Criteo cuts full-year guidance again as ad tech earnings split in two

The August 5 earnings wave split the sector cleanly. Criteo cut full-year guidance for the second time in three months and replaced its chief financial officer, while Magnite raised four separate full-year targets on 36%connected-television growth. The dividing line ran between streaming and first-party identity on one side and open-web performance media and cross-border retail on the other.

A single Google search-spam rule wrote a hard dollar figure into an independent vendor's accounts for the first time. Taboola disclosed that deprecating its Explore More product cost more than $20 million of second-half ex-TAC gross profit, the first quantified cost of Google's back button policy at a named ad tech company.

Publisher economics diverged just as sharply. New York Times digital advertising rose 20.7% to $114 million, while News Corp advertising settled at 15% of revenue with zero constant-currency growth and BuzzFeed handed its ad sales to Allen Media Group after an $11.8 million loss.

The earnings wave that split ad tech in two

Wednesday was the busiest single day of the second-quarter reporting season for advertising technology, and the results did not point in one direction. They pointed in two. Companies exposed to connected television and first-party identity raised their outlooks. Companies exposed to open-web performance media and cross-border retail cut theirs. The dividing line ran straight through the sector, and it ran through individual company income statements too, separating the parts of each business that are growing from the parts that are shrinking.

The clearest casualty was Criteo. The commerce media company reported second-quarter revenue of $428 million, down 11%, and cut its full-year outlook for the second time in three months, publishing the results on August 5, 2026 alongside a Form 8-K filed with the U.S. Securities and Exchange Commission and covering the three months ended June 30, 2026. The company now expects Contribution ex-TAC, the non-GAAP profitability measure it uses in place of gross profit, to decline between 10% and 12% at constant currency for the year. In February it had guided to flat-to-up-2% growth on the same metric. In May that became a low-single-digit decline. Today's revision is the third position on one number in six months, and each successive position has been worse than the last.

The internal contradiction is what makes the quarter notable. Media spend, the volume of advertising money passing through Criteo's systems, grew 9% at constant currency to $1.1 billion. Revenue fell 11%. Contribution ex-TAC fell 13%as reported and 12% at constant currency to $255 million, missing the midpoint of Criteo's own May guidance of $260 million to $264 million by roughly $7 million and landing below the bottom of that range. Chief Executive Michael Komasinski did not dress it up. "While our second quarter top line performance was disappointing, our long-term strategy remains unchanged," he said in the earnings release, adding in prepared remarks for the 8:00 a.m. Eastern call that "the second quarter was a challenging one for Criteo. We did not meet the expectations we set for ourselves."

Two pressures produced the shortfall, and they run on different clocks. The first was known. Two Retail Media clients reduced the scope of their engagements, a change first disclosed on May 2, 2025, when Criteo revealed that its largest retail media client would discontinue managed services and curtail brand demand sales. That decision has been metabolising through the accounts ever since, and the modelling published today quantifies it quarter by quarter: $27 million in the first quarter of 2026, $21 million in the second, $20 million expected in the third, and $7 million in the fourth, a full-year total of $75 million. The second pressure is newer and less predictable. Several large enterprise clients in the Performance Media segment cut spending during the quarter, and did so more sharply than Criteo had modelled when it set May guidance. Chief Financial Officer Sarah Glickman attributed the $7 million miss specifically to those Performance Media dynamics rather than to the retail media headwind, which had been forecast and delivered on plan. Komasinski said the causes varied by customer, including travel-sector effects from Middle East conflict, tariff and de minimis import-rule changes, and clients shifting investment toward other parts of the funnel.

The segment detail tells the story. Retail Media revenue was $47.9 million, down 21% as reported and 22% at constant currency, but strip out the two affected clients and underlying Retail Media Contribution ex-TAC grew 20% across the rest of the base, with the take rate holding steady. Same-retailer Contribution ex-TAC retention came in at 84%, a figure that rises to 113% when the largest retailer is excluded. That gap between the two numbers is the entire segment compressed into one ratio. Media spend inside Retail Media reached $527 million, up 31% at constant currency, with more than 4,500 global brands active and a retailer footprint exceeding 235, and Criteo added Loblaw Advance in Canada, Monoprix and Druni in EMEA, and Olive Young and Golf Digest Online in APAC. It also won a Massmart retail media deal covering 280 stores in Africa and now runs auction-based display with more than 85 retailers, up from 60 in the first quarter. Performance Media revenue was $380.1 million, down 10% as reported and 9% at constant currency, with fashion media spend down 21% and travel moderating after several exceptional years.

Profitability held better than revenue. Adjusted EBITDA came in at $73.3 million, down 18% from $89.4 million but above guidance, and Criteo held the margin at 29% against 31%, a two-point compression on a 12% decline in the denominator. Operating expenses fell 9% to $207.2 million, with sales and operations expenses down 21%. Net income was $11.8 million against $22.9 million, a 49% decline, or $0.22 per diluted share, and adjusted net income was $40.5 million, or $0.80 per diluted share, ahead of the roughly $0.70 analysts had modelled. Free cash flow was negative $37.5 million, which the company attributed to seasonality and 2025 tax payments, and it expects positive generation in the second half. Capital expenditure is running to roughly $190 million for the year, up from $101 million in 2025, tied primarily to the renewal of two large data centres, which explains most of the gap between operating and free cash flow. Criteo repurchased $30 million of stock in the quarter and $61.3 million across six months, with $160 million left on the authorisation.

For the third quarter, Criteo guided to Contribution ex-TAC of $237 million to $241 million, a decline of 14% to 15% at constant currency, and adjusted EBITDA of $54 million to $58 million. The construction of the new full-year guidance is explicit about its own conservatism. "Our guidance reflects what we see in the business today and assumes no improvement in spending from those large enterprise clients through the balance of the year," Komasinski said. Glickman was equally direct on the margin figure: "We do not view 30% as the normalized profitability level for the business," she told analysts, adding that the model retains room for margin expansion as growth improves.

Alongside the numbers came a leadership change. Chief Strategy Officer Connor McGogney becomes Chief Financial Officer on August 10, 2026, succeeding Glickman, who has held the role since 2020 and will remain as an adviser through September 30. A guidance cut of unusual magnitude arriving on the same day as a CFO transition is the combination markets tend to punish, and the pre-market reaction bore that out. McGogney keeps his strategy and corporate development responsibilities alongside finance. Criteo also confirmed it completed its redomiciliation from France to Luxembourg on July 29, 2026, with a further transfer of legal domicile to the United States expected in January 2027, subject to shareholder approval. The company noted that its OpenAI advertising integration, now live across seven countries with Mexico and Brazil planned, is not expected to contribute meaningfully until 2027, and described OpenAI as its fastest-growing channel while explicitly excluding it from the 2026 outlook.

