On June 22, 2026, DataBeat published the June edition of its US Programmatic Trends Report and, for the first time in the series, made autonomous, AI-driven buying its featured theme. The comparison it drew - what the report calls agentic buying set against conventional, non-agentic buying - produced a split verdict. Measured on the two metrics DataBeat selected, cost per thousand impressions and fill rate, neither approach won outright. The programmatic buyers paid publishers more for each impression. Agentic buyers converted a larger share of the auctions they entered. The report presents the gap between the two as narrow. The numbers underneath run wider than the language around them.

What the data shows

DataBeat, a programmatic analytics provider that has partnered with revenue-operations firm MediaMint, builds the report from anonymized data across its partner network, tracking more than $55 million in monthly revenue, 35 billion monthly impressions, and signals from over 200 bidders. The same network and methodology sit behind the earlier monthly editions of the series, which reported an overall CPM decline of 32.5% month over month in January against 23.6% annual growth. The June edition benchmarks May 2026 results against April 2026 and May 2025, and its agentic section compares two buyer populations transacting across that same marketplace.

On price, non-agentic demand led. Programmatic buyers cleared at an average CPM of $6.95, against $6.13 for agentic buyers, a 13.4% premium for conventional demand in the report's own calculation. CPM, or cost per mille, measures what a publisher is paid to serve one thousand impressions. On fill rate, the share of bid opportunities a buyer converts, agentic buying led, at 0.204% versus 0.183%, an 11.5% edge. The third figure is the one DataBeat leans on hardest: agentic buyers reached those results while taking part in what the report describes as "86% fewer auctions" than their non-agentic counterparts: programmatic.

According to DataBeat, agentic buying "delivers CPM and fill rate performance broadly in line with Programmatic (Traditional) Buying." For a category still in early adoption, the report concludes, the numbers show that agentic demand has "evolved beyond a niche channel into a credible market participant."

Where the framing and the numbers diverge

The report describes the price shortfall as marginal. Agentic CPMs are "only slightly lower than Non-Agentic Buying CPMs," it states, crediting the near-parity to selective bidding: "AI-driven buyers are targeting inventory more selectively while maintaining comparable pricing." Yet by the report's own arithmetic the gap it minimizes is no smaller than the gap it highlights. Non-agentic CPMs run 13.4% above agentic; agentic fill rate runs 11.5% above non-agentic. Both figures take the lower value in each pair as the base, and on a like-for-like footing the two spreads sit within a couple of points of each other - a $0.82 difference on CPM and a 0.021-point difference on fill rate. The report presents the fill-rate gap as evidence of agentic efficiency, and the CPM gap, comparable in size and the larger on its own figures, as a rounding detail.

That asymmetry matters because the two metrics do different work. Fill rate measures how often a buyer wins what it bids on. CPM measures what the publisher collects when it does. For a publisher deciding whether to court agentic demand, the price per impression is the direct revenue input, and on that input the report's own data shows conventional demand commanding a 13.4% higher CPM than the agentic buyers it is measured against.

A second tension sits inside the selectivity explanation. DataBeat argues that agentic buyers concentrate spend on "opportunities that best align with campaign objectives," language that implies a move toward higher-value inventory. Premium impressions, in an open auction, tend to attract more competition and clear at higher prices, not lower. Agentic buyers in this dataset did the reverse. They paid less, won more often, and did so across far fewer auctions. That combination - lower clearing prices, higher win rates, thinner participation - is equally consistent with a buyer concentrating on lightly contested inventory, where competition is thin and impressions are cheap to win. The report does not separate the two readings. It advances a confident interpretation and then, a few lines on, hedges it: the higher fill rate, DataBeat notes, is "worth assessing whether this stems from inventory quality or win-rate optimization."

