Two forces ran through the marketing news on July 2, 2026, and they pointed in opposite directions. One was the slow machinery of European competition law, which reached its end point after eight years and confirmed a fine measured in billions. The other was the fast, still-forming economics of machine-read content, where the unit of value is being rewritten in real time. Between the Court of Justice of the European Union in one corner and Cloudflare's engineering blog in another sat a set of stories that shared a single theme: the numbers everyone had relied on were being recalculated, and the recalculation carried a cost.

Luxembourg closes the Android case and leaves the doctrine standing

The Court of Justice of the European Union on July 2, 2026 dismissed Google's appeal in full and confirmed a fine of 4.125 billion euros for anticompetitive practices tied to the Android operating system. Alphabet, Google's parent company, remains jointly and severally liable for 1.52 billion euros of that total. The judgment, delivered in Luxembourg in Case C-738/22 P, closed the last legal avenue available to the two companies in a proceeding that began when the European Commission opened its investigation into Android in April 2015. The Second Chamber, presided over by Judge Kirsty Jurimae, rejected all six grounds of appeal.

The case reaches back to a Commission decision of July 18, 2018, which identified four sets of contractual restrictions and characterised them together as a single and continuous infringement of Article 102 of the Treaty on the Functioning of the European Union, the provision that prohibits abuse of a dominant market position. The first two restrictions came from the Mobile Application Distribution Agreements, known as MADAs, under which device manufacturers that wanted access to the Play Store had to pre-install both Google Search and the Chrome browser. The Commission dated the tying of Google Search to the Play Store from January 1, 2011, with Chrome added from August 1, 2012. The third restriction came from Anti-Fragmentation Agreements, or AFAs, which stopped manufacturers selling devices that ran versions of Android Google had not approved. Amazon's Fire OS served as the practical example: six major manufacturers declined to build devices running it, citing their AFA obligations. The fourth restriction, contained in revenue share agreements, or RSAs, paid manufacturers and operators a share of advertising revenue in exchange for not pre-installing any competing search service, and ran from January 1, 2011 to March 31, 2014.

The arithmetic of the fine has a history of its own. The Commission originally calculated 4.342 billion euros. When the case reached the General Court in 2022, that court found procedural errors in the handling of the RSA element, specifically that the Commission had substantially revised the scope of its objections in letters of facts rather than in a supplementary statement of objections, depriving Google of an oral hearing on the revised objections. The General Court annulled the RSA finding and cut the fine to 4.125 billion euros, while leaving the search tie, the browser tie, and the anti-fragmentation obligations intact. The Court of Justice has now confirmed both that approach and that number.

What matters for the marketing sector is less the sum than the doctrine the ruling leaves standing. Google argued that the General Court should not have taken the RSAs into account when assessing the MADAs, given that MADA signatories were not obliged to enter into RSAs. The Court rejected the separation. Between 2011 and 2016, more than 50 percent of Android devices sold in the European Economic Area were covered by RSAs requiring Google Search as the default, which meant the theoretical option to pre-install a rival search service alongside Google Search was, in practice, unavailable on at least half of all devices. The Court also endorsed the General Court's treatment of what the judgment calls the status quo bias: once the Commission establishes that a user's choice has been distorted by mandatory pre-installation, it is not then required to prove separately that the distortion, rather than the intrinsic quality of Google's products, explained the observed behaviour.

The most consequential legal finding concerns method. Google contended that the Commission was obliged to run an as-efficient-competitor test, the analytical benchmark that asks whether an equally efficient rival could have survived the conduct at issue. According to the judgment, no such test is universally required under Article 102. In digital markets marked by high barriers to entry, network effects, multi-sided structures, and ecosystem lock-in, conduct that does not amount to competition on the merits can make entry by an equally efficient competitor practically impossible, and the absence of an explicit as-efficient-competitor test does not invalidate the abuse finding. That holding, together with the ruling that the combined effects of complementary agreements can be assessed jointly even where each agreement is not independently unlawful, hands enforcement agencies a template that extends well past Android.

The template already has somewhere to go. The Android judgment runs in parallel with a dense field of Google antitrust proceedings tracked by PPC Land. In September 2025 the Commission imposed a separate 2.95 billion euro fine on Google for abusing dominant positions in publisher ad servers and programmatic buying markets, a case still in its early stages. The Court of Justice upheld a 2.4 billion euro fine in September 2024 in the Google Shopping case, which followed the same institutional path of Commission decision, General Court review, Court of Justice appeal, and final dismissal. Outside Europe, Australia's competition regulator secured a 55 million Australian dollar penalty in December 2025 over exclusive pre-installation deals with Telstra and Optus, arrangements structurally close to the RSAs. In the United States, Google filed an appeal in May 2026 seeking to reverse Judge Mehta's August 2024 finding that it had illegally maintained a search monopoly, partly through exclusive distribution agreements worth more than 26 billion dollars in 2021 alone. The core question threading all of these, whether mandatory pre-installation, default settings, and exclusivity payments constitute abuse by a dominant firm, has now been settled at the highest EU level in a way unfavourable to Google. The remedial obligations imposed in 2018, which required Google to end the practices within 90 days, remain in place, and the fine is now beyond further appeal.

