The Trade Desk published platform figures on October 5, 2026 claiming that campaigns using retail data targeting averaged 41% higher return on ad spend and 25% lower cost per acquisition globally than campaigns using none, with gaps of 125% in APAC and 150% in EMEA. The post gives no campaign count, no attribution model and no control for which advertisers chose to buy retail data.

In Short

The Trade Desk, a company that buys online ads for brands, says ads aimed using shoppers' purchase records earned back more money per dollar than ads that did not use that data. It matters because brands and retailers are deciding how much to pay for this kind of shopping data, and the figures come from the company that sells access to it. The numbers are the company's own and are not an independent test, so they show a pattern on its platform rather than proof of cause.

What The Trade Desk published

The post, titled "How retail data is helping to deliver outsized returns across markets" and filed under the company's Data and Measurement section, is dated October 5, 2026. Its headline figures rest on what The Trade Desk calls "new platform insights," drawn from its own platform data covering June 2025 to June 2026. A footnote attached to every statistic reads: "The Trade Desk platform data, Global, June 2025 - June 2026. Note: Excludes WMT DSP."

According to The Trade Desk, "Globally, campaigns using retail data targeting saw 41% higher average return on ad spend (ROAS) and 25% lower average cost per acquisition (CPA) compared with campaigns using no retail data." The post then adds two regional figures: "In APAC, advertisers saw 125% higher average ROAS. In EMEA, the average campaign's ROAS increased by 150%."

That is the whole of the performance claim. There is no North America figure in the text of the post, even though the United States is where The Trade Desk's retail data partnerships are concentrated. An interactive map carousel titled "Advertisers around the world are achieving strong retail data outcomes" sits below the paragraph; the copy of the page reviewed by PPC Land displays only its first slide, a world map, so any additional regional figures it may hold could not be read.

Two different kinds of comparison

The regional sentences are not phrased the same way. The global and APAC figures describe a comparison between two groups: campaigns with retail data against campaigns without it. The EMEA sentence says the average campaign's ROAS "increased by 150%," which reads as a before-and-after change within the same campaigns. The post does not say whether the EMEA number uses the same comparison as the other two, or a different one. Nor does it define how ROAS was calculated, which attribution window applied, or whether sales were measured by the retailer, by The Trade Desk, or by a third party.

What the averages leave out

Averages across groups that chose their own treatment are hard to read as cause and effect. Advertisers that buy retail data are, by construction, more likely to be selling products in categories where purchase records exist, and more likely to run campaigns with sales objectives that produce a measurable return in the first place. A campaign aimed at brand awareness with no retail data would sit in the comparison group with a low or unmeasured ROAS. The post does not say whether campaigns were matched by objective, vertical, budget or channel, and it gives no count of campaigns, advertisers or impressions behind the averages. The return on ad spend figure also depends on what counts as a sale, and retail data products often come bundled with the retailer's own closed-loop reporting.

The Trade Desk includes its own caveat at the bottom of the page: "This information is provided solely for background and is not a representation or guarantee of any future performance."

The exclusion of Walmart DSP matters too. Walmart's buying platform is a white label product, built on The Trade Desk's technology under a partnership the two companies announced in 2021. Removing it from the dataset means the figures describe retail data bought through The Trade Desk's own interface, not through the largest retailer-branded deployment of the same technology. The post does not explain the exclusion.

Adoption and funnel figures

Beyond performance, the post reports adoption numbers for the same period. According to The Trade Desk, "Over a third of advertisers globally activated retail targeting data on The Trade Desk's platform, up nearly 20% year over year." The phrase "up nearly 20%" is not defined: it could mean the share of advertisers rose by about 20% relative to the prior year, or by close to 20 percentage points. The first reading is more consistent with a share that now sits just above one third, but the post does not say which.

The company names consumer packaged goods as "one of the most penetrated categories," and lists e-commerce, retail and financial services "among the fastest-growing categories adopting retail data." No category shares are given.

