There was no show this week, and the reason is older than programmatic, older than the internet, older than the entire American advertising business by roughly two thousand years.

It was Tisha B'Av yesterday. The Ninth of Av. A full fast, sundown to nightfall, no food and no water, sitting low to the ground reading Lamentations by candlelight while the rest of the world went about its Thursday and someone in Slack asked if I was "around for a quick sync." I was not around. I was on the floor, mourning the destruction of two Temples and a long grim parade of other catastrophes the Jewish calendar has filed under the same date, because the filing cabinet only has so many drawers.

My friend Kym Frank at Fox posted a Voronoi map this month showing where Americans are taking GLP-1 drugs, and it is the kind of chart you look at twice and then look away from. West Virginia, roughly one in four residents on an insurance-covered prescription for something like Ozempic or Wegovy in 2024. Hawaii, five percent. The heat clusters through the South and Appalachia, where obesity and diabetes already run highest. And the real numbers are higher, because the data comes from insurance claims and therefore misses everyone paying cash, everyone using a compounding pharmacy, and everyone uninsured. So: a national map of a phenomenon we are measuring with one eye closed.

Here is the part nobody wants set in type. In roughly 95 out of 100 cases of obesity, there is no identifiable disease driving it. No thyroid disorder. No Cushing's. No single broken gene. No weight-gaining prescription. The remaining sliver looks about like this: monogenic and syndromic causes (MC4R, leptin receptor, Prader-Willi) at maybe two to five percent of severe obesity and less across the general population; endocrine disease at one percent or less; medication-induced gain from antipsychotics, steroids, some antidepressants and insulin, real but slippery, call it a few percent. Add it up and you have a rounding error carrying an enormous amount of narrative weight.

Which is the ADOTAT thesis and always has been. The truth is not hidden. It is just commercially inconvenient.

A country can know precisely what is happening to it and still find an injection easier than a system change. An industry can know precisely which vendors are stealing and still find the renewal easier than the complaint.

Here is what I keep circling back to. I owned two ad networks. Sold both. One in 2007, one in 2017. One programmatic, one aggressively not. One was a comScore top 20 property, the other was not rated at all. Both would have handed full access to anyone who asked, log level, placement level, whatever you wanted, at any hour, because there was nothing in there to find. And both made good money. Not generational money. Not "we invented a new adjusted metric to explain this" money. Good money, made by the radical strategy of not committing crimes.

That is the whole argument. You can run a profitable, honest business in this industry. Thousands of people do it every day and nobody writes them up. The problem is that the market does not want profitable and honest. The market, and the public market especially, wants growth, forever, at a rate no legitimate business sustains. Which means somewhere down the org chart, several layers below anyone who will ever testify, a person is quietly told to go find the difference. That person is the reason we have MFA sites, phantom inventory, undisclosed fees, resold impressions and a supply chain laid out like a corn maze designed by a tax attorney.

Nobody wakes up planning to defraud advertisers. They wake up with a number.

So, no show. But there is news, and the news is the same story wearing eleven different suits.

And now, the news.

Check My Ads got the agreement, got its spending minimum waived, and was one signature from a DSP seat before the platform decided its supply partners required "protection." Protection from what, precisely? An advertiser running a campaign and then looking at what she bought? The DSP would not say, which is generally the moment you notice a smell coming from the kitchen and the chef insisting everything is fine.

Platforms have every right to vet buyers. Absolutely. But this was not an anonymous shell shop moving counterfeit supplements out of a mailbox in Limassol. It was a well-known critic whose intentions were discussed openly before the paperwork was drawn up. The industry adores transparency the way it adores a charity gala: enthusiastically, annually, and from a safe distance. This DSP loved transparency right up until transparency showed up with login credentials and a free afternoon.

OpenAI's Yelp licensing deal hands ChatGPT reviews, photos and local business data, with quote requests that can turn an answer into a transaction. Genuinely useful for users. Genuinely valuable for Yelp. And spectacularly convenient for an AI company quietly collecting every component of an advertising business while maintaining, with a straight face, that it does not have one. So the crowning achievement of machine intelligence is helping you find a plumber who calls back. Fine. Honestly, that may be worth the data centers.