A randomized experiment gives AI Overviews a number that holds up

If the Android ruling settled an old argument, a study covered by PPC Land the same week reopened a newer one on firmer ground. A randomized field experiment involving 1,065 desktop Chrome users found that Google's AI Overviews cut outbound organic clicks by 39.8 percent and raised zero-click searches by 34.5 percent when the feature appeared, with no improvement in how users rated their search experience. The paper, written by Saharsh Agarwal of the Indian School of Business and Ananya Sen of Carnegie Mellon University's Heinz College, was first posted to the Social Science Research Network on April 3, 2026 and last revised June 17, 2026.

The method is what sets the work apart. Earlier research on AI Overviews leaned on observational comparisons, either of search traffic before and after the feature rolled out, or of queries that triggered it against queries that did not, designs that cannot rule out simultaneous shifts in user behaviour, competing products, or Google's own algorithms. This study used a custom browser extension, built on the open-source Webmunk tool developed with support from the National Bureau of Economic Research, to randomly assign real users to see or not see the summaries Google places atop many results pages. Random assignment meant the only systematic difference between groups was whether an overview that would otherwise have appeared was displayed or suppressed.

The execution was exacting. Recruitment through the platform Prolific ran from January 7 to January 27, 2026, narrowing 3,710 initial respondents to a final analytic sample of 1,065 after eligibility checks, browser-history validation, and data-integrity exclusions, which removed, among others, 19 workers on Google's own RaterHub search-quality program. Two primary groups, a control seeing the standard results page and a treatment group in which the extension hid AI Overviews, were each assigned with 36 percent probability; a third arm, redirected to Google's AI Mode conversational interface, took the remaining 28 percent. Across 396 control users and 374 hide-overview users, the researchers observed 68,089 unique searches, with AI Overviews triggered in roughly 41 percent of queries.

The core numbers are precise. According to the paper, for queries where an overview was meant to appear, removing it lifted outbound organic clicks from 0.37 to 0.62 per search and cut the probability of a zero-click search from 0.73 to 0.54. Sponsored clicks did not move, and neither did clicks within Google's own properties, a pattern the authors read as AI Overviews substituting specifically for organic website visits rather than reshuffling attention across the page. A placebo check reinforced the design: for queries where no overview was triggered, clicks did not differ between groups. The effect concentrated where the feature sits, since more than 87 percent of overview appearances placed the summary in the very top position, and removing it from that slot produced an 88 percent relative increase in outbound clicks against no measurable effect when the overview sat lower on the page. Informational queries, 71 percent of the sample, drove the result with 0.26 additional clicks per search, while navigational and transactional queries produced noisier estimates not distinguishable from zero.

The quality findings cut against a claim Google has made repeatedly. The company's Vice President of Product for Search, Liz Reid, characterised AI Overview clicks in July 2025 as higher-quality clicks showing greater purchase intent and longer engagement. The experiment tested that argument on three downstream measures, the probability of navigating back via the browser's back button, a bounce measure defined as a session under ten seconds with no further interaction, and total time on page, and none differed meaningfully between groups. The additional clicks generated when overviews were hidden looked, by those measures, just as engaged as the clicks that already existed. An endline survey completed by 90 percent of active participants found no difference in overall satisfaction, perceived information quality, or ease of finding information, results the authors describe as precisely estimated nulls. The AI Mode arm, treated as exploratory because of high attrition, pointed the same way more steeply, with satisfaction falling 1.14 points and compliant participants recording 0.36 external clicks per search against 0.53 in the control group.

The regulatory audience for these figures is not theoretical. The paper lands while the European Commission's formal antitrust investigation into Google's AI content practices, opened December 9, 2025, examines whether the company used publisher and YouTube content for AI purposes without appropriate compensation or a workable opt-out. The UK Competition and Markets Authority has proposed publisher opt-out rights for AI summaries, and in the United States Penske Media's federal antitrust lawsuit, filed September 12, 2025, alleges Google coerces publishers into supplying content for AI systems while cutting the traffic those publishers depend on. Randomization is a different standard of evidence than the correlational work that shaped the debate so far, and it arrives exactly as regulators in several jurisdictions weigh whether opt-out or compensation mechanisms should be mandatory rather than voluntary.

