IndustryPace
Ad industry commentary from Pesach Lattin. Since 2010.

Pace Notes

Wednesday, November 18, 2026

The ad business, read one turn ahead.

Four ADOTAT stories and one from outside, and most of them are about who grades whom. The Trade Desk calls thirteen departures "several," Netflix trims $300 million off its forecast, and ADOTAT corrects its own math on Nielsen. The Pick is The Trade Desk.

From ADOTATThe Trade Desk Meltdown: “Several” Leaders Moved On. The Count Is Thirteen.

The Trade Desk Says “Several” Leaders Left. The Count Is Thirteen.Pick

By ,

On Monday, Nov. 17, The Trade Desk closed at $14.93, four cents under the strike on the 7 million options its board gave Jeff Green in September. After the bell, Adweek reported three more senior people out: Waseem Basheer, SVP of engineering, gone that day, with Whitney Rifkin of client service and Bryce Poll of commercial strategy leaving Friday. The company's statement called them "several long-tenured leaders." ADOTAT kept the ledger. By its count it is thirteen senior exits since early 2025, with the CFO seat turning over four times in about fourteen months. Several is a word. Thirteen is a list.

Engineering is where the pattern is cleanest. Aravind Chandrasekharan left the chair in late August after twelve years. Basheer took it. On Nov. 3 co-founder Dave Pickles came back to run a new CTO Council, and two weeks later the head of engineering walked. The CTO job is still open. Rifkin and Poll matter for a different reason: they are the people a holding company calls about the 2027 contract, and their replacements will be learning the accounts while the accounts are learning the new price.

Because there is a new price, or at least a new posture. On the Nov. 9 call Green said that if the company could lower its take rate and make up more growth, it was open to that. For ten years the fee was the fee and transparency was the product. Now the fee is a negotiation and an "open measurement" product, due before year-end, is the argument for paying it. Meanwhile the board struck Green's grant at $14.97 and repriced roughly 15.6 million employee options to $12.26, with a snap-back if holders leave before April 19, 2027. Three senior people left five months early anyway.

Why it matters: ADOTAT's read is that the people who would explain any of this have stopped talking. None of the thirteen has said publicly why they left. OpenAI handed ChatGPT inventory to Magnite and PubMatic, not the DSP that courted it. A named defendant in the securities case, Samantha Jacobson, sits on the board while working at OpenAI. ADOTAT's advice to buyers is blunt: get the all-in rate, the switched-on toggles and a key-person clause in writing this month, while the account team still needs the business. Then ask the company to say how many.

The full story is at ADOTAT

Expanded from ADOTAT's Nov. 18 Wednesday investigation, published by The Adotat Staff; the departures ledger and the top sections are free, and a buyer's playbook, rescore and a 44-page Third Edition report sit behind the ADOTAT+ paywall. Per ADOTAT, its sources were The Trade Desk's 8-Ks of Aug. 6, Sept. 4, Sept. 14 and Nov. 9, 2026, its 10-Qs for Q2 and Q3, the Sept. 18 proxy and Oct. 19 special-meeting results, the Q3 call transcript, the SEC litigation release and the Nov. 9 guilty plea in United States v. Mitchell, the Central District of California docket, Publicis's June 12 joint statement as reported by Ad Age and Digiday, PubMatic's and OpenAI's announcements, Madison and Wall's Nov. 14 recap and Next in Media, and Adweek's Nov. 17 report by Trishla Ostwal. The count of thirteen is ADOTAT's own tally of senior executive exits since early 2025 and excludes two departed directors. ADOTAT notes its January and July editions were wrong on Jacobson's employment, the Publicis timing and the departure count, and that its revenue estimates were too generous three times running. ADOTAT's publisher wrote the earlier forecasts it scores here. The Trade Desk, Magnite, PubMatic and Publicis are public companies; this is commentary, not investment advice.

From ADOTATNetflix Opened Four Doors to Its Ads. Its Forecast Fell $300 Million.

Netflix Let Four DSPs In. Its Ad Forecast Fell $300 Million.

By ,

Netflix started its ad business renting Microsoft's sales machinery, decided in January 2024 to build its own, and says the Netflix Ads Suite was live in all 12 ad markets by November 2025. By this fall it listed four buying doors: Amazon, Google's DV360, The Trade Desk and Yahoo, with Magnite on supply. Ad chief Amy Reinhard promised to meet advertisers "regardless of what DSP" they use. The doors are open. The rooms behind them are still being furnished.

Then the Oct. 20 shareholder letter cut the 2026 ad forecast from about $3 billion to $2.7 billion, blaming a slower rollout of ad product capabilities. Co-CEO Greg Peters also pointed at partner integrations. ADOTAT's point is precise: this is a revision, not a revenue decline, and the ad business is still growing fast. But S&P Global had Q2 ad revenue at $618 million, 7.2% under consensus, so the spreadsheets were running ahead of the business before anyone wrote the letter.