If Criteo was the wave's low point, Magnite was its high. The independent sell-side platform reported second-quarter revenue of $192.8 million and raised four separate full-year targets after connected television contribution ex-TAC grew 36% year over year to $97.1 million, filing a Form 10-Q the same day. "We significantly beat consensus expectations on both the top and bottom line in the second quarter," said Chief Executive Michael G. Barrett. Magnite beat the top end of its own guidance on all three revenue-adjacent measures it forecasts: total contribution ex-TAC of $189.6 million against a $177 million to $181 million range, CTV of $97.1 million against $90 million to $92 million, and DV+ of $92.5 million against $87 million to $89 million.

The channel split is the structural story. CTV now accounts for 51% of contribution ex-TAC against 44% a year earlier, and of the $27.6 million of additional contribution ex-TAC generated year over year, $25.6 million came from streaming. Mobile grew 3%, desktop was flat. Effectively all the growth came from one channel. The 36% CTV figure sits inside an accelerating sequence: it compares with 30% in the first quarter of 2026, when connected television first crossed half of Magnite's contribution ex-TAC at $82.3 million, with the 20% headline in the fourth quarter of 2025, and with 18% in the third quarter of 2025. Four consecutive quarters of acceleration, not a single strong period. Barrett attributed the performance to breadth rather than a handful of accounts, anchored by the SpringServe video ad server. The DV+ segment, covering everything outside connected television, produced $92.5 million, up 2%, reversing a 5% decline three months earlier and beating guidance that had anticipated a further 2% to 4% fall.

Adjusted EBITDA of $70.6 million was 30% higher, at a 37% margin, three points above a year earlier. Net income reached $19.4 million, or $0.13 per diluted share. Magnite raised full-year contribution ex-TAC growth guidance to 13% to 14% from at least 11%, adjusted EBITDA growth to greater than 20% from the mid-teens, adjusted EBITDA margin to at least 37% from at least 35.5%, and free cash flow growth to the high-40% range from the mid-30s. "Given this strong execution and ongoing shift toward programmatic streaming, we are also raising both our full-year top-line and margin expectations," Barrett said. One technical detail explains why 11% revenue growth understates 17% contribution ex-TAC growth: revenue reported on a gross basis, tied to managed-service insertion orders, fell to 3% of total revenue from 11% a year earlier as advertisers moved to automated buying, suppressing the headline line without touching contribution.

Magnite also addressed the same open-web erosion pressuring Criteo and Taboola head-on. Its 10-Q states plainly that AI is changing how users access information and could reduce the volume of open-web display inventory it monetises. The context is well documented: small publishers lost 60% of their search referral traffic over two years, and 69% of publishers on the Index Exchange saw year-over-year declines in ad opportunities during 2025. The difference is that Magnite has a streaming business large enough to absorb the shock, and Criteo does not. AdExchanger's coverage put Magnite's CTV revenue at $63 million, up 12% year over year on a GAAP basis, a reminder that the contribution figure and the revenue figure diverge because of the gross-versus-net treatment.

AppLovin sat in the middle, growing fast but slowing. The company reported second-quarter revenue of $1.92 billion, a 53% increase that marks the slowest growth in its own restated quarterly series, while guiding third-quarter adjusted EBITDA margin down to 83% from 84%. The headline that carries the most information is not the absolute number but the rate of change beneath it. Year-over-year growth across the disclosed comparison quarters ran 66%, 73%, 71%, 77%, 68%, 66%, 59% and now 53%. The 77% peak arrived in the second quarter of 2025, the same period AppLovin completed the sale of its gaming business to Tripledot Studios for $400 million, and every quarter since has printed a lower rate. Sequentially, revenue advanced just 4.5% from $1.84 billion in the first quarter, the smallest quarter-on-quarter gain anywhere in the twelve-quarter series the company published, against roughly 11% three months earlier and 18% in the fourth quarter of 2025.

None of this describes a business in retreat. A 53% annual growth rate at a $7.7 billion revenue run rate remains an outlier across public advertising technology, and net income of $1.27 billion grew 55%. But two things stood out beneath the top line. Research and development spending more than doubled, reaching $99.9 million against $44.0 million a year earlier, a 127% increase that lifted R&D from 3.5% to 5.2% of revenue while no other cost line behaved that way. Stock-based compensation grew 148% to $85.8 million. And cash conversion weakened sharply. Free cash flow of $863.3 million grew only 12.4% against adjusted EBITDA growth of 58%, dropping the free-cash-flow-to-adjusted-EBITDA ratio from roughly 75% to about 54%. The drivers were a more than sixfold jump in cash taxes paid across six months, to $639.8 million, and accounts receivable rising $351.7 million to $2.17 billion, now more than a full quarter of revenue, the kind of line that expands when a customer base broadens rather than deepens.

Third-quarter guidance of $2.055 billion to $2.085 billion implies year-over-year growth of 46.3% to 48.4%, a further step down, and the 83% margin guide is the first guided step-down in the disclosed sequence. AppLovin declined to publish forward-looking GAAP equivalents, citing the variability of stock-based compensation, a caveat that carries more weight than usual given that line grew 148% in the quarter just reported. The reporting period contained the June global opening of the Axon platform, closing what Chief Executive Adam Foroughi had described as a fourteen-year closed system, but the written materials disclosed no advertiser count, spend-per-advertiser figure or e-commerce split, deferring commentary to a 2:00 p.m. Pacific webinar. The company had earlier published a consumer social application called Gistwithout a press release, read as an attempt to build first-party inventory.

Zeta Global supplied the wave's most interesting disclosure, and it had nothing to do with the headline beat. The company published second-quarter revenue of $442.8 million, up 44%, and put a number on how much of that revenue comes from customers who have actually adopted its AI tools. The answer, from Chief Financial Officer Chris Greiner: the roughly one fifth of customers who have comprehensively adopted the AI tools account for about 70% of revenue. "We're introducing an initial framework to measure adoption and monetization of Zeta's AI, one underpinned by revenue," Greiner said, framing it as a response to investor requests, "because moats are ultimately proven by how long customers stay and how much more they spend over time."

The framework has four parts. Across all customers, including pilots and proofs of concept, the 20% of comprehensive adopters drive 70% of revenue. Among super-scaled customers, those generating at least $1 million on a trailing-twelve-month basis, 50% have comprehensively adopted and that half drives 75% of super-scaled revenue. Comprehensive adopters, described as AI super users, grew four times faster than the 80% still early in adoption. And their year-to-date net revenue retention runs 400 basis points above the company-wide figure and more than 20 percentage points above customers still ramping. None of these ratios are audited, and none appear in the financial statements. They are management disclosures whose usefulness to anyone comparing vendors depends entirely on whether rivals adopt comparable definitions, which they have not. A separate disclosure sat alongside them: super-scaled customer relationships now average 56 months, up from 48 months a couple of years ago, an eight-month extension in average tenure across the highest-value cohort.

