Adelaide, the attention measurement company, has put a self-service Media Quality Calculator online that scores advertiser site lists against its AU rating and emails back a PDF report, a tool its co-founder Zach Kubin promoted on LinkedIn in late August 2026 alongside a claim that more than $27 billion a year flows into lower-quality programmatic inventory.

In Short

Adelaide built a web page where an advertiser uploads a list of the websites and apps its ads run on, and the company sends back a report grading that list with its own 0 to 100 quality score. It matters because many brands buy ads through automated systems and do not always know how much of their money lands on poor placements. What changes is that a first read on media quality now takes an email address and a spreadsheet rather than a sales call, although the grades come from the same company that sells the scoring.

What the calculator does

The product sits on a single landing page under the heading "How much of your media is working against you?" and works in three steps, according to Adelaide. A visitor enters a work email, which the company uses to deliver the results. The visitor then uploads a list of domains or app bundle IDs. Finally, the analysis arrives, which Adelaide describes as showing where a list "is strong, and where it isn't."

Kubin, who is listed as co-founder of Adelaide, set out the contents of the report in his LinkedIn post. According to the post, the custom PDF covers four areas: how a list's average AU compares with Adelaide benchmarks, how much of the list falls below AU benchmarks, how AU varies across Display, OLV and CTV, and where the largest gaps and reinvestment opportunities sit. The landing page splits the same output into two blocks. Overall quality insightscomprise the average AU across the uploaded list, the share of domains falling below AU benchmarks and a single "biggest optimization opportunity." Channel-level insights comprise average AU by channel, benchmark comparisons and the distribution of the list across high and low-AU quality tiers.

Kubin framed the purpose bluntly. The calculator, he wrote, "shows brands how much low-AU media is hiding in their plans." He added that "everything is automated," and that results were expected within minutes of an upload.

No price is attached. The landing page describes the output as a free quality breakdown, and the post states that no spend data is required. The exact release date is not given in either document. Kubin's post carried a relative timestamp of three weeks and an "Edited" flag when captured, and it describes the tool as recently released, which places its availability in the second half of August 2026.

Inside the upload: 200 rows minimum, 100,000 maximum

The technical constraints are narrow, and they say a good deal about what the tool can and cannot see. According to the requirements printed on the landing page, the file must be a single-column CSV containing "No channel, placement or performance data." Each row carries one domain or one app bundle ID. A list must contain at least 200 entries and no more than 100,000 rows.

The 200-row floor rules out small curated deals and tight inclusion lists. The 100,000-row ceiling, by contrast, accommodates large open-exchange delivery logs; for comparison, the Association of National Advertisers' December 2023 transparency study found the average campaign running across 44,000 websites.

Because the file holds nothing but identifiers, the calculator is scoring sites and apps, not impressions. That is a meaningful distinction. A domain that carries both a well-built article page and a cluttered slideshow template receives one grade in this exercise, whatever share of a buyer's delivery went to either. Nor can a list weighted 90% to a handful of large publishers be distinguished from one spread evenly across its long tail, since the file carries no volumes. The "share of domains falling below AU benchmarks" is therefore a count of entries, not a share of spend or impressions.

A discrepancy between the post and the page

The two source documents do not describe the input in quite the same way. Kubin's post invites users to "upload a domain list, whitelist, or DSP delivery report." The landing page repeats the delivery-report option in its introduction, referring to a report "straight from your DSP." Yet the file specification on the same page excludes channel, placement and performance data, and a standard delivery report from a demand-side platform is built around exactly those columns: impressions, spend, line items and formats. Neither document explains whether the calculator strips extra columns automatically or whether a user is expected to reduce a delivery report to its domain column before uploading. The distinction has practical weight, since an unmodified report would appear to breach the stated format.

A related question concerns channel. The post promises a breakdown of AU across Display, OLV and CTV, and the landing page promises average scores by channel. The upload, however, is prohibited from containing any channel field. Adelaide must therefore infer channel from the identifier itself. The company has not published how it does so for the calculator. The inference is not trivial: in a July 29, 2026 blog post on CTV and linear TV measurement, Adelaide said it counts as CTV only impressions that ran on an actual television device, with streaming served to desktops, phones and tablets classified under its OLV model. A streaming app's bundle ID or a publisher domain can serve both. How a single identifier is assigned to one channel bucket, or split between them, is not stated.

Three countries only

Coverage is limited. According to Adelaide, the calculator "is currently available in the US, UK, and Canada," and lists containing domains from other regions "may return limited or less actionable results." The company directs anyone wanting AU measurement elsewhere to its sales team. For European advertisers outside the UK, and for global plans that mix local-language publishers with English-language inventory, the report's coverage of a list could be materially lower than its row count.