The fill-rate figures deserve a closer look

The fill-rate comparison also rests on unusually small absolute numbers. A rate of 0.204% for agentic buying and 0.183% for non-agentic buying describes, in each case, a fraction of a single percent. The 11.5% difference the report highlights is the distance between those two fractions, not a gap in overall inventory monetization. A conventional programmatic fill rate, measured as filled impressions against total ad requests, typically runs far higher than one percent. DataBeat does not define how it derived these figures or explain why the base rates land where they do, which makes the relative comparison hard to place in context. A relative advantage of 11.5% carries very different weight depending on whether the underlying numbers describe most of a publisher's inventory or a sliver of it.

Efficiency, and for whom

The report's central word for agentic buying is efficiency. Agentic buyers achieve "more efficient bidding and supply conversion," DataBeat writes, and the section closes on the argument that "efficiency is becoming as important as scale." As adoption grows, it adds, "inventory quality and signal strength are likely to matter more than auction volume."

Read from the buy side, that case holds together. Fewer auctions, higher win rates, and comparable outcomes describe a buyer spending less effort to reach its goals. Read from the sell side, the same figures raise a question the report does not resolve. Efficiency for a buyer that pays less per impression is not obviously the same thing as higher yield for the publisher serving it. Whether agentic demand lifts or dilutes a publisher's revenue depends on what it displaces - on whether agentic bids capture impressions that would otherwise have gone unsold, or impressions a higher-paying non-agentic buyer would have taken.

That distinction is not academic for publishers. The friction between buy-side efficiency and sell-side yield has recurred across the programmatic market for years; publishers have restricted the transaction data buyers use to map cheaper routes to inventory, arguing that supply-chain transparency can compress the prices they collect. Agentic buying, which the report describes as bidding more selectively across a smaller set of auctions, folds a new variable into that same equation.

How the report defines agentic buying

DataBeat draws a specific line between the two categories. Conventional buyers, it writes, tend to purchase "across broader inventory pools and audience segments." Agentic buying, by contrast, uses "coordinated AI agents to autonomously plan, execute, and optimize campaigns in real time," adjusting bids, inventory selection, and budget allocation against predefined goals as live performance signals shift. The line the report draws is autonomy at the point of execution, not the presence of automation, which programmatic buying has depended on for more than a decade. Both sides of the comparison are programmatic, a point the report itself makes when it frames the contrast as occurring "in programmatic advertising," and both are automated. That makes the standalone label "Traditional Buying" imprecise. The genuinely traditional channel - insertion-order direct buying, negotiated by hand at fixed prices and reserved placements - is not programmatic and sits outside this comparison entirely. What the report labels traditional or non-agentic buying is conventional programmatic demand: algorithmic, auction-based, and automated, but run by human traders rather than autonomous agents. The contrast the data captures is between two modes of automated programmatic buying, not between programmatic and the direct channel that preceded it.

The comparison uses SSP marketplace data, DataBeat says, "including leading SSPs such as PubMatic," and it points to "emerging solutions like AgenticOS" as the mechanism bringing agentic demand to market. That reference identifies part of the pipe the agentic demand in the dataset likely flows through. PubMatic launched AgenticOS on January 5, 2026, positioning it as an operating system for autonomous advertising execution, with WPP Media, Butler/Till, and MiQ among its first participants.

A short history behind the numbers

The DataBeat comparison arrives roughly six months into the commercial life of the infrastructure it measures. The Ad Context Protocol, an open standard built on Anthropic's Model Context Protocol, launched on October 15, 2025 with six founding companies, giving AI agents a shared language for discovering inventory and activating campaigns across platforms. Sell-side platforms built agentic layers on top of that standard through late 2025 and into 2026. Magnite introduced its Orchestration coordination layer for agentic buying on June 11, 2026, with dentsu and DIRECTV Advertising among its beta partners.

Early vendor-reported results ran ahead of the near-parity DataBeat now records. In a December 2025 campaign on AgenticOS, PubMatic and the agency Butler/Till reported 40% more impressions than comparable non-agentic activity, a 5.5x cost efficiency for the agency, and a 98% average video completion rate - figures the vendor cited for a single test. DataBeat's marketplace-wide reading, drawn from May 2026 data across its network, describes something more measured: agentic demand tracking close to conventional programmatic buying rather than beating it, and trailing on price.