The rest of the quarter was a clean beat-and-raise. Revenue exceeded the guidance midpoint of $419 million to $422 million by $23 million. Excluding the Marigold enterprise contribution of $48.2 million, organic growth was 28%, and Zeta used the total-revenue arithmetic to claim what it calls the rule of 64, revenue growth plus adjusted EBITDA margin. Super-scaled customer count reached 197, up 17% and a seventh consecutive quarter of sequential expansion, and super-scaled average revenue per user reached $1.8 million, also up 17%. Adjusted EBITDA was $91.7 million and net income $8.2 million, reversing a year-ago loss. Free cash flow guidance for 2026 rose $20.3 million. Two usage figures drew attention: Chief Executive David A. Steinberg said users interacting with the Athena agent by voice engaged at a level 500% higher than non-Athena users, and Greiner said 83% of customer interactions with Athena are now spoken, a figure that, if it holds, changes the design assumptions behind dashboards and permissioning. Steinberg also said 89.6% of new code generated in the quarter was automated, up from 75% in the first quarter, through an internal tool called Spade, and that total internal AI spending remains well under 1% of revenue.

Pinterest completed the wave with a beat undercut by a regional warning. The company reported second-quarter revenue of $1.180 billion, up 18%, but Europe grew just 12% on a reported basis and 7% stripped of currency after Asia-based cross-border retailers pulled back under regulatory pressure. Revenue of $1,179.7 million cleared the guided range, and adjusted EBITDA of $311.3 million beat by a wide margin, lifting the full-year margin outlook to approximately 30%. But a GAAP net loss of $46.7 million, against net income of $38.8 million a year earlier, sat underneath, driven by share-based compensation rising to $324.5 million and $14.3 million of restructuring charges. On a non-GAAP basis the picture inverts, with net income of $249.5 million and EPS of $0.43 against $0.33.

The regional split is the read-across. United States and Canada revenue accelerated to 18% growth from 13% in the first quarter, reaching $879.9 million, but Europe more than halved from 27% growth in the first quarter to 12%, at $212.7 million. Chief Financial Officer Julia Donnelly named three forces: a lapping effect from cross-border spend diverted into Europe in 2025 after United States tariff changes, a self-inflicted go-to-market reorganisation, and a mid-quarter shock. "We faced incremental pressure mid-quarter from Asia-based cross-border retailers impacted by regulatory actions, particularly in Europe," Donnelly said, adding that the pressure is continuing into the third quarter. Pinterest did not name the retailers, but the timing lines up with the European Union abolishing its 150 euro customs duty exemption on July 1, 2026, replacing it with a flat 3 euro per-item charge, a change that has already halved Temu's presence in Google Shopping and pushed Shein toward a near-exit. When a category of high-volume, low-margin cross-border sellers withdraws from paid acquisition across a continent, the platforms carrying that lower-funnel spend register it in a single quarter. Third-quarter guidance of 13% to 15% growth extends the drag, and Donnelly attributed further Q3 caution to the World Cup ending and Amazon's Prime Day shifting into the second quarter, together roughly a point and a half of headwind. The regional monetization gap is stark: United States and Canada generated $8.30 per user, Europe $1.35, and Rest of World $0.23, against 640 million monthly active users, an eleventh consecutive quarter of double-digit user growth.

Google's gravitational pull on everyone else's profit and loss

If one theme connected otherwise unrelated August 5 stories, it was the extent to which Google's product and policy decisions now show up directly in other companies' results, dashboards and distribution. The clearest example carried a dollar figure.

Taboola reported second-quarter revenue of $476.8 million, below the bottom of its own guidance range by $15.2 million, after a Google policy change eliminated a product it had expected to contribute more than $20 million of ex-TAC gross profit in the second half of 2026. The product is Explore More, a session-depth tool Taboola has sold to publishers for years, which let users discover additional sponsored content after clicking the back button. National World, one of the largest regional publishers in the United Kingdom, signed a multi-year deal in October 2024 that included Explore Morealongside the Taboola Feed unit. Chief Financial Officer Stephen Walker said Google policy no longer permits such products.

That policy has a date and a paper trail. Google named back button hijacking as an explicit spam violation on April 13, 2026, with enforcement beginning June 15, placing it in the malicious practices category alongside malware distribution. The conflict then spread inside Google itself, when an AdSense vignette trigger was found to match the company's own definition of the violation, a collision Google resolved by removing the back button trigger from vignette ads on the same June 15 date. Taboola is now quantifying the cost. Walker said that without the policy change, ex-TAC gross profit would have exceeded the high end of guidance. As reported, ex-TAC of $192.372 million sat inside the $189 million to $194 million range rather than above it, which makes the qualification material rather than rhetorical. It is the first hard dollar figure attached to Google's back button policy at a named ad tech company, and it establishes a precedent: a search spam rule written by one team at Google removed a monetization product sold by an independent vendor to thousands of publishers. Publishers running comparable session-extension mechanics carry the same exposure, and the enforcement deadline has already passed.

The rest of Taboola's quarter pulled in two directions. Ex-TAC gross profit rose 11.8% to $192.4 million, adjusted EBITDA reached $55.5 million, up 22.8% and well above the $36 million to $43 million guided, and net income of $4.3 million reversed a year-ago loss. But revenue grew only 2.4%, dragged by a second, chosen headwind: Taboola exited publisher relationships that did not meet its standards for advertiser success, booking a one-time non-cash write-off of $12.2 million in publisher prepayments it no longer expects to recoup. Reducing supply while holding demand steady raises prices, and Walker attributed the ex-TAC outperformance to exactly that mechanism, plus a mix shift toward higher-margin areas and Taboola News. For media buyers, the implication is concrete: buyers running performance campaigns across the network are transacting against a smaller, more expensive inventory pool than three months ago, and the company intends to keep pruning. Adjusted EBITDA margin expanded to 28.8% from 26.2%, or 33% at constant currency, since foreign exchange took roughly $7.5 million off the figure.