The $27 billion figure

The headline number is the most prominent claim in both documents and the least documented. The landing page states that "$27 billion is wasted on low-quality programmatic inventory every year." Kubin's post phrases it differently: "Over $27B a year flows into lower-quality programmatic inventory, and few brands know how much of it is theirs." One version describes waste; the other describes spend on lower-quality supply. Those are not the same claim, and neither document names a source, a methodology, a geography or a year.

A figure of similar size has circulated recently. Gracenote research, covered by PPC Land in August 2026, cited ANA benchmark data putting global ad waste attributable to programmatic inefficiencies at $26.8 billion. Whether Adelaide's $27 billion is a rounding of that number, a separate calculation or an estimate built on its own AU data cannot be determined from the materials. The distinction matters because the ANA's definitions of waste cover fees, invalid traffic, non-viewable impressions and made-for-advertising sites, whereas Adelaide's framing on the landing page concerns "low-quality" inventory measured against its own metric.

The broader direction of the ANA work is well established. The association's December 2023 study found that only 36% of post-transaction programmatic budgets reached valid, viewable, measurable, non-MFA impressions. IAB Spain's April 2026 supply-side platform guide, citing the ANA Programmatic Transparency Benchmark 2025, put that share at 41%, with transaction costs consuming 26.1% of every dollar invested.

AU, the metric underneath

AU is described on the landing page as "Adelaide's 0-100 media quality rating that predicts a placement's ability to capture attention and drive business outcomes." According to the company, the score is built on "placement-level exposure signals, eye-tracking data, and real campaign outcome data across billions of impressions." The page lists no model version, update frequency, training period or error range.

It is a predictive score rather than an observed one. PPC Land's July 2026 report on the DV360 integration set out the distinction: Adelaide's methodology analyses signals including ad size, time in view, clutter and position to predict a placement's likelihood of capturing attention, and predictive scores can be computed for every impression in an auction, whereas panel-derived scores cannot. That property is also what makes a domain-list calculator possible. A panel-based attention metric could not grade 100,000 arbitrary sites on request; a model that has already scored placements across its client base can look them up.

The same property sets the calculator's limits. A list-level report can only be as complete as Adelaide's historical coverage of the sites on the list. The documents do not say what happens to domains the model has not observed, how many observations a domain needs before it is graded, or whether unscored entries count toward the 200-row minimum.

The page positions AU explicitly against conventional quality signals. "Low-quality inventory can look efficient on paper, but cheap CPMs and high viewability rates don't mean it's worth buying," it states. Standard viewability confirms only that an ad met a technical on-screen threshold; it says nothing about whether anyone looked.

Vendor figures, without the underlying studies

To support the case, the landing page presents four results by advertiser category, each under the heading that "high-AU media consistently delivers stronger outcomes":

  • Tech: 3x more conversions on high-AU placements.
  • Apparel: 60% higher conversion rate and 28% lower cost per conversion.
  • CPG: 108% lift in ad recall for placements above the AU quality floor.
  • Automotive: 61% lift in awareness using AU-optimized media.

These are Adelaide's own figures. The page does not name the advertisers, the campaign dates, the sample sizes, the comparison groups or who measured the outcomes. It does not say whether "3x" refers to absolute conversions, a rate, or conversions per dollar. Each number compares high-AU media with something, but the baseline is undefined in every case. The figures have not been independently verified.

Independent research has not settled the relationship between attention and effectiveness. A Kantar study of 873 campaigns representing more than $3.2 billion in media spend, published in July 2024, found that overall channel attention levels did not directly correlate with channel-level brand contribution or cost-effectiveness, and suggested some advertisers may be overinvesting in high-attention placements. The Media Rating Council and IAB, in attention measurement guidelines released in November 2025, standardised vocabulary without mandating a single methodology, which means AU remains one proprietary scale among several.

The landing page also displays logos including WPP, OMG, Publicis Media, Coca-Cola, Mars, Prudential and Diageo beneath a statement that AU is "trusted by leading brands and agencies." The page does not say whether any of those companies used the calculator or contributed to the benchmarks against which uploaded lists are graded.

Cheap and viewable, or expensive and viewable?

The calculator's pitch rests on the idea that poor inventory hides behind efficient-looking numbers. Recent data complicates one half of that premise. An analysis published on July 28, 2026 by the Trustworthy Accountability Group, the ANA and Fiducia found that AI-generated inventory graded as premium more than 70% of the time, with an invalid traffic rate of 0.05% against 0.32% for clean supply, viewability of 77.2% against 74.9%, and a TrueCPM of $7.08 against $6.15. The high viewability matches Adelaide's argument. The price does not: that category of low-value supply was more expensive than the clean inventory it displaced, not cheaper.

