is expected and worth stating plainly rather than burying: ad revenue held at a 23% increase, CTR at 21%, and conversion value at 47%, sustained across Search, Browse, product detail pages, app, mobile web, tablet, and desktop.

The gap between the initial test numbers and the sustained numbers is not a red flag. It is what an honest rollout looks like. Controlled tests on a small traffic slice tend to outperform full-scale deployment, because early traffic samples and easy wins get captured first. A vendor who only ever shows you the pilot number is not showing you the whole picture. This case study shows both, because the retailer measured both. For more on why the number retailers see after a pilot is rarely the number they should plan around, see What Actually Keeps a Retailer on Retail Media Ad Tech After the Pilot.

Why this needed to be judged, not just published

Vendor case studies are usually self-graded. A retail media team reading one has no real way to compare it against what a genuinely different, less impressive deployment would have looked like. Award judging exists specifically to close that gap. A panel with no stake in Pentaleap's growth evaluated this submission against every other entry in Ad-Tech Innovation and Retail Media Infrastructure/Platform of the Year, and it won one and placed in the other.

That does not make the underlying architecture right for every retailer. It does mean the mechanism described here, unifying ranking logic and increasing auction density without adding ad load, held up to outside scrutiny rather than only internal marketing review. For the broader framework behind why this architecture works the way it does, see The Three Architectures: How Retail Media Tech Actually Works. For how these results compare against industry-wide trends, see the H1 2026 Sponsored Products Benchmarks Report.

Key Takeaways

• A major US department store retail media network added an optimization layer on top of its existing ad server without adding new ad slots, per a case study submitted to The Drum Awards for Commerce Media 2026.
• The initial controlled A/B test showed a 78% increase in ad revenue, an 85% increase in CTR, and a 152% increase in conversion value, per the published case study.
• At scale, sustained gains were a 23% ad revenue increase and a 21% CTR increase, holding across Search, Browse, and product detail pages.
• The revenue lift came from two mechanisms: ranking sponsored and organic products against one relevance standard, and increasing the number of bids competing for the same fixed set of ad positions.
• Brand-facing workflows in the incumbent ad server's UI did not change during the rollout.
• The deployment won Ad-Tech Innovation and was highly commended for Retail Media Infrastructure/Platform of the Year at The Drum Awards for Commerce Media 2026.

Frequently Asked Questions

Did this retailer replace their existing ad server?

No. The incumbent ad server remained the control plane for campaign, budget, and bid management. The optimization layer was added on top of it to govern ranking and serving decisions, per the case study.

‍

Did revenue growth come from adding more ads?

No. Ad load stayed fixed throughout, at a maximum of four sponsored placements on mobile and eight on desktop. Growth came from improving which products won those positions and from increasing the number of bids competing for them.

‍

How much of the revenue lift held up after the initial test?

The initial controlled A/B test showed a 78% ad revenue increase. After expanding unified ranking to full scale across channels and devices, the sustained increase settled at 23%, still without adding ad load.

‍

What was the source of the additional bid competition?

Per the case study, the incumbent demand source expanded from returning 8 bids per request to 24, and connecting additional demand sources brought competition up to 40 bids for the same 8 positions.

‍

Is this result specific to department stores?

The case study is specific to one US department store retail media network. The mechanism, unifying sponsored and organic ranking under one relevance standard while increasing auction density, is not category-specific, but results vary by retailer based on existing relevance overlap and how much of the page is open to unified ranking versus reserved for direct deals.

‍