A chart in the post breaks down retail data spend by funnel stage, again citing global platform data from June 2025 to June 2026:

  • Upper funnel: 29%
  • Mid funnel: 26%
  • Lower funnel: 45%

On these numbers, more than half of retail data spend, 55% by PPC Land's own addition of the upper and mid funnel shares, went to campaigns that were not classified as lower funnel. The Trade Desk uses that split to argue that retail data "is often associated with lower-funnel activation, but platform insights show advertisers are applying it across the full customer journey." The post does not say how a campaign was assigned to a funnel stage, whether by the advertiser's stated objective or by the company's own classification.

That split also complicates the headline comparison. If 29% of retail data spend sat in upper-funnel campaigns, a ROAS average across all retail data campaigns would blend awareness work with conversion work. The post does not break the 41% figure down by funnel stage.

The 11-brand incrementality study

The strongest-sounding claim in the post comes from a separate analysis. According to The Trade Desk, it "ran an analysis spanning 11 brands and 8 retailer datasets across grocery, big-box, and specialty categories, using Attain as a retail-agnostic measurement layer." The question, in the company's words, was "Does retail data on The Trade Desk generate incremental sales?"

The company reports that "Every campaign in the study drove statistically significant incremental sales and transaction lift." It adds that "Ninety percent of campaigns beat Attain's industry benchmarks for new-buyer ROAS, outperforming them by an average of 122%." A third finding states that "Retail data also lifted average transaction value from attributed consumers by 23% relative to those brands' trailing 12-month norms."

Incrementality is the right question for retail data, and testing it through an outside measurement company is a step beyond attribution alone. The disclosure, however, stops short of what a reader would need to assess it. The post does not name the 11 brands or the eight retailers, give the study period, describe the control group or holdout design, state confidence levels, or say how many campaigns were included. "Statistically significant" is asserted without a threshold. The 122% figure is measured against Attain's own industry benchmarks, which are not published in the post, and only the 90% of campaigns that beat the benchmark are averaged in that number.

The 23% transaction value figure uses a different baseline again: the brands' own trailing 12-month averages, rather than an unexposed control group. Seasonal effects, price changes and product launches during the campaign period would all move that number regardless of advertising. Attain is not described further in the post, and PPC Land could not find previous coverage of the company in its own archive.

Recommendations in a vendor post

Roughly half the post is guidance on how to use retail data, written in the second person and addressed to advertisers. It lists audience strategies, including using retail data "as a seed to identify people who resemble your highest-value customers," reaching category buyers, reconnecting with lapsed customers, engaging repeat buyers and reaching "life-stage audiences" such as "new pet owners, frequent travelers, or recent movers."

It also lists the channels across which The Trade Desk applies retail signals for offsite buying: CTV, display, online video, audio and digital out-of-home. "Instead of limiting retail media to sponsored placements on a retailer's site or app," the company writes, advertisers can "apply retail signals to different channels." A sidebar promotes an interview on The Current, The Trade Desk's own editorial site, with Paul Frampton-Calero, chief executive of Goodway Group, under the headline "Legacy structures and silos get in the way." The post does not quote him beyond that headline.

The closing section restates the commercial pitch. "Retail data connects what people buy with how advertisers plan, activate, and measure media," according to the company, which says the advertisers that "get the most from it will be the ones that use it as a signal for building audiences, shaping full-funnel strategy, improving off-site activation, and measuring real business outcomes."

Why the timing matters for The Trade Desk

The post arrived during a difficult quarter for the company. On August 6, The Trade Desk reported second-quarter 2026 revenue of $715 million, up 3% year over year, and guided to at least $650 million for the third quarter, roughly 12% below the prior year. Shares fell 24% after the results, and on the same earnings call management said participating retailers now represent more than 80% of US retail sales, while confirming that Walmart had renewed its partnership with The Trade Desk without exclusivity. A month later, the company cut about 575 jobs, about 15% of its workforce.