Scaled advertisers, those with more than $100,000 of cumulative gross spend over a trailing four quarters, numbered 2,081, up about 2%, and accounted for roughly 86% of revenue, but average revenue per scaled advertiser was flat at approximately $197,000, down from a $208,000 peak. Distribution is holding, wallet share is not expanding. Chief Executive Adam Singolda named Fox News as a competitive win, describing it as a top-five United States publisher, and teased an unnamed full-page expansion with a premier media and entertainment company, where display alone is estimated to be worth two to three times the native placements Taboola historically monetized there, continuing a direction begun when it added display inventory across TIME, Gannett, Nexstar and Slate in October 2025. Both wins begin contributing in the fourth quarter. Realize+ has drawn more than 300 advertisers, and Singolda disclosed that Taboola's Model Context Protocol and Claude integrations carry a few million dollars of advertiser spend, one of the few disclosed spend figures attached to an agent-mediated buying interface. Full-year ex-TAC gross profit guidance rose $7 million at the midpoint to $772 million to $783 million even as third-quarter revenue guidance implies a 6% year-on-year decline at the midpoint. The revenue miss sits inside a broader compression of open-web inventory, with Alphabet's Network segment falling 4% to $6.97 billion in the first quarter and Google itself arguing in court that open-web display dropped from over 40% of AdWords display impressions in January 2019 to 11% in January 2025.

Google's gravitational pull showed up next in the mechanics of the auction itself. Twelve days before a scheduled bidding change, Google published a question-and-answer post on the August 17 transition hours after an ecommerce analyst told advertisers the date is not a cliff. The post, from Ads Product Liaison Ginny Marvin, sets out that campaigns limited by budget and over-achieving their targets will see performance adjust to their set targets, and that the system will aim to hit the advertiser's CPA or ROAS target regardless of budget settings. The operationally specific passage concerns inaction: if no action is taken, the system will aim to find conversions at the average target set in the campaign, which, in Marvin's words, "may mean entering different auctions than you previously competed in in order to achieve that higher ROAS or the lower CPA target." That describes a change in inventory, not only in arithmetic. A campaign converting well inside its stated target has, in effect, been buying a narrow, cheap slice of the auction, and pulling delivery toward a looser stated target means bidding into queries and placements it previously skipped. Marvin's framing: in campaigns where budget is not the binding constraint the target already functions as the spend control, but in budget-constrained campaigns the target now becomes the efficiency lever.

Two hours earlier, Mike Ryan, Head of Ecommerce Insights at Smarter Ecommerce, had argued the reaction had outrun the mechanics. "Google advertisers are losing their minds over August 17th," he wrote, calling the cliff framing a misconception and adding that "luckily, it isn't." His technical objection turned on a verb: Google does not say actual ROAS will immediately revert to target, only that performance will trend toward it. He put the evaluation window at 30 to 60 days, or one to two conversion cycles, before advertisers judge the change, matching the timing guidance in Marvin's own post and earlier documentation covered when Google Ads got promotion mode and a bidding overhaul scheduled for this August. "Your limited budget campaign isn't going to break overnight," Ryan wrote, arguing that what might break it is overreacting and moving budget or target in one large jump. He attached three criticisms: the change was miscommunicated, some Google sales decks set unrealistic expectations, and scheduling the switch for peak northern-hemisphere vacation season was, in his word, borderline offensive. A Google sales slide he reproduced made the platform's own case, headed with the admission that a budget-constrained campaign can appear to overachieve even while missing high-quality demand, because the budgeting system lowers bids to stretch spend and produce cheaper conversions. In that reading, cheap conversions are a by-product of throttled bids rather than evidence of a well-run account.

The gravitational pull turned punitive on Blogger. Owners of blogs hosted on Google's Blogger platform began receiving removal notices on August 4, 2026, citing a malware policy, with dashboards displaying an 89-day countdown to permanent deletion. Google told PPC Land on August 5 that a bug had incorrectly flagged some Blogger-hosted sites as malware for less than a day, and that a fix was in progress. The discrepancy between interfaces was more than cosmetic: notification emails named the malware and malicious content policy specifically, while the dashboard cited Community Guidelines generically, and neither surface told owners what had been detected, on which URL, or at what time. The most consequential technical detail was the absence of any corroborating security signal, with affected owners reporting that both Search Console and Safe Browsing showed nothing. That gap matters because Google's malicious practices category normally surfaces through Search Console, and when Google began sending Search Console warnings for back button hijacking in April 2026 those notices included sample URLs and policy links. Blogger's enforcement surface produced a padlock and a countdown. Sites aged 12 and 18 years were locked, active blogs were hit while inactive ones in the same accounts were spared, some owners noted Cloudflare-fronted blogs appeared to escape, and no figure for the number of affected blogs was published. Glenn Gabe of G-Squared Interactive brought the incident to a wider search-industry audience, writing that "renting versus owning is a dangerous game" and "Poof, gone."

Google's gravitational pull also drew a rare on-record denial. Google communications representative Jennifer Kutz told The Verge that Reddit receives no special preference in ranking systems or AI search features, a statement reported by Search Engine Roundtable at 7:51 a.m. on August 5, days after Reddit's stock fell 12.49% on referral fears. That is what has changed since the earlier coverage of Reddit's earnings. The platform's traffic troubles and stock decline were already known; what is new is Google addressing the ranking question directly. Kutz said forums carry a mix of quality and that Google's systems detect helpful versus low-quality user-generated content. On spam, her wording was broader: "There will always be bad actors trying to game the system," she told The Verge, adding that Google has "kept results 99% spam-free for years." Three claims sit inside that answer, none testable from outside Google. The denial is narrow in a way a reader comment on Search Engine Roundtable flagged: "no special preference" is not "no preference," and the qualifier is left undefined.

The commercial backdrop is a matter of record. Reddit signed a data licensing agreement with Google on February 21, 2024, reported at roughly $60 million a year, and the arrangement gave Google effective exclusivity over indexing recent Reddit content through a robots.txt configuration that blocked most other crawlers, supplying training material for Gemini. Google's own documentation records a more specific admission than Kutz's general capability claim: the company only extended its spam policies to cover AI Overviews and AI Mode on May 15, 2026, after those surfaces had run at scale for roughly two years, applying existing rules on scaled content abuse, inauthentic mentions and site reputation abuse to the generative layer for the first time. The gap between policy text and enforcement has been visible in Google's own products, from fabricated earnings stories persisting on Google Finance to self-promotional listicles surfacing inside AI Overviews. The Verge report that prompted the question examined whether Reddit can contain promotional posting aimed at AI search, opening with a skincare thread in which a marketing-style comment sat three upvotes deep in a 199,000-visitor community, close to invisible to human readers but not to a retrieval system assembling an answer from indexed text.

Finally, the gravitational pull created an opening for a rival. Brave reported 122.55 million monthly active users and 2.46 billion monthly search queries for July 2026, and said its search engine was added to Google Chrome's global default search engine list during the same month. The figures arrived through LinkedIn rather than a press release. Monthly users rose 1.3% month over month and queries rose 1.9%, with 51 million daily active users. The engagement ratio held steady at 0.416 daily-to-monthly, against 0.42 in September 2025, meaning new arrivals use the browser as often as existing ones rather than accumulating as dormant installs, a meaningful distinction for anyone modelling reachable inventory. Search grew faster than the browser carrying it: 2.46 billion monthly queries annualise to roughly 29.5 billion against nearly 20 billion reported for September 2025, and query intensity rose to about 20.1 searches per user per month from roughly 15.8.

