Magnite's new integration with Truthset, announced yesterday, addresses a problem most CTV buyers know exists but rarely quantify: by the time an audience segment reaches the bid stream, it has already lost roughly half its accuracy. Every identity resolution hop between data provider and impression introduces error. Truthset CEO Scott McKinley put it bluntly to AdExchanger:
You basically lose about half your data, and you introduce about 50% error every time you move it through one of these ID resolution hops.
Scott McKinley, CEO of Truthset
For a channel projected to hit $38 billion in U.S. ad spend this year, that degradation translates into billions of dollars chasing the wrong households. The Magnite-Truthset deal doesn't eliminate the problem, but it does something more useful for buyers: it makes the problem visible and, for the first time, configurable at the SSP level.
The Compounding Error Problem
Audience data doesn't degrade linearly. Truthset's own analysis shows that a typical data provider audience might start at 55% accuracy, but after onboarding into a marketplace, it drops to 33%. Then it passes through CDPs, DMPs, DSPs, and SSPs, each with its own identity resolution layer. IP-to-household matching, cookie syncs, MAID lookups: every translation introduces noise.
The result is that by the time a buyer activates an audience segment in CTV, the segment may bear only passing resemblance to the original intent. A campaign targeting 35-year-old suburban parents might actually be reaching a mix that includes 65-year-old empty nesters, college students, and households that moved six months ago. The buyer pays premium CTV CPMs (roughly 3.4x higher than open-web display) for impressions that never had a chance of converting.
This isn't a new problem, but CTV's fragmented supply chain makes it worse. Unlike display, where cookies at least provided a consistent (if flawed) identity layer, CTV relies on a patchwork of IP addresses, device graphs, and authenticated logins. Truthset's 2026 State of Data Accuracy report documents how these inaccuracies compound across the CTV supply chain, from targeting through measurement.
What the Integration Actually Does
Magnite's integration pipes Truthset's Data Rated Audiences directly into the SSP. This means buyers can now filter inventory by data accuracy tier before bidding, rather than discovering accuracy problems after the fact in post-campaign analysis.
Truthset's rating system assigns accuracy tiers (AAA, AA, A, B) to audience segments based on its patented Truthscore methodology, which measures demographic accuracy at the individual record level across 95% of U.S. Census coverage. The Data Collective includes major providers like Experian, Alliant, MRI-Simmons, and TransUnion, who give Truthset access to their data for independent accuracy scoring.
For sellers, the appeal is yield optimization. As McKinley explained, publishers can now "parse it into super premium, premium, regular and inexpensive" tiers, pricing data based on quality rather than volume. A smaller, more accurate audience pool commands higher CPMs; a larger, less precise pool goes for less to buyers prioritizing reach over precision.
For buyers, the appeal is workflow integration. Mike Treon, head of CTV and video strategy at independent agency PMG, told AdExchanger that the integration lets his team "put it through this lens of accuracy" rather than matching audiences directly into platforms and hoping for the best.
The CFO Question: What's the Actual Lift?
The business case for data quality is intuitive but rarely modeled. If 50% of your audience data is wrong, you're wasting 50% of your media spend, right? Not quite. The math is more nuanced.
Consider a campaign with a $1 million CTV budget targeting in-market auto intenders. If the audience segment is 55% accurate at the source and degrades to 33% by activation, roughly two-thirds of impressions reach households with no purchase intent. At a $30 CPM, that's $670,000 in wasted spend.
But the waste isn't evenly distributed. Some of those "wrong" households might still convert through serendipity or brand awareness effects. The real question is incremental lift: how much additional conversion do you get from accurate targeting versus random exposure?
Truthset's Databricks integration announcement cited 30-60% of ad spend as "wasted" due to mis-targeting. That range is wide enough to suggest the actual number depends heavily on category, creative, and campaign objectives. Brand campaigns with broad appeal waste less than performance campaigns with narrow targets.
The pilot design here matters. A proper test would run identical creative against AAA-rated audiences versus unrated audiences, with holdout groups, measuring incremental conversions over a 60-90 day window. The expected outcome: higher CPMs for AAA inventory, but lower effective cost per acquisition if the accuracy claims hold.
PMG's Angle: Alli and the Accuracy Stack
PMG's involvement in the announcement isn't coincidental. The agency has been building its Alli platform as an operating system for marketing data, and Truthset is already a foundational partner in the Alli Marketplace launched at Cannes last year.
The strategic logic: if you can validate audience accuracy before activation and measure it after, you can build a closed-loop system that optimizes for actual outcomes rather than proxy metrics. Treon's comment about CTV "further isolating these islands of data and inventory" points to the real challenge. Fragmentation isn't going away. The question is whether you can build quality controls that work across fragmented supply.
For PMG's enterprise clients, this means the ability to audit data quality at the segment level, adjust bids based on accuracy tiers, and attribute conversions back to data quality as a variable. That's a meaningful capability for brands spending eight figures on CTV annually.
Risks and Limitations
Three caveats worth noting before treating this as a solved problem.
First, Truthset's accuracy ratings are based on its own methodology and data collective. The ratings are independent in the sense that Truthset doesn't sell the underlying data, but they're not third-party audited in the way financial statements are. Buyers should treat the ratings as directional rather than absolute.
Second, the integration addresses data quality at the SSP level, but it doesn't fix upstream problems. If a brand's first-party data is dirty, or if the CDP introduces errors before the data reaches Magnite, the Truthset layer won't catch it. Data quality is a supply chain problem, and this addresses one link in the chain.
Third, premium accuracy tiers will command premium prices. The economic question is whether the lift justifies the cost. For some campaigns, the answer will be yes. For others, particularly broad awareness plays, the unrated inventory might deliver acceptable results at lower CPMs.
The Pilot Checklist
For teams considering a test, here's a starting framework:
- Run a 4-6 week A/B test comparing AAA-rated inventory against standard inventory, same creative, same frequency caps, same dayparts.
- Measure incremental conversions using a holdout methodology, not just attributed conversions.
- Calculate effective CPA at each tier, accounting for the CPM premium.
- Document match rates and reach at each tier to understand the scale tradeoff.
The expected finding: AAA inventory delivers higher conversion rates but at smaller scale. The optimization question becomes how to blend tiers to maximize total conversions within budget, not how to maximize accuracy in isolation.
Magnite and Truthset have built a useful tool. Whether it changes your CTV economics depends on whether you're willing to run the experiment and follow the math wherever it leads.
