Monday note. This slot is normally an interview. The ADOTAT Show is dark until Tuesday, September 8, when Season 10 opens with Jared Belsky of Acadia. Until then you get me, unsupervised, with a standards document and a grudge. Oh this is bright.

June 2019. The IAB Tech Lab ships the Data Transparency Standard.

A nutrition label for audience data. Yummmm… Tasty Data.

It was a good idea. It is still a good idea. That is the tragedy.

The label tells you what you are actually renting. Observed or inferred. Declared or modeled. Whether a lookalike engine quietly tripled the size of your "auto intenders." How stale the qualifying event was allowed to get before somebody still counted it. Which ID types are in there, and how many. What the refresh cadence is. Whether the whole thing got cross-device expanded into a household guess and then sold to you as a person.

Everything a buyer would want to know before wiring the money.

Seven years later, the public compliance registry lists four companies.

That is it. That is the list. You could seat the entire certified universe of the open internet at one restaurant table and still have room for the person who booked it.

Not one is a DSP. Not one is a holding company.

Not one is an advertiser.

The entire class of buyer this instrument was built to protect does not appear anywhere on the roster of people using it.

Note the versions while you are here. Version 1.2 has been current since April 2024. Half the compliant population is still certified against 1.1. The room is not just empty, it is a year and a half behind.

The label went to public beta in October 2018 with ten participants: Alliant, LiveRamp, Oracle, Neustar, Dstillery, Epsilon, Hearst Magazines, Meredith, Relevanc and Pandora.

Count the survivors on today's registry. Three.

Oracle is not there, though Oracle's advertising business is not really anywhere. Neustar is not there. Dstillery is not there. Hearst is not there. Meredith is not there, and Meredith is not so much a company now as a brand portfolio in someone else's basement.

LiveRamp is not there.

That one absence matters more than the other six combined, and here is why.

LiveRamp is not a participant in the identity market. It is close to the plumbing of it. Onboarding, matching, the connective layer a very large share of third-party audience data crosses on its way to becoming addressable. If the company sitting at that junction had certified, the standard would have had a distribution mechanism. Every provider routing through it would have had a commercial reason to produce a label, and every buyer would have had one place to check.

Instead, the firm best positioned to make the standard ambient sat in the beta, watched it ship, and never appeared on the compliance list.

A standard adopted by the plumbing becomes infrastructure. A standard adopted by four data sellers becomes a plaque.

The same logic runs the other direction, and it gets worse. The industry's real answer to "how do I know what is in this audience" turned out to be the clean room, an architecture whose entire selling proposition is that nobody sees the underlying data. That is not a company declining to label a segment. That is a market relocating the segment somewhere a label cannot follow, and then billing the move as a privacy improvement.

Hold that thought. It is the actual answer to the question this series is chasing.

Anthony Katsur has not been coy about any of this, which is genuinely to his credit and simultaneously the most damning material in the file.

January 2025. Yahoo becomes the first DSP to adopt the labels. First. Nearly six years after final v1.0. Asked about adoption across DSPs, SSPs, data brokers and DMPs, the man who runs the Tech Lab called it a travesty.

A month before that, on the broader Seller-Defined Audiences framework the label plugs into, he had already conceded uptake was lagging after nearly three years in market, and wondered aloud whether the branding was the problem. Maybe we shouldn't have branded it "seller-defined," he wrote, then noted he was not kidding. The framework was subsequently renamed Curated Audiences, on the theory that the industry might adopt a thing it had ignored if the thing had a friendlier name.

Reader, it did not.

March 2026, still going: the standard had been in place about four years, adoption remained limited, broader adoption was needed.

When the head of the standards body has to keep announcing that his own standard is not being used, the news is not that the standard failed. The news is that the industry watched it fail in slow motion, on the record, for six years, and never once stopped putting the word "transparency" on the slide.

Here is what I could not find, and I looked hard. A published adoption target. Not a number of certified providers. Not a percentage of data sellers. Not a share of programmatic spend. Not a deadline after which labels become table stakes. The public language is all accelerate, support, broaden.

You cannot miss a target you never set. That is what not setting one is for.

There are three different things in this industry that get called adoption, and only one of them is public.

A company can implement the schema. A provider can upload segment metadata to the Transparency Center. A company can complete certification and appear on that registry.

I have hard public evidence for exactly the third. The registry is a roster of certified firms, not a directory of every implementation, and it publishes last-certified dates rather than original enrollment, so it cannot even tell you when this stalled.

Yahoo proves the gap. Yahoo publicly implemented the labels in January 2025 and does not appear on the compliance registry. One documented implementer, uncertified. If there is one, there are others.

So the fair version of the Tech Lab's answer is: your number counts audits, not usage, and usage is higher than four.

That answer is not wrong. It is also not a defense.