OpenAI does not run a conventional ad network today. But look at the parts inventory: local intent, commercial queries, business listings, conversion tools, self-serve buying, CPC bidding, a Conversions API, and job listings mentioning inventory setup and yield issues. Those are not scattered puzzle pieces that happened to end up in the same drawer. That is an ad network wearing a lab coat and hoping nobody checks the pockets.

Alphabet posted another quarter of double-digit growth, its twelfth straight, with Search, YouTube and Cloud all up and Gemini usage climbing. Reports of Google's death continue to be enormously profitable for Google. The company also intends to spend roughly the GDP of a mid-sized nation on infrastructure, because nothing signals capital discipline like building data centers fast enough for the electrical grid to retain counsel.

The interesting question was never whether Google can print money. It can. It does. It will. The question is how long publishers, platforms and users keep shoveling coal into a furnace whose exhaust is AI summaries that answer the question and keep the click. Google is flourishing financially while steadily degrading the ecosystem that supplies its raw material. People keep calling that a contradiction. It is not a contradiction. It is the business model, functioning exactly as designed.

Reddit's data arrangement with Google is reportedly wobbling, with the platform weighing tighter limits on how its posts and comments feed AI products. This matters more than the dollar figure suggests, because Reddit has become one of Google's most dependable sources of information that sounds like it came from a human being, on a web increasingly filled with machine-written oatmeal. Google loves community content nearly as much as it loves not having to make any.

The standoff exposes the arithmetic underneath AI search, which has never quite closed. Platforms need fresh human material. Communities and publishers need traffic or licensing money. Users want answers for free. Everybody wants to get paid, ideally by somebody else, ideally next quarter. Eventually the parties supplying the actual useful substance stop accepting exposure as legal tender, particularly once the exposure no longer even includes a click.

Major publishers are reportedly reassessing their relationship with Google as AI results resolve queries without ever sending anybody onward. The traffic firehose has narrowed to a cocktail straw. This gets described as a product improvement, which is a lovely way of saying Google keeps the audience and the publisher receives a citation-shaped pat on the head.

Publishers built search, in the sense that they produced the material people were searching for. The current proposal is that they keep producing it while the search engine converts it into a finished answer and the reader never arrives. As arrangements go it is elegant, efficient, and works beautifully for everyone except the people writing checks to reporters, editors, photographers and lawyers.

Pinterest's Inclusion Fund returns for a sixth year, backing creators who use its features for skin tone ranges, hair pattern search and body type. The company reports that people using these tools save substantially more Pins, which is a data-backed way of saying that treating users like actual human beings turns out to be good for engagement. Inclusion with a dashboard attached. I can live with that.

Some executive will inevitably describe this from a stage as purpose-driven innovation, with a slide, possibly with a moon landing metaphor. The unglamorous version is better and truer: Pinterest built search that helps people find things that look like them, creators who use it get discovered, and the platform makes money when the product stops being annoying. Inclusion does not have to be charity when the product team is competent.

TiVo Ads and OpenGlass are opening home-screen inventory to programmatic buyers, monetizing the fragile moment between sitting down and choosing something. Some formats cover nearly the whole screen. Congratulations: your television now has a lobby, and the lobby has been leased.

Media buyers will call it a high-attention environment. Households will call it the reason the remote is now inside the drywall. Both are right. The commercial logic is unarguable, though: the home screen is scarce, unavoidable, and sits directly upstream of every content decision anyone makes all night. In an industry starving for premium inventory, we have arrived at the moment where the most premium content in television is the menu.

Jaysen Gillespie in Adexchanger makes the point most transparency advocates skip: log-level data, fee disclosure and supply-path visibility explain what happened. They do not guarantee that the buying system was ever pointed at incremental business value. Which is a useful distinction in a business where "transparency" frequently means being handed a much larger spreadsheet some weeks after the money left.

You can see every auction, every intermediary, every fee, in beautiful detail, while paying an algorithm handsomely to chase people who were going to buy anyway. Perfect visibility into a misaligned system is not control. It is high-resolution security footage of your budget being mugged, timestamped, exportable, and available in CSV.