The finding cuts both ways for a tool like this one. It supports the case for signals beyond viewability and invalid traffic, since the standard checks rewarded the wrong supply. But it also shows that quality problems now include sites engineered to pass whatever grading apparatus buyers adopt. A domain-level score is harder to game than a viewability rate only if the signals behind it cannot be manufactured. The underlying economics of arbitrage, in which traffic is bought cheaply and resold as impressions at a margin, give operators every reason to optimise toward the metrics that determine budget.

What the tool collects

Although no spend data is required, the transaction is not without exchange. Access is gated behind a work email address, and the form requires consent for Adelaide "to store and process" personal data in order to deliver the results. A separate, optional box covers the company's newsletter, updates and promotions.

The uploaded list is itself information. A site list or inclusion list shows where an advertiser or agency buys, and the landing page does not state how long uploaded files are retained, whether they are used to refine benchmarks, or whether they are shared. The consent language references Adelaide's privacy policy but the policy terms themselves are not part of the materials reviewed.

Free diagnostic tools are a familiar entry point in measurement sales. The approach contrasts with FouAnalytics, which in August 2026 priced unlimited verification at a flat $2 million a year, arguing that impression-based fees reward vendors for measuring more traffic. Adelaide's calculator sits at the opposite end of the pricing spectrum: a no-cost sample whose natural next step is a paid engagement.

Where the calculator fits in Adelaide's distribution

Over the past fifteen months Adelaide has placed AU inside a growing set of buying and measurement platforms. In June 2025, Comscore added Adelaide's metric to its Certified Deal IDs through PubMatic. In October 2025, Nielsen integrated AU with reach data inside its Outcomes Marketplace. Later that month, Uber Advertising built a custom AU model with Adelaide and Kantar. In February 2026, Vevo tied its Attention Guaranteed offering to Adelaide measurement, reporting that its YouTube CTV inventory scored 28% above average benchmarks.

Activation followed. On June 10, 2026, Adelaide made AU pre-bid segments available in Amazon DSP, including a Quality Floor that excludes MFA-designated sites and the bottom 10% of AU-scored placements in one step. From mid-July, Google allowed Display & Video 360 advertisers to use AU as a signal in Custom Bidding. By late July, InMobi's CTV attention bundles accepted outcomes reported by either Lumen Research or Adelaide.

The calculator sits upstream of all of these. Where the Amazon and Google integrations apply AU at the point of the bid, the calculator applies it to a plan before any campaign is set up, using the same benchmarks. It creates a route from a diagnostic PDF to pre-bid filtering in the platforms where AU is already available, which makes it as much a distribution tool as a measurement one.

Why this matters for buyers and publishers

For agencies and brands, the report offers a benchmarked read of site lists that previously required a tagged campaign and a contract. That lowers the cost of an initial comparison, and it arrives when the industry has a documented reason to question conventional quality metrics. Yet the output is a count-based grade on a proprietary scale, produced by the vendor that sells the remedy, from benchmarks that are not published. A list judged weak by AU is weak by AU's definition; the same list might score differently against IAS's Quality Attention, DoubleVerify's attention product or Lumen's observational measures, none of which share a common unit.

For publishers, the implications are less visible but real. Every uploaded list is graded site by site, and a publisher that falls below AU benchmarks could appear as a named line in a buyer's "biggest gaps" section without knowing it. A free buy-side tool that ranks domains extends the reach of that scoring well beyond the campaigns Adelaide measures directly, and publishers have no visibility into how their properties are presented in the reports.

The geographic limit keeps the effect concentrated in three English-speaking markets for now. And the unanswered questions, from the provenance of the $27 billion to the treatment of unscored domains and the retention of uploaded lists, are the sort that procurement teams tend to raise before a diagnostic becomes a line in a media contract.

Timeline

Summary

Who: Adelaide (Adelaide Metrics Inc.), an attention measurement company, with co-founder Zach Kubin promoting the tool. The calculator targets brands and agencies buying programmatic media in the US, UK and Canada.

What: A free, email-gated Media Quality Calculator that grades a single-column CSV of 200 to 100,000 domains or app bundle IDs against Adelaide's 0-100 AU benchmarks, returning a PDF with list-level averages, the share of entries below benchmark and a channel breakdown across Display, OLV and CTV. The promotion cites more than $27 billion a year in lower-quality programmatic inventory without naming a source.

When: The tool became available in the second half of August 2026, based on the relative timestamp of Kubin's LinkedIn post; Adelaide has not published an exact release date.

Where: Online through Adelaide's website, with coverage limited to the US, UK and Canada; lists from other regions may return limited results, according to the company.

Why: Adelaide argues that viewability and low CPMs disguise poor inventory and that its outcome-trained metric identifies it. The calculator offers a no-cost entry point to that scoring, extending AU beyond the Amazon DSP, DV360, Nielsen and Comscore integrations, while leaving open questions about benchmark transparency, channel inference, unscored domains and the retention of uploaded lists.