Cloudflare moves the meter from the crawl to the citation

That question of voluntary versus mandatory frames the third story, which is the market's own attempt to build a compensation mechanism before regulators impose one. Twelve months separate two Cloudflare announcements, and the distance between them measures how fast the economics of machine-read content have moved. On July 1, 2025, Cloudflare told AI companies they would have to pay each time a crawler fetched a page. On July 1, 2026, the same company declared that model insufficient and proposed paying publishers only when their content actually appears inside an answer, a shift PPC Land covered on July 2.

The argument rests on a specific figure. According to Cloudflare, more than half of the crawl traffic generated by bots it classifies as legitimate goes toward re-fetching pages that have not changed since the last visit, and that share is likely to grow as crawl volumes rise. The waste runs both ways: AI companies burn compute retrieving stale content, and publishers pay hosting costs to serve pages that produce no new information. Cloudflare, which says it sits behind more than 20 percent of websites on its network, casts that footprint as the reason it can address the problem, since no single publisher or AI company observes traffic at comparable scale. The centerpiece is a research program that pairs voluntary publisher signals about genuine content changes with Cloudflare's own observation of human and bot traffic, measuring both whether the combined signals help answer engines surface fresher content and how much unnecessary crawling they eliminate. Broad availability is planned for later in 2026, without a specific date.

The commercial half of the announcement runs through two named partners with two different mechanics. Ceramic.ai, led by founder Anna Patterson, operates a pay-per-query model, under which opted-in publishers are paid when their content appears in Ceramic's search results rather than each time a crawler fetches a page, and participating publishers gain query-level reporting on the exact snippet shown and its average ranking position. The second partner, You.com, lets AI agents pay on demand for a specific piece of premium content at the moment it is needed. No pricing was disclosed for either. The reframing is deliberate, since Pay Per Crawl, launched in private beta on July 1, 2025, let content owners charge crawlers using HTTP 402 Payment Required responses, and the company now treats that entire apparatus as a first step. A single page might be crawled once and cited in thousands of answers, or crawled repeatedly and never cited at all, so the unit of compensation moves from the crawl to the outcome it produces.

A second policy change published the same day extends the logic to defaults. From September 15, 2026, crawlers Cloudflare classifies as Training and Agent will be blocked by default on pages that display ads for new domains onboarding to the network, while Search remains allowed by default. An ad, the company reasoned, signals that a site owner meant a person to land on the page, so on those pages human attention is treated as the goal. The pressure behind both moves is a documented decline. Cloudflare's post cites 2025 Pew Research Center data finding that when Google shows an AI summary, users click a traditional result just 8 percent of the time, roughly half the rate seen without a summary, figures Pew published in July 2025 and Google disputed at the time. Subsequent measurements moved one way. Ahrefs put the click reduction at 34.5 percent in April 2025 and 58 percent in a February 2026 follow-up, while Seer Interactive reported that organic click-through rates on AI Overview queries fell 61 percent between June 2024 and September 2025, with paid rates down 68 percent over the same window.

What the announcement avoids claiming deserves equal weight. Cloudflare describes the whole effort as an experiment, states there is much to learn about how it holds up at internet scale, and positions itself as an infrastructure layer providing plumbing for payment rather than a party setting prices. That distinction separates July 1, 2026 from July 1, 2025, when its dashboard let publishers set a domain-wide price per crawl directly. Whether a citation is worth more or less than a crawl remains, for now, a question the market has not priced.

The recalibration reached individual campaign settings as well. Google Ads is preparing to change how its bidding systems handle campaigns that have spent months, sometimes years, converting well below the cost target their advertisers set. Starting August 17, 2026, and continuing over several weeks, budget-limited campaigns running Target CPA or Target ROAS bidding will begin delivering closer to the figures advertisers typed into their account settings. Google sent notification emails to affected advertisers on July 2 and published updated help documentation, detailed in a recorded conversation between Greg Finn of Cypress North and Barry Schwartz of Search Engine Roundtable published in the "It's New" daily video series that day.