The Chrome listing is the most consequential line and the least detailed. Brave said only that it was added to Chrome's global default search engine list, offering no specification of which Chrome versions or channels carry it, whether it applies to desktop or mobile, what position Brave occupies, or whether the change stems from a Google product decision or a regulatory obligation. Scale explains why it matters regardless: Cloudflare's measurement placed Chrome at 66.282%of global web traffic and Brave at 1.083%, and Google at 90.8% of the global search market against Brave Search at 0.279%. Being present in a settings menu inside the browser that carries two thirds of the world's web traffic is a distribution event of a different order than anything Brave can achieve through its own installed base. Whether presence converts into use is a separate question, and the evidence urges caution: a University of Antwerp analysis found Digital Markets Act choice screens delivered Firefox roughly 6 million additional EU users while leaving Google at approximately 90% of the search market. A menu entry is not a default, and the distinction is the entire lesson of a decade of European enforcement. Brave attributed part of July's strength to the FIFA World Cup, which concluded with the July 19 final and lifted usage above a typical summer slump.

Publisher economics, pulling apart at the seams

The publisher results reported on August 5 diverged as sharply as the ad tech vendors, and along a similar fault line: subscription-led businesses held up, advertising-led ones did not.

The New York Times Company reported second-quarter digital advertising revenues of $114.0 million, a 20.7%increase that cleared the high-teens growth it had guided to. Digital advertising has now grown at 18.7%, 20.3%, 24.9%, 31.6% and 20.7% across the last five quarters, a sequence few publishers of comparable scale have matched. Total revenue reached $762.5 million, up 11.2%, and adjusted operating profit rose 16.1% to $155.3 million. Print advertising fell 11.1% to $35.2 million. Chief Financial Officer Will Bardeen attributed the digital gain to demand: "Marketer demand in the quarter for our high performing ad products exceeded our expectations," he said. He tied the one line that missed guidance, adjusted operating costs up 10.0% against an 8% to 9% guide, directly to that beat: "Cost growth exceeded our guidance range primarily due to incremental variable compensation tied to financial outperformance." Advertising beat, incentive compensation rose accordingly, the cost line moved past guidance as a consequence.

President and Chief Executive Meredith Kopit Levien described the drivers in portfolio terms, naming news, games and sports as the areas of greatest scale while insisting growth came from across the portfolio. "Campaigns renew because the ads perform," she said. The company staffed a new middle-market advertising sales team in 2026 to reach demand below the enterprise threshold, a detail relevant to any agency operating there, because it signals coverage for accounts that would previously have transacted only programmatically. On video, she was explicit that it is not yet an advertising story, playing "a relatively minor role in the growth in advertising so far," with production, then engagement, then monetization the stated sequence. Subscriptions decelerated even as revenue per user accelerated: net digital-only additions of approximately 280,000 sat at the low end of the recent run, below the 310,000 of the prior quarter and analyst estimates near 295,000, while digital-only ARPU reached $9.94, up 3.1%, helped by a bundle price increase moving a tenured cohort from $25 to $30. The company ended the quarter with approximately 13.35 million total subscribers, including 12.80 million digital-only. Third-quarter digital advertising guidance points to slower mid-to-high-teens growth, with Kopit Levien noting the company is lapping strong supply growth from the back half of 2025, and the shares fell in premarket trading on a total-subscription-revenue outlook that moderates to 9% to 11%. Her closing remark framed the whole publisher thread: the results were delivered against a backdrop shaped by a small number of big tech companies whose moves result in less traffic to publishers.

News Corp sat at the opposite pole. The company reported fiscal 2026 revenues of $9.03 billion, a record on a continuing-operations basis, and a year in which advertising settled at 15% of the revenue mix and grew nothing once currency is stripped out. Consolidated advertising revenue for the fourth quarter was $363 million, but on a constant-currency basis it was $353 million, identical to the prior-year period. Growth was zero. The full year told the same story: advertising of $1,391 million was up 2% as reported but 0% at constant currency, and every dollar of the reported increase came from exchange rates, with foreign exchange adding $71 million to fourth-quarter revenue as the Australian dollar averaged $0.71 against $0.64 a year earlier. Advertising now represents 15% of group revenue, down from 48% in fiscal 2014, while circulation and subscription is 36%, and digital revenue reached 61% of the total against 22% in fiscal 2014. The largest single revenue line at one of the world's biggest news publishers is now a subscriber relationship rather than an impression.

The segments underneath tell where the buyer relationship has migrated. Dow Jones grew fourth-quarter revenue 7% to $644 million, with digital advertising up 10% while print advertising fell 6%, and digital now 69% of the segment's advertising against 65%. Risk and Compliance revenue grew 11% to $102 million and Dow Jones Energy 4% to $76 million, both faster than any advertising line, and digital was 84% of Dow Jones total revenue. Digital Real Estate Services carried the quarter, with revenue up 19% to $553 million and Segment EBITDA up 46% to $222 million on the strength of REA Group, up 21% to $386 million, and Move, operator of Realtor.com, up 13% to $167 million, though Realtor.com's monthly unique users fell 6% to 68 million. News Media revenue rose 5% to $574 million, but segment advertising fell 5% in constant currency to $197 million, and the masthead audience figures were the sharpest data in the filing: The Sun's digital reach fell to 65 million global monthly unique users from 87 million, and New York Post's network to 77 million from 90 million. The segment carried FIFA World Cup costs at News Broadcasting and the newly launched California Post; the tournament ran June 11 to July 19 and, per iSpot, generated more than 25 billion United States television advertising impressions, with United States men's national team matches drawing 70% of FOX tournament viewers, after Video Advertising Bureau research projected 63.9 million American adults would watch.

The World Cup also framed Fubo, whose quarter was the clearest case of a big event failing to move the advertising line. FuboTV reported advertising revenue of $108.9 million for its fiscal third quarter, a shade below the $109.4 million pro forma figure a year earlier, even though the tournament generated three times the advertising revenue of the 2022 competition. It was the first full quarter Fubo inventory ran through the Disney Ad Server, following the platform's affiliation with The Walt Disney Company, and the first men's World Cup under that structure. The tripling refers to tournament-specific revenue against 2022, not the quarter as a whole, and a tournament that generated more than 25 billion United States television impressions produced a Fubo advertising line $500,000 lighter than the prior-year pro forma equivalent. Sequentially the advertising line improved, from $91.5 million to $101.6 million to $108.9 million across three quarters, which the company attributed to Disney Ad Server integration lifting fill rates and CPMs. New Chief Executive Alisa Bowen, appointed in July, cited "encouraging improvements in advertising capacity utilization and CPMs since completing our integration with Disney Advertising." Adjusted EBITDA fell 38% to $19.1 million and the margin dropped to 1.3%. Total North America paid subscribers recovered to 5.75 million from a spring dip, up roughly 20,000 sequentially, though the base remains well below the 6.18 million of two quarters earlier. Advertising accounted for 7.4% of total revenues of $1.482 billion.