If a transparency program cannot tell you who is using it, that is a second transparency problem stacked on the first. Four is the only public number that exists. If the real figure is higher, the single organization on earth positioned to publish it has chosen not to.

This is the comparison that should end the argument.

Ads.txt, released May 2017 by the same standards body. By September, 12.8% of the top ten thousand ad-selling domains had a file. By November, 44%. By early 2018, one measurement put 51% of the top five thousand programmatic-volume sites on it.

Four months to double digits. Six months to roughly half the supply that mattered.

The Data Transparency Standard has had eighty-six months and four certified companies.

The difference is not virtue. Nobody in this business woke up in 2017 and got principled. The difference is enforcement.

Ads.txt handed the buy side a free, machine-readable authorization check they could drop straight into a filter. A publisher either authorized you or did not, the answer was a text file, and a seller who skipped it lost money that same afternoon. No committee. No seal. No sales call. A rule with teeth wired directly to a budget.

Money is the only compliance department this industry has ever respected.

Watch what happens as the enforcement pressure drops off. App-ads.txt, March 2019, same idea ported to apps. Adoption among top-ranked apps runs in the fifties. Across the whole Google Play catalog it sits near 24.8%. Same standard, same body, and the number turns out to be a readout of how much each publisher had to lose.

Sellers.json, also 2019. One exchange announced adoption near a hundred percent by that December, which is a vendor describing vendors. Independent looks later found about 22% of several thousand exchanges hosting a valid file, and a 2026 European scraping effort pulled usable files roughly 61% of the time.

The pattern is not subtle. The standards this industry adopts are the ones that block a transaction. The standards it praises are the ones that don't.

The Data Transparency Standard is a disclosure regime. Nothing in the bidstream refuses a trade because a label is missing. Nothing in the P&L notices. The label repository was never a buying venue, never carried pricing, and metadata is retrieved out of band, which is a technical way of saying it sits somewhere else while the auction happens without it.

A standard nobody can enforce is a suggestion with a logo.

Tech Lab's public pricing page lists fixed pricing for OM SDK compliance. Fixed pricing for podcast compliance. For Data Transparency, members and non-members alike, it says: pricing determined based on scope of work.

Read that again with the subject matter in mind.

A transparency standard with an undisclosed price.

I want to be fair. Scoped pricing is normal for audit work, effort genuinely varies with the size and mess of a data business, and the program's own archived materials describe providers prepaying against the volume of segments they had in market. There is a defensible reason for it.

There is also a consequence. A published number is something a mid-sized data provider drops into next year's budget. A scoped quote is something they have to call sales about. One of those produces adoption. The other produces a pipeline.

And it is not quick. Tech Lab describes engagements generally running two to five months before a seal. Earlier program documents describe a three to five month audit involving the IAB and a third party, with random label audits through the year and no seal if the third-party audit fails.

So the ask is: several months, an unpublished sum, an annual renewal, for a credential no buyer has ever publicly required anyone to hold.

I went looking for one documented instance of a DSP making label compliance a mandatory bid-request field, a holding company writing it into an RFP, or an advertiser making it a contractual condition of buying curated inventory.

I found none.

That is a negative finding and I will flag it as one. Programmatic contract terms are frequently confidential and a private clause could exist. But no public procurement document, no agency RFP template, no advertiser contract language, no bidstream mandate. Even Yahoo's move was an implementation announcement, not a no label, no buy rule.

Every buyer in this industry says they want provenance. Not one has ever made it a term.

The entity monetizing the segment fills in its own fields. Tech Lab's documentation says so, and then, in the same breath, acknowledges the economic incentive a seller has to misrepresent a cohort's attributes or drape a more attractive classification over it.

The document identifies the fraud incentive in the same file where it proposes the fix, and then leaves the fix voluntary.

The certification audit does add something real, and I want to be precise about what. It assesses whether label information is reliable, whether the organization has the systems and staff to produce labels consistently at scale, and whether it can produce them for all in-market segments.

That is a process audit. It tests whether a company can manufacture labels correctly and at volume.

It does not test whether the people in the segment are the people on the label. Nobody knocks on a door.

How we reported this: built from IAB Tech Lab's own primary documents, Katsur's on-record remarks between December 2024 and March 2026, and peer-reviewed research on segment accuracy. The registry records certification, not implementation, so companies may use the schema without appearing on it, as Yahoo does; no public count of implementers exists. Adoption rates for ads.txt and its siblings vary by denominator and are cited to their original samples, and curated-market size figures are vendor-supplied. That no buyer has publicly required label compliance is a negative finding from public documents. The LiveRamp point is my analysis, not a claim about any decision the company made. ADOTAT has no financial relationship with any company named here, no subject received advance review, and Anthony Katsur has previously appeared on The ADOTAT Show.