Magnite today published a case study claiming that a connected TV campaign for an unnamed retailer reached its target audience 43% more efficiently with Comscore's ID-free Predictive Audiences than with the ID-based audiences used in the same campaign. The campaign ran in the United States during the third quarter of 2026 on inventory sold by Trusted Media Brands (TMB) through SpringServe, Magnite's ad server for streaming publishers. When Comscore announced the integration on October 28, its only performance evidence was a reference by Future Today to "preliminary findings", without numbers. The case study does not define the efficiency metric, name who measured audience delivery, or disclose prices, impressions or spend.

In Short

Magnite says a retailer's streaming TV ads on Trusted Media Brands reached homeowners and DIY enthusiasts 43% more efficiently with Comscore audiences built from viewing content than with audiences built on tracking identifiers, and gained 10% more target-audience delivery. The claim matters to TV buyers and publishers deciding whether content-based predictions can replace device and household IDs, and to Comscore, which has reported falling usage of the product line. The numbers come from the companies selling the integration, with no method published, so they show what Magnite and Comscore claim rather than an audited result.

What Magnite published

The document sits under Insights and Case Study on Magnite's site, carries the byline "Magnite Team" and is dated November 9, 2026, with a stated reading time of three minutes. Its opening sentence sets out the claim: "Through Magnite's real-time data integration with Proximic by Comscore, a leading retailer used ID-free Predictive Audiences to efficiently reach its target audience at scale via SpringServe."

The campaign details are short. A summary box lists the location as the United States, the campaign date as Q3 2026, the deal type as a Publisher Direct Deal and the ad format as CTV. The target audience is given as "HomeOwners, DIY, Appliance Shoppers", although the narrative text names only "a core audience of homeowners and DIY enthusiasts across TMB's premium CTV inventory."

According to the case study, TMB "was exploring how contextual signals could strengthen campaign performance for a leading retailer." To reach the audience "without relying on identifiers," TMB activated Comscore Predictive Audiences through SpringServe, "layering audience and contextual segments across its CTV footprint." It configured its targeting stack inside SpringServe to include the segments, "extending reach through content-level signals."

Trusted Media Brands publishes titles including Reader's Digest, Taste of Home and Family Handyman, and bought the viral video company Jukin Media in August 2021. The case study does not say which TMB channels, apps or programs carried the campaign. The retailer is not named either.

The two numbers

The results section makes two claims, each presented as a headline figure with a short explanation.

43% "targeting efficiency"

"Compared with ID-based audiences in the campaign, Predictive Audiences delivered the target audience 43% more efficiently, allowing more of the advertiser's budget to go toward working media while maintaining performance," according to the case study.

The metric behind "efficiently" is not defined. It could describe cost per on-target impression, the share of impressions that reached the target audience, or some combination. The difference is material. If the figure means 43% more on-target impressions per dollar, the cost of each on-target impression would have been about 30% lower than with the ID-based segments, by PPC Land's calculation. If it means the cost fell by 43%, the gap is larger. The case study does not say which.

Nor does it say who decided which impressions counted as reaching homeowners and DIY enthusiasts. If Comscore measured the result, the same company built the segments and graded them. The document also does not say whether the ID-based and predictive segments ran on the same inventory over the same weeks, whether impressions were split at random between them, or which company supplied the ID-based audiences. "Maintaining performance" is not tied to any named measure, such as completion rate, brand lift or sales.

10% more core audience delivery

The second figure concerns scale. "Predictive Audiences extended delivery beyond the core ID-based audience, contributing an additional 10% of target audience scale and helping the campaign achieve full delivery," according to the case study.

The base is unclear. The 10% could be a share of the total target-audience delivery or an addition on top of what the ID-based segments delivered. The wording implies that the ID-based segments alone would have left the campaign short of its delivery goal, though the case study does not say by how much.

Where the saving sits

The strategy section offers the reason for the price advantage: "because there's no ID dependency, it runs at a meaningfully lower CPM, freeing up more budget for working media without sacrificing reach or performance." "Meaningfully" carries no number. Working media usually refers to the share of a budget that buys ad space rather than data, technology and fees, which suggests the saving lies in the cost of the audience data rather than in the price of TMB's inventory. The case study does not split the CPM into a media price and a data fee, so that reading cannot be confirmed from the document.

How the segments are built

According to the case study, "Predictive Audiences pair Comscore's consented consumer insights with AI to identify the content patterns most closely tied to a given audience." The segments, in other words, target programs and content that the model associates with an audience, rather than devices or households matched to a profile.

Three example segments appear in the document, each with its full taxonomy path:

  • Predictive Audience > Eyeota > Demo > Life Events - New Homeowners
  • Predictive Audience > Epsilon > Lifestyles & Interests > Home Improvement & DIY Enthusiasts
  • Predictive Audience > AnalyticsIQ > Interest > DIY Doers

Each path carries the name of a third-party data provider, which indicates that the predictive segments start from those providers' audience definitions. That matches what Comscore said in September 2025, when Proximic launched its AI-powered Data Partner Network with 13 named providers, among them AnalyticsIQ and Eyeota, described as a Dun & Bradstreet company. Epsilon was not among the 13; the announcement listed it among previous integrations. Epsilon belongs to Publicis Groupe, which confirmed its $4.40 billion cash acquisition of the company in April 2019. At that launch, Comscore said MiQ had reported 30% stronger engagement with Comscore predictive segments than with traditional ID-based segments across CTV campaigns. That earlier comparison, like today's, came without a published methodology.