The industry's standing answer to complexity is to consolidate spend into fewer preferred platforms and strip out intermediaries. This genuinely does simplify contracts, reporting and the number of people you have to be polite to at Cannes. What it does not do is align anyone's incentives with yours, which was the actual problem you were trying to solve.

A short supply chain optimizes toward cheap conversions, flattering attribution and demand capture just as cheerfully as a long one. It simply does it faster and with better UI. Removing three middlemen is real progress. Replacing them with one enormous middleman who owns the inventory, the auction, the measurement and the report card is not reform. It is fewer vendors reaching the same bad decisions more efficiently.

Most ad systems are extremely good at locating people with strong purchase intent, for the excellent reason that predicting an imminent conversion is vastly easier than causing a new one. The result is performance marketing that looks brilliant in the deck while spending its budget on consumers already standing at checkout with a card out.

This is not artificial intelligence so much as artificial opportunism. The model watches a man carry a pizza to the table, files a report, and announces that it drove dinner. Everyone applauds. Nobody asks the pizza.

Beehiiv is layering programmatic advertising across its newsletter network, giving publishers another revenue line beside direct sales and subscriptions, and sparing creators the daily indignity of emailing brand managers who will not respond until the third follow-up. Initial units are text-and-image formats styled to match each publication's design, with publisher floors, CPC or CPM buying, and contextual controls. Done well, that produces sponsorship-quality creative at scale. Done badly, it produces ads that resemble editorial closely enough to send every media ethicist reaching for the antacids, and native advertising's oldest bargain reasserts itself: make the ad look less like an ad, then act astonished when readers feel handled.

The irony deserves a moment of silence. Search and social spent a decade convincing publishers that email was ancient infrastructure for people who still print directions. Now AI is eating referral traffic, the algorithms have gone feral, and the inbox is suddenly one of the last places on earth where a publisher can identify its own audience by name.

Google's Meridian and Meta's Robyn revived marketing mix modeling right as privacy restrictions were strangling user-level attribution, and open-source code gives marketers something they can inspect, modify and describe in board meetings as independent. Google sales teams also reportedly carry performance targets tied to Meridian adoption, which makes the ensuing organic groundswell feel slightly less spontaneous. The code is open. The incentives remain fully enclosed.

None of that makes MMM bad. MMM is a substantial improvement over pretending a last click explains a purchase. But the models draw enormous power from platform data, platform integrations and platform assumptions, and Google is not merely participating in the measurement ecosystem, it is deciding where everybody starts counting. A model can be perfectly transparent and still reflect the worldview of whoever wrote it. A casino can publish the rules of blackjack. The house edge does not evaporate because the brochure is on GitHub.

DV360 buyers will be able to use Adelaide's Attention Unit inside custom bidding, optimizing straight toward predicted attention or folding it in as a multiplier. Useful, and worth stating plainly: the score is a prediction, assembled from placement, clutter, size and time in view. Nobody is watching your pupils through the television. That is precisely why it scales to billions of impressions at bid time, and precisely why buyers must resist upgrading "likely to receive attention" into "a human being watched and was moved." Ad tech's favorite trick remains turning an estimate into a noun and the noun into an invoice.

The genuinely interesting wrinkle is that Google and Amazon took opposite roads with the same vendor. Amazon DSP uses Adelaide segments to decide what is even eligible before the auction. Google adjusts what an impression is worth inside the auction. Exclusion says bad attention is unacceptable. Bid-time pricing says bad attention is a negotiation. That is a real strategic fork rather than a feature comparison: is low-quality media waste at any price, or can sufficiently cheap garbage become efficient garbage? Somewhere a holding company has already booked the forty-eight-slide meeting.

Attention vendors use different methodologies, inputs and scales, so one provider's score cannot be honestly compared to another's, which will not stop a single person from selling it as a currency. Viewability walked this identical path: useful concept, necessary measurement, years of methodological trench warfare, and enough acronyms to stun livestock. The consolation prize is real, though. MFA sites run on clutter, weak placements and competing units, all of which crush predicted attention, so optimizing toward attention defunds junk supply without anyone hand-labeling ten thousand domains. Not foolproof. A well-designed garbage site is still garbage. But anything that makes ad-stuffed content farms less profitable earns applause.