The fine print is where buyers live. Amazon's own documentation says its audiences work on Netflix in private marketplace run-of-network deals only, that existing Netflix deals are not compatible with audience campaigns, and that stacking extra audiences can stop delivery. The Trade Desk now lets buyers target Netflix with their own data in Kokai, but measurement is "expected to follow," with no date. And the MRC accreditation from Aug. 4 covers U.S. impression counts, not targeting, reach, viewability or attribution. A counted impression is not a proven audience.

Why it matters: the question Netflix has not answered is whether the missing $300 million was signed money that never ran, cheaper impressions, spend that moved to rivals, or dollars slipping into 2027. Netflix has not broken it down. It named ex-Google executive Darshan Kantak to run the ads platform on Nov. 2, which gives the tools an owner. Buyers told Variety they are still waiting on plans and pricing, especially for sports. An owner is not a calendar, and engineering cannot send a proposal.

The full story is at ADOTAT

Expanded from ADOTAT's Nov. 18 report by Pesach Lattin. Per ADOTAT, its sources were Netflix's Oct. 20, 2026 shareholder letter and later disclosures, Amazon's buyer documentation for Netflix, The Trade Desk's Nov. 9 release, the MRC's Aug. 4 accreditation letter, S&P Global's Q2 analysis, Tinuiti's July overview, Jack Myers's Oct. 21 analysis, Texas and Florida filings, and reporting from Digiday, Variety, The Current and Reuters. Netflix's programmatic share is not public and the figures for it are secondhand estimates; CPM figures come from different dates and deal types and are not a rate card. Anonymous buyer comments in the original are background only. The Texas and Florida suits are allegations. ADOTAT notes that The Current, source of the Ovative buyer quote, is a Trade Desk publication. Netflix received written questions on Nov. 16 and had not answered by publication; ADOTAT says it has not confirmed any Netflix ad executive's departure, including the reporting on Nicolle Pangis. ADOTAT is reader-supported. Netflix, Amazon, Alphabet and The Trade Desk are public companies; this is commentary, not investment advice.

From ADOTATKelly Abcarian Is Now CEO of Star Media Enterprises

Star Media Runs TV's Ratings Plumbing. Its CEO News Came in a Job Post.

By ,

Kelly Abcarian, who built products at Nielsen and then spent two years at NBCUniversal telling the industry its measurement was out of date, is the new CEO of Star Media Enterprises. She confirmed it to ADOTAT. The industry found out through a LinkedIn hiring post listing three jobs. Neither Star Media nor its owner, Banyan Software, has issued a release naming her, and the last official word named someone else as president.

Star Media is a 1988 company in Dunedin, Florida, with a LinkedIn headcount of 11 to 50, and it does the unglamorous middle job in TV: it takes daily data from Nielsen, Comscore, VideoAmp and iSpot and turns it into estimates, tracking and post-buy reports networks can bill on. A Banyan job listing says its StarTrak system is used by most ad-supported networks that get national overnight ratings. That is the owner describing its own company, not an audited share. Estimate, track, post: the whole life of a TV deal runs through software almost nobody has heard of.

Banyan, an Atlanta holding company with more than 120 software businesses, bought Star Media on June 12 for an undisclosed price. It already owns broadcast traffic and workflow tools like vCreative and Mr. Master, plus identity and attribution firm Launch Labs. It advertised the CEO job at $240,000 to $275,000. Abcarian's post now seeks a New York chief revenue officer to sell to media companies and agencies. Nobody hires that person to keep a company the same size.

Why it matters: Abcarian argued for a world of multiple currencies. In that world the company that ingests, checks and reconciles all of them is the one place the numbers sit side by side. The more measurement fragments, the more the translation layer is worth, and she now runs it. Open questions remain: when her role took effect, what founder Jeff Stern still holds, and what Banyan wants Star Media to sell next. In a four-currency market, the discrepancies are the product.

The full story is at ADOTAT

Expanded from ADOTAT's Nov. 17 report by Pesach Lattin. Abcarian confirmed her role to ADOTAT. Per ADOTAT, its other sources were Banyan Software's June 12, 2026 acquisition announcement and job listings, Star Media's website and LinkedIn pages, Abcarian's LinkedIn post, a June 19 InfoTechLead roundup, PitchBook, a Better Business Bureau profile, coverage from Marketing Brew, Ad Age, AdExchanger, Digiday, MediaPost, Variety, Multichannel News, Cablefax and LBB, Advertising Week and MediaVillage biographies, and a 2025 TVision report. The claim about StarTrak's reach is Banyan's own description. Other executives' titles come from LinkedIn and are self-reported; when Abcarian's appointment took effect, the founder's current role and the deal price are not established. Star Media was sent questions and had not answered by publication; Banyan was not contacted before publication. ADOTAT is reader-supported with no paid placement in editorial.

