AudienceProject released Elevate 3.0 on August 27, 2026, adding an AI assistant called Audrey, automatic CTV benchmarking and a rebuilt interface to the measurement environment it opened to broadcasters and publishers across ten markets two months earlier.

The update, published on the company blog by Aske Ostergaard, arrives roughly nine weeks after AudienceProject made Elevate available to media owners for the first time. That June release ended a period in which the environment served only large platforms. Elevate 3.0 does not extend the footprint. It reworks what sits inside it.

Three changes define the release. A redesigned user interface replaces the previous reporting layout. Scaled benchmarking compares a campaign's on-target percentage against the CTV benchmark for the same target audience in the same country. And Audrey, described by the company as its AI assistant, reads the data inside a report and writes observations about what it contains.

What the interface changes

According to AudienceProject, the refreshed interface introduces clearer visualisations, redesigned reporting widgets and a more consistent experience across campaigns and markets. The stated goal is navigation rather than new measurement: helping users move through reports and interpret reach, frequency, on-target performance, demographics and TV viewership without assembling the picture manually.

The metric set is unchanged from June. When Elevate opened to broadcasters and publishers on June 22, 2026, the same five categories anchored the product, alongside reach efficiency expressed as a cost per thousand within a defined target group and a non-TV viewer figure showing how much of a reached audience does not watch linear television. Elevate 3.0 changes how those numbers are presented, not what they measure.

Erlingur Jonsson, VP User Experience at AudienceProject, framed the redesign in terms of data volume. "Audience measurement can involve a lot of data, so our focus has been on making the experience as intuitive as possible," he said. "The new interface is designed to reduce complexity and make it easier for users to quickly understand what matters, move between insights and communicate campaign performance clearly."

Interface work is rarely the headline in measurement releases. It matters here for a specific reason. Elevate exists so that a broadcaster's commercial team can put a number in front of an advertiser without commissioning a bespoke study. A report that requires interpretation before it can be shown is a report that gets shown less often.

Benchmarking shifts its reference point

The second change carries more analytical weight than the first.

An on-target percentage records how much of a campaign's measured audience fell inside the intended target audience. AudienceProject notes in the announcement that the percentage alone does not always show whether the result was strong for that particular audience. A 60 percent on-target figure means something different for a broad demographic than for a narrow one, and different again in a market where addressable inventory is scarce.

Scaled benchmarking addresses that by comparing performance automatically against the CTV benchmark for the target audience in the same country. Users see whether a campaign landed above, in line with or below that reference. Because the benchmark reflects the target audience rather than a general market average, AudienceProject argues it enables fairer comparisons.

Some form of benchmark flag was already visible in June. PPC Land's reporting on that launch documented an interface panel showing a 66.2 percent on-target percentage marked as 17 percent above benchmark. What Elevate 3.0 describes is the systematisation of that comparison across audiences and countries rather than its introduction. The announcement does not state how the CTV benchmarks are constructed, how frequently they refresh, what sample sizes underpin them in smaller markets, or whether a benchmark exists for every target audience a user might define.

The reference class has also moved. AudienceProject's Q4 2025 roadmap, published on October 2, 2025, described a benchmarking feature built on historical comparison, letting advertisers measure current campaigns against their own previous efforts. That is self-referential: it answers whether a campaign improved on the last one. Scaled benchmarking in Elevate 3.0 answers a different question, comparing a media owner's delivery against a market-level CTV standard. For a sell-side team arguing that its inventory delivers efficiently, the second comparison is the one an advertiser is more likely to accept.

Audrey and the boundaries drawn around it

The third component is the one most likely to attract scrutiny.

Audrey analyses the measurement data within a report and presents observations about campaign performance. According to AudienceProject, it can summarise results, highlight trends and surface notable audience delivery patterns, reducing the time spent reviewing reports manually.

Two constraints in the announcement deserve attention. The first is scope: Audrey's insights are based only on the data within the relevant campaign report. That rules out cross-campaign inference, external market context and comparison against data the user has not loaded. It is a narrow window, and narrowness is the point - a system confined to one report has fewer opportunities to assert something the underlying data does not support.

The second is positioning. AudienceProject states that Audrey is designed to support, rather than replace, users' own analysis and decision-making. The company does not describe a review step, an accuracy measurement, or a mechanism by which a user can trace an observation back to the specific figures that produced it.

Rahil Shaikh, Product Owner, Elevate at AudienceProject, tied the three components together. "Elevate 3.0 is an important step towards making audience measurement not only more accessible, but more actionable," he said. "By combining clearer reporting, relevant benchmarks and AI-powered insights, we are helping media owners move faster from data to understanding, and we will continue building on this foundation to make measurement an even stronger part of how they demonstrate value."

Several details are absent. The announcement names no underlying model, states no availability date beyond the release itself, and specifies no market or customer-tier restrictions on Audrey. No pricing appears, consistent with the June release, which also directed interested parties to a demo request form.

Ten markets, unchanged

Elevate remains available to broadcasters and publishers in the United Kingdom, Germany, Italy, Spain, Poland, Sweden, Denmark, Norway, Finland and Mexico. The list is identical to the one published in June.

That stability is worth stating plainly, because AudienceProject's release cadence over the preceding year has been driven substantially by geography and integrations. Netflix measurement went live across five European markets on October 1, 2025. Disney+ measurement followed on January 21, 2026 across the same territories. Poland arrived in March 2026. Pinterest became measurable in the UK and Germany on June 18, 2026. Elevate 3.0 breaks that pattern by adding capability inside an existing footprint rather than extending the footprint itself.