Finn described the arithmetic in round numbers. A campaign with a target CPA of 10 dollars that has actually been bringing in acquisitions at 5 dollars, and has done so for perhaps 18 months, may see that 5 dollar figure move up toward 10. The update applies to campaigns across Search, Shopping, Performance Max, Demand Gen, Travel, and Display that carry a "Limited by budget" status while running Target CPA or Target ROAS. App campaigns, Video reach campaigns, and Video view campaigns are excluded, and Hotel and Display campaigns already operate under the new logic. Google's Help Center documentation states that budget-limited campaigns using a target-based bid strategy will more consistently perform toward the bid target, including when advertisers make budget adjustments. Under the prior system, PPC Land's account of the mechanics explains, such a campaign could deliver conversions substantially below its stated target and simply keep the gap as apparent efficiency. Finn offered a sharper reading, describing the update as arriving under the guise of predictability and predicting higher cost-per-click, on the logic that acquisition costs cannot double without upward pressure on clicks.

The rollout comes with a tool and a window. A Bid Target Adjustment Tool becomes available inside Google Ads on July 6, 2026, triggered by account-level notifications sent to advertisers whose campaigns carried a "Limited by budget" status over the prior 12 months while running an affected strategy. Inside it, advertisers can keep the existing target and let delivery drift toward it, lower the target to match recent performance, or set a new one, while switching toward Maximize Conversions or Maximize Conversion Value removes the target constraint altogether. The gap between July 6 and August 17 amounts to roughly six weeks, and nothing in the rollout adjusts targets automatically. Layered on top sits a cosmetic relabeling that began earlier in June 2026, in which "Maximize conversions with a Target CPA" became simply "Target CPA," a naming change PPC Land covered separately whose arrival in the same month as the functional shift creates real risk that advertisers conflate the two.

The same recording gathered a cluster of smaller Google and Microsoft changes. Google Search will now route users directly to the publisher-hosted version of an Accelerated Mobile Pages result rather than through a Google-hosted cached copy, reversing a caching model that had long frustrated publishers by limiting analytics, tracking, and monetization on those pages, though early testing surfaced a page-load error. On the Microsoft side, SEO consultant Glenn Gabe noticed on June 1, 2026 that Bing Webmaster Tools AI performance reports across many profiles showed a marked jump in citation and impression data, which a Microsoft representative characterised as routine backfilling rather than any real change. Continuous backfilling means a spike in reported citations cannot be read reliably as improved visibility without first ruling out a data-processing explanation, a caution sharpened by the timing, since Fabrice Canel, the Microsoft principal product manager credited with shaping Bing's crawler infrastructure over nearly three decades, retired on July 1, 2026, one day before the recording, according to Search Engine Roundtable.

OpenAI drafts the formats while the content it monetizes is still being priced

The buyer side of the same economy showed its own construction site. Three job listings on OpenAI's careers page point to text, image, video, native, conversational, and interactive ad formats in development, Digiday reported on July 1, 2026. The company is seeking an ad formats software engineer with at least seven years of experience for what the listing calls a foundational role across the full stack, sitting within the monetization team and responsible for building the infrastructure that defines how ads are structured, rendered, and delivered across surfaces. Two further roles focus on iOS and Android experiences.

The listings arrive four months into the commercial life of OpenAI's ad business and mark a shift from distribution to product. The pilot formally began on February 9, 2026 with a closed roster of large brands and holding companies including Omnicom, WPP, Publicis, and Dentsu. A self-serve Ads Manager opened in beta to US advertisers on May 5, 2026, alongside cost-per-click bidding with recommended starting bids between 3 and 5 dollars per click, a Conversions API, and pixel-based measurement, while the original 60 dollar cost-per-mille model stayed live in parallel. The distribution layer consolidated just as fast, with ad tech companies lining up behind OpenAI, and by an AdExchanger daily roundup on July 1, chief revenue officer Denise Dresser had stated plainly that OpenAI is in the advertising business, a formulation the company would have avoided a year earlier. The financial logic is public: an internal advertising revenue target of 2.4 billion dollars for 2026 against an estimated 14 billion dollars in projected losses for the year.

What the listings reveal, beyond intent, is sequencing. Text ads with a label already run, while conversational and interactive formats, in which the ad becomes part of the dialogue rather than an insert beside it, remain unbuilt, which suggests the format layer is being designed before the demand it will carry has fully arrived. That ordering mirrors the platform playbooks of the prior two decades, with one difference the Cloudflare story makes concrete: the inventory OpenAI monetizes is generated from content whose owners are, this same week, negotiating what a citation is worth. The connection is not incidental. A market that cannot yet agree on the value of a machine-read page is simultaneously building the ad formats that will run on top of pages read exactly that way.