From ADOTATUnmeasured: We Said Nielsen's Math Hadn't Caught Up. It Was Ours.

ADOTAT Said Nielsen's Math Was Off. It Read a Plus as a Minus.

By ,

On Nov. 17, ADOTAT's Unmeasured told readers Nielsen's holiday sales lift for cosmetics on connected TV was minus 22.9%, set it against Nielsen's 3.2x headline, and built a section on the contradiction. The number was plus 22.9%. There is no negative figure anywhere in Nielsen's data. A stray dash in Nielsen's written answers got read as a minus on one row and as punctuation on the next. Same dash, two readings.

The worse part, which ADOTAT printed itself: Nielsen answered at 1:06 PM Eastern and said plainly the lift was positive. Unmeasured published at 3:52 PM. The correction was in the inbox when the piece went out. Nielsen wrote again that evening and apologized for the dash. ADOTAT's reply is the right one: a stray dash is a typo, publishing after being told is ours.

Do the division the original skipped. Cosmetics went from 7.2% to 22.9%, which is 3.2 times. Department stores, 11.2% to 27.3%, about 2.4. Restaurants, 6.9% to 10.6%, about 1.5. Those are Nielsen's headline multiples. The second complaint fell too: Nielsen had explained that its lift is a modeled value on top of observed sales, and that its Predictive Sales Lift model was not used for these figures. One answer in two vocabularies, not a conflict.

Why it matters: a newsletter whose whole job is asking measurement companies to show their work has to show its own, at the same volume, and this one did. The web article is corrected with dated notes; the email cannot be recalled, so the correction went to the same inboxes. The questions left standing are ordinary ones about sample sizes and method. The check runs both ways, or it is not a check.

The full story is at ADOTAT

Expanded from ADOTAT's Nov. 18 Unmeasured correction by Pesach Lattin. Per ADOTAT, it draws on the Nov. 17 article as emailed, the corrected web version, and written correspondence with a Nielsen spokesperson on Nov. 16 and 17, given on background and attributable to a Nielsen spokesperson. ADOTAT has not seen the underlying sales lift studies; the lift figures are Nielsen's and the division is ADOTAT's, rechecked here. Nielsen did not review the correction before it was sent. The original piece also quoted Frank Kowalczyk, whom ADOTAT identifies as a fictional commentator; no line from him is used here. Unmeasured is reader-supported with no paid placement in editorial.

ElsewherePPC Land: Trade Desk picks Attain, with 15 million consumers, to anchor ad measurement

The Trade Desk Promises Measurement on Real Sales. No Method Yet.

By ,

On Nov. 17, the same day three more senior people left, The Trade Desk named Attain the first data partner for Open Measurement, the product Jeff Green promised on the Nov. 9 call for later this quarter. PPC Land's Luis Rijo read the release closely. Attain says it reaches about 15 million U.S. consumers through its own financial wellness and shopping rewards apps, which it does not name. Two weeks earlier, announcing an Amazon deal, it said over 14 million. Nobody explained the change.

The pitch, from chief commercial officer Kristi Argyilan, is that the product rests on purchases, "not self-graded homework." Fair line. But per PPC Land, the release gives no methodology, no price, no channel list, no exact launch date, no word on how ad exposures get matched to Attain's users, whether Attain is exclusive, or whether anyone outside audits the results. An earlier Trade Desk retail post claimed 90% of Attain-measured campaigns beat benchmarks, with no brands, period or control group named. Homework graded by a partner is still homework until somebody shows the rubric.

Context matters here. Attain was also one of Amazon DSP's early offline sales lift partners, and Amazon offers dozens of third-party measurement products. So the data source that is supposed to set The Trade Desk apart is one its biggest rival already buys. And the product arrives in the quarter the company's revenue fell for the first time since it listed in 2016.

What ADOTAT has already written: its Nov. 18 investigation argued that once Green called the take rate negotiable, Open Measurement became the argument for paying the fee, and told buyers to demand the spec before the pitch and to make sure outputs export to their own auditors at no extra charge. ADOTAT's Netflix report shows the same gap from the other side: Trade Desk targeting on Netflix is live, measurement is "expected to follow." The test is the first campaign it grades, and who else gets to see the answer key.

Read the original at PPC Land

Commentary on Luis Rijo's Nov. 17, 2026 PPC Land article, which is built on the Attain and Trade Desk release datelined New York, Nov. 17, and on PPC Land's earlier coverage; the release itself was not read separately here. The 15 million and 14 million consumer figures are Attain's own claims, not verified. The 90% benchmark claim comes from a Trade Desk post that PPC Land notes named no brands, period or control group. The list of what the release leaves out is PPC Land's reading; neither company is reported to have commented beyond the release. The point that Attain also works with Amazon DSP comes from PPC Land's background. The ADOTAT references in the last paragraph come from the linked ADOTAT reports. The Trade Desk and Amazon are public companies; this is commentary, not investment advice.