Scrutiny from the publisher side has grown alongside the buy-side momentum. Prebid, the open-source header-bidding project, has spent 2026 building agentic tooling from the publisher's side of the transaction, on the argument that publishers without representation in the protocols governing machine-to-machine buying risk becoming price-takers in a system built for buyers. The same stretch has seen publishers pressing structural questions about where automation sends revenue, a debate sharpened by products that reorganize monetization around session value rather than impression volume as search referral traffic erodes.

Why it matters for the marketing community

For publishers and the buy-side teams weighing agentic demand, the DataBeat report offers a network-scale benchmark rather than a single vendor's case study, and its value lies partly in the questions its own data raises. The conclusion is unambiguous: agentic buying has, in the report's words, "moved beyond the experimental stage" and stands to become "a larger contributor to publisher monetization" as adoption widens. The figures beneath that conclusion are more mixed. Agentic demand pays less per impression, wins a marginally larger share of the fewer auctions it enters, and does so on absolute fill rates the report leaves unexplained.

None of that contradicts the claim that agentic buying is now a real market participant. It does complicate the assumption that its arrival is straightforwardly good for the sell side. On the single metric that determines what a publisher earns each time an impression clears, the report's May 2026 data shows conventional programmatic demand commanding a 13.4% higher CPM than the agentic buyers displacing it, a gap the report describes as slight. Whether the gap narrows, widens, or holds as agentic volume scales is the question the June report sets up and does not answer.

Timeline

  • October 15, 2025: The Ad Context Protocol launches with six founding companies, built on Anthropic's Model Context Protocol, establishing a shared standard for AI agents to transact across advertising platforms.
  • December 2025: PubMatic and agency Butler/Till run an early agentic campaign on AgenticOS, reporting 40% more impressions, a 5.5x cost efficiency, and a 98% average video completion rate for a single brand test.
  • January 5, 2026: PubMatic launches AgenticOS, an operating system for autonomous advertising execution, with WPP Media, Butler/Till, and MiQ among early participants.
  • May 2026: The data period underlying DataBeat's agentic-versus-conventional comparison.
  • June 11, 2026: Magnite launches its Orchestration coordination layer for agentic buying, with dentsu and DIRECTV Advertising among beta partners.
  • June 22, 2026: DataBeat publishes its June US Programmatic Trends Report, making agentic buying the report's central theme.

Summary

Who: DataBeat, a programmatic market-intelligence provider that has partnered with revenue-operations company MediaMint, whose network data underpins the report. The analysis sets AI-driven agentic buyers against conventional non-agentic buyers, with PubMatic named as a data source and its AgenticOS platform cited as a route agentic demand reaches the market.

What: DataBeat's June US Programmatic Trends Report found agentic buyers clearing at a $6.13 average CPM against $6.95 for conventional non-agentic buyers, a 13.4% premium, while agentic buyers posted a higher fill rate, 0.204% against 0.183%, an 11.5% edge, in what the report describes as 86% fewer auctions. DataBeat frames the near-parity as evidence agentic buying is now a credible market participant, though its own figures show a wider CPM gap than the "only slightly lower" language implies and leave the sub-one-percent fill rates undefined.

When: The report was published on June 22, 2026, drawing on May 2026 marketplace data benchmarked against April 2026 and May 2025.

Where: The United States programmatic advertising market, measured across DataBeat's partner network of more than $55 million in monthly revenue, 35 billion monthly impressions, and over 200 tracked bidders.

Why: Agentic buying, autonomous AI agents that plan, bid, and optimize campaigns without step-by-step human control, has moved from pilots to live spend over roughly six months, and the report is among the first network-scale attempts to measure whether it monetizes publisher inventory differently. Its split result, a higher fill rate but a lower CPM, leaves publishers to weigh whether growing agentic demand lifts yield or discounts it.