The plumbing underneath: adoption without consumption, and duplication at 46 percent

Two further July 2 items describe the state of the infrastructure the machines are now entering, and both turn on the same disconnect between a count and its meaning. The first concerns a file that site owners keep adopting and AI systems keep ignoring. Originality.ai's June 2026 update to a year-long tracking study found that llms.txt adoption grew 8.8 times over twelve months, from 4,088 instances in June 2025 to 36,120 by May 2026, reaching a combined 38,980 sites across the llms.txt, llms-full.txt, and ai.txt formats. The file, proposed by data scientist Jeremy Howard in 2024, is a plain-text markdown document placed at a site's root that gives AI models a structured summary of the site's purpose and most important pages, on the reasoning that language models have limited context windows and struggle to process most websites in full.

Adoption growth has not translated into use. A separate Ahrefs study, published in June 2026 and drawing on server-log data from 137,000 domains, found that 97 percent of llms.txt files received zero requests in May 2026. According to Ahrefs, even where AI retrieval bots were active, they accounted for just 1.1 percent of requests, with SEO audit tools leading at 21.7 percent, unidentified bots at 14.9 percent, and general web crawlers at 13.1 percent. The assistants the file was designed to serve were largely absent from the logs. The major platforms send conflicting signals, since each advises site owners to rely on robots.txt for crawler management while each publishes its own llms.txt for its developer documentation. Google's guidance states the file will not affect visibility or rankings, a position PPC Land covered in detail when the AI search guide appeared on May 15, 2026, even as Chrome Lighthouse added an optional agentic browsing audit for llms.txt. Originality.ai concludes the file looks far more like agent-readiness infrastructure than a search visibility tool, useful for sites with substantial documentation or APIs that autonomous agents might navigate, and not the mechanism for general AI search visibility.

The second infrastructure item quantifies a structural fault in the auction machinery. Established supply-side platforms carry a 46 percent duplicated-domain rate, meaning nearly half of the publisher domains they connect to can be reached through more than one supply path at once, per a Sellers Report that DataBeat, a programmatic market intelligence unit under MediaMint, published on June 11, 2026 and PPC Land covered on July 1. The mechanism is near-mechanical: smaller SSPs list an established SSP as a reseller partner on a publisher's ads.txt file to reach that platform's demand, and when the established SSP already holds a direct relationship with the same publisher, its inventory becomes reachable through two declared paths, both able to send a bid request for the identical impression into the same auction. Because the requests originate from different ads.txt paths, downstream systems have no inherent way to recognize them as the same inventory. A companion DataBeat report, its June US Programmatic Trends Report published June 22, 2026, found conventional programmatic buyers clearing at an average CPM of 6.95 dollars against 6.13 dollars for agentic buyers, a 13.4 percent premium for conventional demand according to the report, while agentic buying led on fill rate at 0.204 percent versus 0.183 percent, achieved in what the report describes as 86 percent fewer auctions. Read together, the two reports describe a marketplace onboarding autonomous demand into pipes that still bill the same impression twice.

What July 2 adds up to

The stories share a single structural feature: each replaces an old proxy with a measured outcome, and the measurement carries a bill. Luxembourg confirmed that default settings and exclusivity payments are abuse when a dominant firm deploys them, closing a proxy Google had defended for eight years. The Agarwal and Sen experiment replaced observational correlation with randomized causation because the policy stakes had outgrown the older evidence. Cloudflare abandoned the crawl for the citation because a crawl no longer predicts value. Google Ads narrowed the gap between a stated CPA target and actual delivery that advertisers had banked as efficiency. OpenAI began drafting a native taxonomy of six ad formats rather than borrowing the assumption that AI ads look like search ads. The llms.txt and DataBeat findings exposed the distance between a count, whether of adopted files or of bid requests, and the thing the count was supposed to represent. Markets reprice when their measurements change, and across a single day the marketing economy changed several of the numbers it lives by.

Also noted

  • July 2, 2026: The Court of Justice confirmed the 4.1 billion euro Android fine in Case C-738/22 P, with Alphabet jointly liable for 1.52 billion euros, ending eight years of litigation, PPC Land reported.
  • July 2, 2026: The Trade Desk signed Amsterdam-based Massarius as its first Dutch publisher on OpenPath, giving advertisers direct access to display inventory and bypassing several intermediary layers.
  • July 2, 2026: LinkedIn data showed advertisers running five or more ad variants gained 20 percent higher click-through rates than single-ad campaigns on Campaign Manager, alongside new Brand Kit and drafting tools aimed at smaller teams.
  • July 1, 2026: The EU abolished the 150 euro customs duty exemption for low-value parcels, replacing it with a flat 3 euro charge per item rather than per parcel, after 5.9 billion duty-free items entered the bloc from third countries in 2025.
  • July 1, 2026: Pew Research Center found almost six in ten US adults support banning social media access for people under 16, based on a survey of 9,750 adults conducted May 26 to June 1, MediaPost reported.