The case study does not explain how a segment such as "DIY Doers" is translated into content patterns, which content signals SpringServe passes to Proximic, or how often the model is updated. It also leaves open what "consented consumer insights" covers.

A campaign that ran before the launch

The dates are worth noting. When Comscore announced on October 28 that Proximic's Predictive Audiences were coming to SpringServe, it described the deal as the first real-time contextual integration across SpringServe and named Future Today, a free ad-supported streaming publisher, among the first sellers. Future Today cited "preliminary findings" with no figures. The TMB campaign ran in the third quarter, between July and September, so it predates that announcement. Today's document adds the numbers that announcement lacked, and it comes from a different publisher.

Jessica Trainor, identified in the case study as Head of Partnerships at Proximic by Comscore, is the only person quoted. Comscore's October 28 announcement gave her title as SVP Partnerships.

"As advertisers look for efficient ways to connect with their audiences across CTV, the ability to reach relevant consumers without relying on identifiers can be valuable," Trainor said. "This campaign demonstrates how Comscore's Predictive Audiences can help advertisers extend their reach while making their media investments work harder. Through our integration with Magnite SpringServe, we're making these capabilities more accessible and actionable across premium streaming environments."

No TMB executive and no representative of the retailer is quoted. The document closes with a sales line inviting readers to contact Magnite about the product.

Why ID-free targeting is being pushed in CTV

Audience targeting on television screens has long depended on identity resolution - linking a household's devices and IP address to a profile held by a data company. The accuracy of that chain is disputed. A study commissioned by CIMM and Go Addressable found IP-to-postal address linkages accurate only 13% of the time on average, a result PPC Land reported in November 2025.

Buyers have also been moving where they apply data. OpenX's first annual supply-side intelligence report, released on October 15, put supply-side targeting at 51% of connected TV deal spend on its US marketplace in 2025, up from 9% in 2022. Within those deals, third-party audience data fell from 78% of spend in 2022 to roughly 40% in 2025, while quality signals rose to 35%. The TMB campaign fits that pattern: the seller attached the audience data in its own ad server before the inventory reached the buyer.

Demand for contextual tools is growing more slowly than it was. Proximic's own 2026 State of Programmatic Report, based on more than 200 decision-makers surveyed in November 2025, found that 74% planned to maintain or increase contextual targeting, but only 43% planned an increase, down from 54% a year earlier. The same survey expected CTV to take 26% of media budgets in 2026.

Comscore has been selling ID-free audiences to TV platforms for some time. In August 2024, it agreed to make more than 1,000 Proximic ID-free interest and purchase-intent audiences available in FreeWheel's Audience Manager. With SpringServe, the segments now sit in Magnite's publisher ad server as well as FreeWheel's.

The business stakes for both companies

For Comscore, evidence that the product works has commercial weight. The company reported on November 3 that third-quarter revenue fell 14.3% to $76.2 million and that it lost $65.5 million after a $57.3 million goodwill write-down. Its Cross-Platform segment, which includes Proximic, slipped 0.9% to $12.2 million as lower usage of Proximic products was offset by new content measurement business.

Magnite reports from a stronger position. On November 4, it posted CTV contribution ex-TAC of $105.4 million for the third quarter, up 39%, or 53% of its total. Data and targeting tools built into SpringServe give publishers more ways to package that supply.

Magnite has led with a large percentage before. On October 21, Magnite said ad recall lift on its campaigns ran 136% above Cint's benchmarks, without disclosing the number of campaigns, sample sizes or dates.

Why this matters for the marketing community

For CTV buyers, the case study offers a specific claim that ID-free segments can match or beat identity-based audiences on cost, at least on one publisher's inventory in one quarter. Without the definition of efficiency, the measurement source or a comparison design, the 43% cannot be compared with results from other vendors, or even with Comscore's earlier MiQ figure.

For publishers, the more practical point may be the 10%. A seller that can fill the last tenth of a campaign's audience goal from its own inventory, without the identifiers its supply may lack, keeps that budget instead of losing it to under-delivery. The case study does not show how often that happens, or at what price TMB sold the extra impressions.

The open questions are concrete: what the efficiency metric is, who measured the audience, what the CPMs were for each line, whether the data fee for the predictive segments differs from the fee for the ID-based ones, and how many impressions the campaign served. None is answered in today's document.

Timeline

Summary

Who: Magnite, which published the case study; Comscore's Proximic division, which supplies Predictive Audiences; Trusted Media Brands, the publisher that ran the campaign; and an unnamed retailer. Jessica Trainor of Proximic by Comscore is the only person quoted.

What: A case study claiming that Comscore's ID-free Predictive Audiences, activated through Magnite's SpringServe on a publisher direct deal, delivered the retailer's target audience of homeowners and DIY enthusiasts 43% more efficiently than ID-based audiences in the same campaign and added 10% more target-audience delivery. The efficiency metric, measurement source, CPMs, impressions and spend are not disclosed.

When: Published today, November 9, 2026. The campaign ran in the third quarter of 2026, before Comscore announced the SpringServe integration on October 28.

Where: On Trusted Media Brands' CTV inventory in the United States, served and targeted through SpringServe.

Why: Identity-based targeting on TV screens rests on household and IP matching whose accuracy is disputed, and buyers are moving data activation to the sell side. Comscore, which reported lower Proximic usage in the third quarter, and Magnite, whose CTV business now makes up 53% of contribution ex-TAC, both have reasons to show that content-based audience predictions can stand in for identifiers.