Notice where all of it lands, though. Google has spent years retiring manual bidding controls while expanding custom bidding, APIs and automation, and Adelaide slots into that pattern like it was measured for the space. Buyers get to express more sophisticated goals through fewer actual levers, inside a system where Google still runs the auction. That is the modern platform bargain in one line. The platform calls it simplification. Buyers call it automation. Finance calls it whatever the dashboard says worked.

Read all of that as one document and the pattern falls out like change from a coat pocket.

A watchdog gets an agreement and then gets uninvited, and nobody will explain why. An AI company assembles every organ of an ad network while insisting it has no ad network. Google writes the measurement code, incentivizes its salespeople to distribute the measurement code, and is then graded by the measurement code. Attention vendors sell predictions dressed as observations. Beehiiv promises the inbox will be different this time, and every single person who lived through 2011 already knows which meeting comes next. Publishers keep feeding a machine that stopped feeding them somewhere around 2023.

None of that is fraud. That is the part I want you to carry out of here. Almost none of the damage in this business is done by criminals. It is done by reasonable people making locally reasonable decisions inside a system that rewards opacity slightly more than honesty, quarter after quarter, year after year, until the accumulated slightly becomes load-bearing. Nobody torched the Temple from the inside. They just stopped being able to hear each other, and the walls came down on schedule.

Which brings me back to the map, and the injections, and the ninety-five percent. The hardest thing about the truth is almost never that it is hidden. It is that it is sitting in plain view, fully documented, thoroughly inconvenient, and considerably easier to medicate than to address. America knows what is making it sick. This industry knows what is making it sick. In both cases there is an extremely large business built directly on top of not saying so, staffed by decent people whose mortgages depend on the silence.

I am not asking anybody to be a hero. Heroes get blacklisted, I hear. I am asking for something smaller and much harder:

Do that and the machine still runs. It just runs with the lights on.

The show returns next week. I have eaten. I am fine. The fast does its work whether or not you enjoy it, which is more than I can say for most of the ad tech I cover.

Somewhere in ad tech there is a demand-side platform that spent April flirting, May proposing, and June ghosting, and the only thing that changed in between was the dawning realization that the client actually intended to read the reports.

Per AdExchanger's Anthony Vargas, who reviewed the email chain, an unnamed DSP was so thrilled to land Check My Ads as a customer that it waived its minimum spending requirement. Waived it. In this economy.

Both parties cosigned a master service agreement in mid-May. The recap email after the April demo was not a legal document so much as a mash note: it cheerfully summarized the watchdog's goal of moving past Google's black boxes into what it called real programmatic with full transparency and control, then signed off with an actual cheer about how honored the DSP was to be considered.

Woot woot, indeed.

Then, before a single impression could be served, the DSP developed scruples. It informed Check My Ads that it could not get comfortable with, and I am quoting the email here because no paraphrase does this justice, "the scope of protections extended to our vendor and supply partners."

Protection from what, precisely? From an advertiser running a campaign and then looking at what she bought? The DSP would not say. That is generally the moment you notice a smell coming from the kitchen and the chef insisting everything is fine.

Rejections are boring. Vetting is legitimate. Any platform can decline any buyer for any reason, and no one is owed a seat.

But look at the sequence, because the sequence is the story.

On May 20, the rep tells Check My Ads the seat should go live the following week, and by the way, could they share a little more about the planned test campaigns, brands, budgets, objectives. Fine. Normal. The reply comes back: a brand lift campaign, and a cost-per-acquisition campaign to drive newsletter signups, with a test budget of one thousand dollars. One thousand. Not a stress test of the global supply chain. A merch-and-newsletter budget that would not cover a mid-tier sponsor's bar tab at Cannes.

Then on June 3, after silence, the rep asks the question that gives the whole game away: if Check My Ads spots any vendor performance concerns on the platform, could they agree to loop the DSP in first before escalating externally.

Read that again. That is not a compliance question. That is not a creditworthiness question. That is a media-relations question wearing a lanyard. It is the ad tech equivalent of a landlord asking whether you plan to call the health department or just mention the mold to him privately.

Check My Ads said yes, obviously, because that is already how they work. Two days later the head of client partnerships wrote to say the partnership was off.