BuzzFeed was the publisher-economics story with the sharpest structural change. The company reported second-quarter revenue of $36.3 million, a 21.8% decline, and disclosed that it has handed responsibility for selling its advertising inventory to Allen Media Group, the private company controlled by its own chairman and chief executive, Byron Allen. It was the first full quarter under Allen's control, and it was not a recovery. Advertising revenue fell 23.4% to $17.3 million, commerce and other revenue fell 31.4% to $9.0 million, content revenue fell 6.5% to $10.0 million, and net loss reached $11.8 million. Adjusted EBITDA swung to negative $1.7 million from positive $2.0 million a year earlier, a margin move from 4.3% to negative 4.8%. Costs fell only 6.5% while revenue fell 21.8%, and general and administrative expense actually rose 28.8%.

The sales representation agreement arrived without financial terms, a start date, a minimum commitment or a commission structure, and was not characterised as a related-party transaction despite Allen's dual role as founder, chairman and chief executive of both companies. Allen said the arrangement engages "AMG's sales organization to expand monetization of BuzzFeed's advertising inventory and give brands and agencies another point of entry." For agencies, the practical consequence is that BuzzFeed inventory is now sold by a sales organisation built around broadcast television and free streaming, becoming a line within a cross-media proposal rather than a publisher pitching audience segments against editorial verticals. BuzzFeed had earlier kept parts of its programmatic supply in premium auctions, joining nine publishers backing The Trade Desk's OpenAds platform in January 2026, leaving a two-track strategy the results do not yet separate. Audience Time Spent fell 10.8% to 62.3 million hours, and direct visits plus internal referrals rose to 65% of United States traffic, up from 61% in the prior quarter, though the measurement basis differs from the 63% figure reported for BuzzFeed.com in November 2025, so the two do not form a clean series. A rising direct share is also consistent with search and social referrals falling faster than direct visits, a different story from direct visits growing.

Bumble closed the publisher-and-demand thread from the buy side, a signal of where marketing budgets are going. The dating app operator halved its selling and marketing budget across the first six months of 2026 to $56.2 million from $91.8 million, a 38.8% reduction, with almost all of the cut in the first quarter, where spend dropped about 54.9% year on year. Dating apps have historically ranked among the heaviest buyers of mobile install inventory, so a participant of Bumble's size cutting its promotional budget by more than a third in six months registers across the demand side, though the company does not disclose channel-level allocation. Second-quarter revenue fell 15.2% to $210.5 million, and total paying users fell 16.4% to 3.16 million, a loss of 620,000 across twelve months, though the sequential bleeding has slowed to near zero. Product development spending rose to $35.5 million, climbing to 16.9% of revenue from 13.1%. Chief Financial Officer Kevin Cook signalled a turn, saying the company is now "deliberately investing across product, technology, and brand as we prepare to deliver on our innovation roadmap." Third-quarter guidance of $205 million to $213 million revenue and $56 million to $60 million adjusted EBITDA implies a margin stepping down to roughly 26% to 29% from 34.6% in the quarter just reported and 38.9% in the first quarter. Spending is going back in somewhere, and the margin guidance is where that decision becomes legible.

The data and identity plumbing

Two August 5 stories concerned the infrastructure underneath everything else: how identity data changes hands, and how conversion events are packaged for the clean rooms that increasingly mediate measurement.

LiveRamp reported first-quarter fiscal 2027 revenue of $214.0 million, up 9.8%, while confirming that its shareholders vote on the Publicis Groupe takeover on August 17, 2026. The data collaboration company held no earnings call and issued no guidance, decisions taken in light of the pending transaction, which valued its equity at $38.50 per share when the definitive agreement was signed on May 16, 2026. This was the first full quarter reported under merger constraints, and the filing shows where they bite: share repurchases that absorbed $194 million across fiscal 2026 have been suspended, and capital expenditure, material transactions, asset disposals and new debt now require the acquirer's consent. Closing is still set for before the end of 2026.

The operating business grew faster on the bottom line than the top. Subscription revenue reached $160.4 million, up 8.1%, and Marketplace and Other revenue reached $53.6 million, up 15.4%, with Data Marketplace alone up 13% to $40 million. Roughly 84% of subscription revenue was fixed and 16% usage-based. GAAP income from operations more than doubled to $20.2 million, lifting operating margin to 9.4% from 3.7%, and adjusted EBITDA reached $51.0 million against $36.4 million. The line that moved most was $6.5 million of legal and professional fees tied to the merger agreement. The key performance metrics told a more restrained story than the revenue line: annualized recurring revenue of $539 million grew 7%, subscription net retention slipped to 103% from 104%, and the count of customers above $1 million in annualized subscription revenue was 132, up from 127 a year earlier but down from 140 at December 31, 2025. For media buyers, the vote matters because Publicis absorbing an independent identity vendor removes a neutral party from the middle of the measurement stack, and the mechanics of the August 17 vote determine when that happens. The earlier clean-room measurement thread is now secondary to the ownership question.

IAB Tech Lab opened the other piece of plumbing to comment. The standards body published a set of technical guidelines telling advertisers how to package conversion events for data clean rooms, with comments closing on September 4, 2026. The document, titled ECAPI Guidelines for Data Clean Rooms, extends the Event and Conversion API standard and is published in the ECAPI repository on GitHub alongside the core specification, a changelog and an examples file. It does not create a new standard; the body says the clean room use case does not warrant one because the events are the same as those already defined in the core specification. What differs is packaging.

The requirements are specific. Values that ECAPI v1.0 designated SHA-256 hashed must be either hashed under the standard's normalization rules or encrypted through Parquet Modular Encryption or application-layer encryption before upload, and unprotected values must not appear in any Parquet file delivered to a clean room, written in the capitalised MUST NOT convention. That prohibition matters because hashing has repeatedly been treated as a sufficient privacy control, a position the Federal Trade Commission rejected in July 2024 when it warned that hashed identifiers remain identifiers, following that with November 2024 guidance that clean rooms are not privacy-preserving by default. The guidelines describe two Parquet layouts: a recommended nested-structs approach using Dremel encoding and producing one row per event, and a fully denormalized wide table where every child array is exploded, so that a generate_lead event for a user with two email addresses and two phone numbers produces four rows. The nested approach avoids combinatorial explosion, and the denormalized layout carries a double-counting risk because the value field repeats across every row from the same event, which is why the guidelines warn against querying without deduplicating on the event identifier first. Partitioning follows Hive-style directory conventions, with campaign-plus-day described as the most common, and once all files for a partition are written the uploading party must write an empty marker file named done, without which a reader cannot distinguish an in-progress upload from a finished one. The guidelines place this in a lineage that includes the Hadoop MapReduce _SUCCESS file, Spark's FileOutputCommitter and the World Federation of Advertisers cross-media-measurement done blob.