Retail data is one of the few areas where The Trade Desk can point to signals that its rivals do not hold in the same combination. Its pitch has been that a single independent buying platform can combine data from many retailers, rather than requiring a brand to buy through each retailer's own network. That pitch has come under pressure through 2026. Walmart Connect, whose own DSP runs on The Trade Desk's technology, took its shopper audiences into Yahoo DSP in May, routed through Magnite and limited at first to VIZIO CTV inventory. In June, Walmart Connect announced a proof of concept bringing purchase data into Google's DV360 for YouTube ads, with measurement of resulting Walmart sales.

The earnings coverage also carried a named retail data example from The Trade Desk: General Mills' Nature Valley campaign in the UK, using data from Tesco, Sainsbury's and Ocado, which the company said produced a 5X sales uplift, a 92% lower CPM versus benchmark and a 2X ROAS improvement over four months. Like the October figures, those were company-supplied numbers.

A pattern in the company's performance claims

This is not the first time in recent months that The Trade Desk has headlined a performance percentage with limited methodology. On August 27, the company released Zuma, an update to its Kokai platform, with a claim that Koa AI optimizations cut CPAs by 32% across 62 campaigns. That figure compared the new model against the previous model version rather than against campaigns without Koa, and no methodology document was published. The October retail data post goes further in one respect, naming an outside measurement partner for its incrementality study, but discloses less in another: it gives no sample size at all for the 41% figure.

The company's third-party data marketplace sits close to this story. The Trade Desk added Adsquare location data to Audience Unlimited in June, the AI-scored data marketplace launched in September 2025 that prices data at 3.3% or 4.4% of impression costs in Control Mode. The retail data post does not say whether retail segments bought through Audience Unlimited are included in the platform data, or how retail data is priced.

The post also landed on the same day Apple asked The Trade Desk to test an iOS 27.2 beta build, weeks after iOS 27 began blocking The Trade Desk from serving ads on Safari. Purchase-based data, which matches on retailer identity rather than on browser cookies, is one of the signals that survives browser restrictions of that kind.

Why this matters for marketers

For brands deciding how to split budget between retailer-operated networks and independent platforms, the October post adds to a running argument about where retail data performs best and who gets to measure it. Measurement remains the weak point across the sector. An Incremental white paper covered by PPC Land in June estimated that siloed, single-retailer attribution misses 36% to 53% of total retail media impact, rising to 67% to 80% for off-site video. In April, an IAB white paper argued that legacy measurement undervalues retail media, describing marketing mix modeling as structurally ill-suited to the channel.

Both findings cut two ways for The Trade Desk's numbers. If single-retailer attribution undercounts the effect of off-site retail media, then a retail-agnostic measurement layer such as the one used in the 11-brand study is a reasonable design. But the headline 41% ROAS figure is platform data, and the post does not say whether it relies on the same retailer-reported sales that those studies found incomplete.

Retail media networks have also been building their own off-site products. Criteo and TripleLift packaged commerce audiences into curated deals in April, and Amazon DSP began naming the specific retailers its ads drive sales to. Each of these companies sells the data, the buying and in many cases the measurement. The Trade Desk's post is a contribution to that contest rather than an independent reading of it, and the absence of a North America figure, a sample size and a matched comparison leaves the 41% number as a description of what happened on its platform, not a measure of what retail data causes.

Timeline

Summary

Who: The Trade Desk, the independent demand-side platform, with Attain as the measurement partner for an 11-brand incrementality study. Walmart DSP, which runs on The Trade Desk's technology, is excluded from the data.

What: A post claiming that campaigns using retail data targeting on The Trade Desk averaged 41% higher ROAS and 25% lower CPA globally than campaigns without retail data, with 125% higher ROAS in APAC and a 150% ROAS increase in EMEA. Over a third of advertisers activated retail data, up nearly 20% year over year, and retail data spend split 29% upper funnel, 26% mid funnel and 45% lower funnel. The post gives no sample size, attribution model or matched comparison.

When: Published October 5, 2026, covering platform data from June 2025 to June 2026.

Where: Global platform data, with regional figures for APAC and EMEA and no North America figure in the post's text.

Why: The figures support The Trade Desk's case that an independent platform is the right place to buy retail data across retailers, at a time when its revenue growth has stalled, Walmart has taken its data to Yahoo DSP and Google, and measurement of retail media remains contested.