The kill shot came with a small, very human coda. The client partnerships lead said the decision was made above their level, driven by supply and legal, and that the objection had first surfaced internally on May 20. May 20 being the same day the rep was cheerfully promising the seat would go live next week.

So somewhere in that building, on that Wednesday, two teams were operating on entirely different scripts. Sales was setting the table. Legal and supply were quietly moving the reservation to a restaurant that does not exist.

And here is the part that should embarrass everyone: according to Check My Ads COO Arielle Garcia, the DSP never ran a financial audit of the organization. The stated concern was never about solvency, never about a policy breach, never about fraud risk from the buyer. It was about downstream impact on partners, and about activity that might intersect with those partners in ways the contract did not cover.

Intersect. What a word. Ships intersect with icebergs.

An anonymous ad tech consultant offered AdExchanger the industry's honest defense, which is that no DSP is going to knowingly hand a seat to someone whose stated purpose is to turn it into an exposé. That is true. It is also a confession. If your supply path can survive a small-business advertiser with a thousand-dollar budget, you have nothing to protect. If it cannot, "protection" is doing a great deal of work in that sentence.

Check My Ads reportedly ran into the same wall twice more.

A second DSP invited Iesha White, the organization's director of intelligence and a former media buyer of ten years, onto a podcast. She recorded it. Content is content. She then asked whether she could get access to the platform to test it. The answer was no. Congratulations, you have been booked as a guest and blocked as a customer, which is a novel way to say we love your voice and fear your eyes.

Then there is Thrad, the startup placing ads inside generative AI responses. Thrad approved Check My Ads, then flagged problems: the campaign had selected sensitive contextual topics where it cannot place ads, including politics, abortion, mental health and LGBTQ, and the merch storefront was not set up to ship to Europe while the campaign targeted European audiences. Both fair catches. Both fixed. Thrad still would not launch. CEO Andrea Tortella later told Garcia that Check My Ads does not fit the ideal customer profile. Thrad did not respond to AdExchanger's requests for comment.

The ideal customer profile. In an industry that has spent fifteen years bidding on anyone with a credit card and a domain, we have finally located the buyer too risky to serve, and she is a nonprofit that wants to sell t-shirts and write a methodology post.

The industry adores transparency the way it adores a charity gala: enthusiastically, annually, and from a safe distance. Vendors have spent years putting Check My Ads in their decks as proof of principle, a logo of accountability, a line item in the trust narrative. That is transparency as merchandising, and it costs nothing.

This DSP loved transparency right up until transparency showed up with login credentials and a free afternoon.

The uncomfortable inference is not that anyone is running a conspiracy. It is duller and worse than that. DSPs are optimized for their own margin, not for your visibility, and the parts of the stack most carefully curtained off are the parts that would make the economics easy to argue with: the fee take, the supply path decisions, the resold and re-resold inventory, the placement-level outcomes that never quite match the dashboard. Complexity gets sold as sophistication. It also functions as camouflage.

The right investigative question is never "are they lying." It is: what becomes visible if they fully disclose the logs, the fees, the identity paths, and where the impressions actually ran?

Somebody at that DSP already knows the answer. That is why the seat never went live.

Garcia's closing position, for whatever it is worth, was an open invitation: any DSP confident in its supply and interested in making programmatic accessible to small businesses should get in touch.

lol.

The line is going to be short. Watch it and see.

The full ADOTAT Scorecards. Vendor comparisons that name winners and losers, with the analysis your planning team needs before Q3 budget meetings.

The takes nobody else will publish. Independent means we don't answer to vendors, agencies, or the trade press establishment.

The archive. Every Scorecard, every analysis, searchable. The reference layer for how this industry actually works.

Read by senior leaders at every major holding company agency, GroupM, Publicis, Omnicom, IPG, Havas, Dentsu, and inside the buying and strategy teams at Netflix, Roku, Pinterest, Disney, NBCUniversal, Spotify, Amazon, Meta, and most of the platforms shaping where ad dollars actually go.

If your competitors are reading ADOTAT+ and you aren't, you're planning with less information than they are.

Expense it.

Founding member rate. Locks in for life.

Also, no ads. Zero ads. No Ads.