Rather than a separate manifest, the guidelines use the key-value pairs a Parquet file can carry in its footer, all prefixed with ecapi. to avoid collisions, and require seven metadata keys on every file: the ECAPI schema version, the layout in use, the advertiser identifier in the clean room, the data set identifier, a unique batch upload identifier expressed as a version 4 UUID, the uploading party, and a key recording whether the data was hashed, encrypted or both. The draft itself carries an inconsistency worth flagging, since the metadata table names the seventh key ecapi.hash_encrypted while the encryption section refers to ecapi.privacy_method for the same purpose. As the document notes, public comment periods exist to catch discrepancies of exactly that kind. The guidelines also recommend Trusted Execution Environments with key-encryption-key and data-encryption-key wrapping, and specify field mappings, with JSON numbers rendered as DOUBLE and extension fields JSON-encoded as UTF-8 strings. The closing section cites IAB Tech Lab research that 72% of publishers identified technical integration challenges as a major obstacle to conversion API adoption and that two-thirds of advertisers reported improved return on ad spend after implementing one, while stressing that consent remains the unresolved layer, since ECAPI standardises transmission and not the legal basis for collection. It also cites a June 21, 2026 New York Times clean room deal reporting a 61% increase in international click-through rate against cookie benchmarks.

AI adoption economics, where confidence meets the data

Several August 5 stories, read together, described the same tension from different angles: enterprises are spending heavily on AI while struggling to connect it to a defensible outcome, and vendors are racing to supply the governance and measurement that gap requires.

TransUnion put the clearest number on it. The identity company published survey research finding that 89% of senior US marketing leaders expect to increase spending on AI-enabled marketing over the next 12 to 24 months, while only 36% rate their data foundations and internal processes as highly ready to support it, a gap it calls the AI Confidence-Readiness Paradox. Only 53% said they had seen meaningful ROI from AI-driven marketing. The research, commissioned from UTA Advisory, the brand advisory division of United Talent Agency, covered 100 senior marketing and technology leaders at US enterprise brands with annual marketing budgets of at least 50 million dollars, all already using AI in core workflows and screened to exclude those confined to basic productivity tasks. These are not marketers deciding whether to adopt. They have adopted, at companies with the budgets to do it properly, and a majority still cannot connect the technology to a defensible business outcome.

The distribution of use cases explains the paradox. Productivity applications reached 92% adoption and basic content development 89%, but the higher-value applications fell away: 45% applied AI to marketing measurement and 33% to customer or market insights. Reported outcomes tracked the same shape, with 75% citing reduced manual effort but only 44% citing reduced media waste and 35% improved conversion rates. The pattern is a technology delivering time savings reliably and revenue effects intermittently. Fragmentation was the most named obstacle, cited by 67%, and readiness ratings on people, process and data ran 42%, 36% and 36% at the high end. "Marketers are increasingly confident in AI's ability to drive business results, but many are still working to build the foundations needed to scale it effectively," said Matt Spiegel, EVP, TruAudience Growth Strategy at TransUnion. "AI isn't a shortcut around data challenges. It's a force multiplier." With a sample of 100, each percentage point is roughly one respondent, placing the figures closer to a directional reading than a precise population estimate. The finding sits alongside a body of research PPC Land has tracked, including Publicis Sapient's confidence-capability gap and DigiCert research finding nearly half of firms unable to fully trace AI decisions to source data. Measurement stayed anchored to cost, with 65% assessing AI impact primarily through estimated time or cost savings, and only 31% using multi-touch attribution.

Zeta's 20%/70% adopter disclosure, covered in the earnings wave above, is the vendor-side counterpoint to the TransUnion finding. It is one of the few attempts to quantify the same question TransUnion posed to buyers, and it points in the same direction: comprehensive adoption and business outcome move together, while shallow adoption does not.

PubMatic addressed the governance half of the gap. The sell-side platform opened a five-step governance layer inside AgenticOS, its autonomous buying platform, restricting what AI agents can transact and recording every decision they make, with Rise, a Quad agency, among the first partners piloting it, one day before PubMatic reports a quarter its own guidance places in decline. The framework's five components each address a failure mode in autonomous execution: a dual-tier permission system setting platform-wide and buyer-configured boundaries; business rules agents inherit before a campaign begins; pre-approved asset libraries from which agents draw only vetted inventory, creative and audiences; authenticated approval workflows where an agent encountering a decision beyond its authority halts and notifies a human approver who authenticates with a one-time passcode; and audit trails with drift detection that flag anomalous behaviour for human review rather than executing it. Presence-based controls, including brand safety and geo-targeting parameters, must be populated before execution, and if a required field is missing the agent halts until the gap is resolved.

The one-time passcode requirement is the most specific technical detail, converting a human approval from an implicit action into a cryptographically verified event, which matters for any agency later needing to prove to a client who authorised what. The guardrails also run as an independent check on Ad Context Protocol traffic, addressing a structural gap the protocol has carried since its October 15, 2025 launch, since AdCP standardises how agents discover inventory but does not adjudicate whether an agent is behaving within a buyer's commercial limits. That gap has drawn industry criticism since the protocol first appeared, with practitioners arguing that a transaction layer built before a transparency layer repeats the sequencing error of early programmatic. Standards bodies have been working the same problem, with IAB Tech Lab's AAMP initiative and AAMP 2.3 shipping on July 30, 2026 with a pricing-provenance field. What separates PubMatic's release from that standards work is enforcement location: a protocol field describes a claim, while a platform check rejects a transaction. AgenticOS, which debuted on January 5, 2026, has since run more than 80 autonomous campaigns across 100,000 or more websites, apps and streaming services, with more than 4,000 direct deals executed, up from more than 30 campaigns and more than 1,000 deals at its first-quarter results. Both remain small numbers against a platform that processed 94 trillion impressions in the quarter. PubMatic sits on the IAB Tech Lab Programmatic Governance Council alongside Omnicom, WPP, Dentsu, Disney, Amazon Ads, The Trade Desk and Magnite.