The version number is itself new. The June 22 announcement carried no version designation, and AudienceProject has not published a public description of what constituted Elevate 1.0 or 2.0. The numbering appears to date from the product's roughly two-year history serving large platforms, a period during which the environment was not marketed to media owners.

The commercial case behind the release

The reason a reporting redesign registers as news sits in what Elevate is for.

AudienceProject's argument, set out in its white paper distributed on April 22, 2026, is that the industry lacks a shared, verifiable definition of reach across media, leaving substantial advertising investment optimised against incomplete data. Elevate applies that argument to the sell side. A broadcaster or publisher that can produce structured, comparable documentation of audience delivery competes on different terms than one producing bespoke reports for each pitch.

That framing has commercial backing. AudienceProject and Omnicom Media Italy announced a cross-media measurement partnership on May 27, 2026, the first such deal for a communications group in that market. The Trade Deskformalised a dual partnership with AudienceProject on February 25, 2026, enabling campaign measurement within AudienceReport and placing AudienceProject socio-demographic segments in the demand-side platform's Data Marketplace across Europe. In France, Mediametrie announced a partnership on September 4, 2025 to accelerate its cross-media video measurement solution. Each of those relationships raises the value of being visible inside AudienceProject's measurement environment.

Benchmarking sharpens that dynamic. A media owner whose on-target delivery sits above the country CTV benchmark for a given audience now has a defensible number. A media owner whose delivery sits below it has a number that an agency can see too. Comparability cuts in both directions, which is a structural feature of any benchmarking system rather than a flaw in this one.

AI assistants arrive across the measurement stack

Elevate 3.0 lands in a period when analytical assistants have become a standard component of advertising platforms rather than a differentiator.

The Trade Desk released Zuma on August 27, 2026, the same day as Elevate 3.0, including a Koa AI assistant in closed beta that coordinates specialised agents across campaign creation, audience building, troubleshooting and performance analysis. Google added prompt-generated dashboards and insight cards to Google Ads and Google Analytics on August 10, 2026, including benchmarking against anonymised peer averages. iSpot launched a pair of creative analysis agents on February 10, 2026 after two years of development.

Audrey is narrower than most of these. It does not build campaigns, adjust bids or query external datasets. It reads one report and describes it. That restraint is defensible in a measurement product, where the commercial value depends on the numbers being trusted by a counterparty who did not generate them.

Trust in the underlying data remains the harder problem. An Adstra and InterMedia study reported in July 2026 found that only 23 percent of residential IP addresses matched correctly at the household level, with IP-to-postal matching reaching 13 percent accuracy in Truthset's parallel work. An assistant that summarises a report accurately still inherits whatever error exists in the measurement beneath it.

What media owners face next

Independent measurement of broadcaster and publisher inventory is being contested by more than one vendor. Nielsenextended Four-Screen Ad Deduplication in Japan to all measurable CTV inventory in Nielsen ONE Ads on August 3, 2026, a move that positioned inclusion in a buyer-recognised measurement framework as a commercial question rather than a technical one. Guideline extended its CPM benchmark framework to CTV, audio streaming and podcasting on April 30, 2026, covering more than 60 US publishers. In Poland, one of Elevate's ten markets, Nielsen's monthly viewing figures now sit alongside a second documented measurement route for the first time.

For a broadcaster sales team, the practical question raised by Elevate 3.0 is narrow and immediate: whether an automatically generated benchmark comparison and an AI-written summary shorten the distance between a completed campaign and a renewed order. AudienceProject has published no data on that. The announcement describes reduced manual review time without quantifying it, and offers no case study, beta cohort or customer count.

What the release does establish is direction. Two months after opening Elevate to the sell side, AudienceProject is iterating on presentation and interpretation rather than on coverage - a signal that the constraint the company sees is not which inventory can be measured, but how quickly a measured result can be turned into an argument.

Timeline

Summary

Who: AudienceProject, a Copenhagen-based independent cross-media measurement company, published the release. Aske Ostergaard authored the announcement. Erlingur Jonsson, VP User Experience, and Rahil Shaikh, Product Owner for Elevate, provided comments. The intended users are broadcasters, publishers and platforms selling advertising inventory.

What: Elevate 3.0, a version update to AudienceProject's measurement environment for media owners. It introduces a redesigned user interface with new reporting widgets, scaled benchmarking that automatically compares a campaign's on-target percentage against the CTV benchmark for the same target audience and country, and Audrey, an AI assistant that summarises results and highlights trends using only the data contained in the relevant campaign report.

When: August 27, 2026, approximately nine weeks after Elevate opened to broadcasters and publishers on June 22, 2026.

Where: The United Kingdom, Germany, Italy, Spain, Poland, Sweden, Denmark, Norway, Finland and Mexico - the same ten markets covered at the June launch, with no expansion announced.

Why: AudienceProject positions structured, comparable audience documentation as a condition of competing for advertiser budgets. An on-target percentage carries limited meaning without a reference point, and scaled benchmarking supplies a country and audience-specific one. Audrey targets the time cost of interpreting reports. Neither addresses coverage, which is where the company's previous releases concentrated, indicating a shift toward speed of interpretation rather than breadth of measurement.