Meta reordered its developer support around AI. The company made its AI Developer Assistant the primary support path on developers.facebook.com, placing a conversational interface in front of the documentation library and the ticket queue, with full availability set for mid-September 2026. Nothing says human support has been withdrawn; what changed is ordering, with the model now sitting between the developer and everything that came before it. Three categories fall inside its scope: onboarding, product issues covering Graph API and SDK behaviour, and compliance and review. That third category is the one with money attached, because app review and business verification are gating functions, and an app that fails review cannot ship while a business that cannot complete verification cannot access the permissions commercial integrations depend on. Meta said the assistant is connected to its documentation, API references and platform knowledge and does not return generic answers, though two of its three worked examples describe outputs that already exist as static documentation. Meta claimed considerable increases in first-time app review approval rate and business verification completion rate for developers who used an earlier Business Messaging version of the assistant, but published no figures, no baselines and no sample sizes, an unusual construction for a company that reports engagement numbers routinely elsewhere. The first metric also carries an unaddressed selection problem, since developers who consult a tool before submitting may differ systematically from those who do not, and version currency is the more testable claim, since an assistant answering against a superseded API version produces an answer that is coherent and wrong.

Optimove rounded out the AI thread from the consumer side. Its research arm published a Holiday Shopping Report 2026 finding that 72% of surveyed United States consumers now consult AI assistants for gift ideas, yet a majority say they will not shop anywhere they did not shop last year. The study rests on a survey of 648 US consumers conducted in Spring 2026, adults aged 18 and over with household incomes of $75,000 or more. Of the 72% consulting tools such as ChatGPT, Claude or Gemini, 49% answered regularly and 23% occasionally, and the 49% regular-use figure is unchanged from Optimove's April Mother's Day report, suggesting regular AI consultation has held steady across seasons. The finding cuts both ways: AI has become a measurable part of the consideration stage, but 53% of respondents intend to buy only from last year's stores, a loyalty signal that constrains how much any discovery tool can redirect. Product recommendation features led actual AI-feature usage at 55%, chatbots 41% and visual search 31%, and data privacy was the leading factor that would increase willingness to use an AI shopping assistant, cited by 37%. Whether stated usage translates into platform share is a separate question, and a Smarty Marketing survey found AI platforms captured roughly 7% of stated shopping search preference while Google retained the lead. The report carried two internal inconsistencies Optimove did not reconcile, describing the sample as both 648 and 647 respondents and its AI-trust finding as both almost three-quarters and three-quarters.

Measurement and distribution

Two August 5 stories widened the surfaces advertisers can measure and the content streaming apps can carry.

Nielsen removed a restriction that had limited its cross-screen measurement in Japan. The company extended Four-Screen Ad Deduplication from YouTube campaigns to every measurable connected TV publisher available through Nielsen ONE Ads on August 3, 2026, two months after four-screen measurement first arrived in the market in June. The distinction matters more than a scope note usually would, because a deduplicated reach figure is only as complete as the inventory feeding it: a product that resolves duplicate exposures across computer, mobile and connected television but can see only one publisher on the CTV side describes a slice of a campaign rather than the campaign itself. "Expanding Four-Screen Ad Deduplication across measurable CTV inventory gives the industry a clearer view of total campaign reach, frequency and incremental audiences, without being limited to a single publisher," said Matty Lin, Nielsen's APAC Regional Sales Leader. The expanded reporting includes unique audience, audience reach, average frequency and gross rating points, spanning both the digital vocabulary of reach and frequency and the broadcast vocabulary of GRPs, and it includes co-viewing in total impression delivery, a variable device-based measurement has historically missed and one Google Ads addressed with a Total Co-view definition on June 2, 2026. Data will initially be delivered through offline reporting with no forced campaign splits, but a caveat sets the real timetable: for active campaigns already enabled for Nielsen measurement, clients must ensure the relevant CTV publisher inventory has also been activated, so the expansion widens what is measurable without retroactively measuring what was never tagged.

The Walt Disney Company and TikTok set out a two-way content arrangement that changes what a subscription streaming session can contain. The two companies announced a deal on August 5, 2026 that moves fan-made short-form video into the Disney+ app, with a United States pilot due in the coming months and no advertising terms disclosed. Two flows sit inside it. The first is a licensing flow, with TikTok giving its creators access to assets tied to hundreds of films and series across Pixar, Marvel, Star Wars and FX, drawing on a daily pool of 6.5 million film and TV posts. The second, and more unusual, is a distribution flow, with opt-in creator videos running inside Verts, the vertical feed Disney began rolling out on the Disney+ mobile app in March 2026. Subscription streaming services have historically treated their catalogues as closed, professionally produced inventory, and placing opt-in user content inside a paid app changes what a Disney+ session can contain. "The best storytellers are fans first," said Asad Ayaz, Chief Marketing and Brand Officer at The Walt Disney Company. "This collaboration creates a new bridge between the stories we tell and the creativity they inspire." Dawn Yang, Global Head of Entertainment at TikTok, described the flow from the platform's side, saying creativity "extends the life of films and shows into conversations that fans discover and share."

Neither company disclosed financial terms, revenue sharing, or advertising plans for the shared inventory, leaving open whether creator videos in the feed will carry advertising or be sellable through Disney's Real-Time Ad Exchange, and whether the ad-supported and ad-free tiers will see the same content. It is the first time TikTok videos will be shown on another platform. Research context explains the pull, with a Media.net study finding 73% of consumers watch short-form video multiple times daily and 81% view it in vertical format. A jointly run Disney Creator Ambassador Programaccompanies the content exchange, tiered to unlock rewards, visibility and career development pathways. The announcement coincided with Disney's earnings and comes as the company continues to build interactive ad infrastructure around its supply, having added BrightLine interactive units on August 4, 2026, none of which has yet been publicly connected to the Verts feed.

Also noted

• August 5, 2026: GreenBanana SEO founder priced a 238-page AI search playbook at $1.99, though Amazon lists June 26 as the publication date while the press release says July 21, and all ten reviews are five-star, a promotional launch landing where Google has argued no new framework is needed. Read on PPC Land.
• August 5, 2026: Autio gained dashboard placement in five Ford and Lincoln models, bringing 12,000 US location-based audio stories to Explorer, Expedition, Aviator and Navigator owners via Google Play, though paid connectivity gates entry and no advertising inventory has been disclosed. Read on PPC Land.
• August 5, 2026: Palm Lake topped Australia's retirement-living advertisers as category spend rose 10.4%, with Nielsen Ad Intel putting total category investment at $53.6 million for the year to June 2026 against 4.68 million Australians already retired. Read on PPC Land.
• August 5, 2026: IKEA drew 243,400 New Zealand shoppers in a single month, Nielsen found, with almost half of the retailer's six-month advertising spend landing in December and 82% of its shoppers also visiting Chemist Warehouse. Read on PPC Land.

For a wider view of the day's programmatic news, AdExchanger's August 5 daily roundup covered People Inc's continued reliance on Google Search traffic, with chief executive Neil Vogel telling investors that Google Search accounted for 21% of its traffic last quarter, down from nearly two-thirds, complementing the publisher-traffic and platform-dependency threads running through this edition.