July 2026
Industry News
Here's what happened this month, why it matters, and what your team should know.
The Rundown
- ›AI search became a paid channel. OpenAI opened self-serve ChatGPT Ads on July 22 and added conversion-optimized (oCPC) bidding, geo exclusions, and bulk APIs two days later. Ads show only to free and lower-tier users, across seven markets. The shift: brands now have to both earn organic citations in AI answers and buy placement alongside them.
- ›The frontier-model race kept sprinting. OpenAI released GPT-5.6 publicly on July 9 (Luna, Terra, and Sol tiers, plus ChatGPT Work), Anthropic shipped Claude Opus 5 on July 24 at the same price as Opus 4.8 with a 1M-token context window, and Google pushed Gemini 3.5 Flash-Lite into Search for agentic tasks. Build for model portability, because capability and availability keep shifting.
- ›Platforms drew a hard line on AI "slop." LinkedIn added an "AI slop" report button and pulled its own AI post-writer on July 30, YouTube clarified that generic and AI-persona content is ineligible for ads, and Google's summer ranking volatility hit AI content farms hardest. Original, human-credible work is the moat.
- ›Measurement matured. Google Search Console added platform properties (Instagram, TikTok, X, and YouTube in Search data), globally available July 29, and Google's rewritten Ads Terms of Service formalized AI automation authority on July 1. The job now is measuring AI's real impact and keeping human oversight on automated systems.
July 2026 Timeline
SEO Updates
Google added platform properties to Search Console, letting you connect Instagram, TikTok, X, and YouTube to see how those posts perform in Search, Discover, and News.
It is the first property type you can verify without owning the domain, which finally brings owned-social visibility into the same reporting surface as your site. It became globally available July 29. Start a baseline now to see how your off-site content surfaces in Search.
More than 14 trackers, including Semrush Sensor, Mozcast, and Sistrix, flagged sharp swings around July 18-19, then again July 23-24. Google did not confirm a core or spam update.
The steady-hand takeaway: track trend lines over two to three weeks rather than reacting to a single day. Summer volatility is common, and unconfirmed movement rarely calls for emergency changes.
On July 16, Google added connected apps that let Search complete tasks, from building an Instacart cart to opening Canva templates or queuing YouTube Music.
On July 22, Google also launched AI Overviews and AI Mode in France, with publisher opt-out and separate measurement. AI search is becoming a place where actions happen, not just answers, so make sure your feed and structured data are ready to be acted on.
Google added new Merchant Listing properties: a Category field based on the Product Taxonomy, plus validFrom and validThrough for sale duration.
Together they let merchants mark up product category and promo pricing directly on the page, giving Google clearer signals for shopping surfaces. If you run sales or a broad catalog, it is a low-effort structured-data win worth implementing.
AI & Search
On July 22, OpenAI launched an "Advertise in ChatGPT" Ads Manager that targets on conversational context across seven markets, with ads shown only to free and lower-tier users.
Two days later, OpenAI added conversion-optimized (oCPC) bidding, geo exclusions, and bulk APIs. AI search is now a two-front effort: earn organic citations in AI answers, and decide whether to buy placement alongside them. A small, measured test budget is the sensible way to learn the channel early.
On July 9, OpenAI released GPT-5.6 publicly, with Luna, Terra, and Sol tiers aimed at enterprise, coding, and research, and Sol powering the new ChatGPT Work.
The public release had been delayed under US export and access restrictions. For teams standardizing on a model, the takeaway holds: build for portability, since availability now shifts with policy as well as roadmaps.
On July 24, Anthropic launched Claude Opus 5 with a 1M-token context window at the same price as Opus 4.8, and said its classifiers trigger far less often.
The pricing and context combination makes large-context work cheaper, such as analyzing full brand guidelines, campaign histories, or long research sets in a single pass. For marketing teams, that lowers the cost of grounding AI work in your own material.
On July 21, Google rolled its fastest and cheapest 3.5-class model, Gemini 3.5 Flash-Lite, into Search to power agentic workflows.
A day later, AI Mode expanded to France. Faster, cheaper models make agentic search cheaper to run at scale, which points to more tasks completed inside the results page rather than on your site.
Perceived helpfulness of AI search fell from 82% in 2025 to 54% in 2026, and self-described skeptics grew three points to 17%, even as usage keeps climbing.
Meanwhile, Perplexity's ad-free bet pushed annualized revenue past $450M. The pattern rhymes with the rest of this issue: adoption is rising, but trust is splitting, so disclosure and credibility matter more, not less.
Paid Media
On July 1, Google's first substantive rewrite of its Ads Terms of Service since 2018 took effect, explicitly authorizing automated systems to format, select, or generate targets, ads, and destinations.
Advertisers keep responsibility for review. The practical move is to tighten the inputs you can control, asset groups, brand exclusions, and negative keywords, and to review AI-driven campaigns regularly rather than setting and forgetting.
On July 29, Meta reported $60.8B in total revenue and $59.36B in ad revenue, up 27% year over year, with impressions up 14% and price per ad up 12%.
Rising price per ad points to tightening inventory. Watch cost trends into Q4, and lean on creative diversity and efficient audiences to keep CPMs manageable as competition for impressions climbs.
Alphabet's Q2 report on July 22 put YouTube ad revenue at $11.1B, up about 13% year over year and ahead of estimates.
Live sports were a key driver, with the World Cup alone drawing more than 1.7 billion views. YouTube's pull on connected TV and live events makes it a strong place to reach lean-back audiences at scale this year.
In its fiscal Q4, the quarter ended June 30 and reported July 29, Microsoft's search and news advertising revenue ex-TAC grew 10%, a deceleration from prior quarters.
The growth came from higher revenue per search plus volume, partly offset by changes in partnership terms. Microsoft search remains a cost-efficient complement to Google, but the slowdown is a reminder to judge the channel on incremental value rather than assumed momentum.
As Advantage+ becomes the default, the Andromeda retrieval engine underneath it is why creative demand keeps climbing. Because it can evaluate far more ad candidates per impression than the old six-ad playbook, the payoff now scales with creative volume: a July 2026 analysis found brands testing 20 or more new ads a month see about 65% higher ROAS than those testing fewer than 10, and top-third advertisers keep roughly 395 live ads running versus 296 for the bottom third.
Volume only works with diversity. Andromeda clusters near-identical ads into a single entity, so fifty look-alikes compete as one. Practitioners now aim for 8 to 15 conceptually distinct concepts per campaign, and consolidation wins: one ad set with 25 diverse creatives outperformed five narrow ad sets by 17% on conversions at 16% lower cost, while Advantage+ creative has been tied to a 22% ROAS lift. The takeaway for teams: build a pipeline for distinct concepts, not more renders of the same idea.
Reddit Roundup
The July 1 Ads Terms of Service rewrite lit up the paid threads. The debate centered on ceding more control to automation now that Google can format, select, or generate ads on your behalf. The practical consensus: you still own the inputs, so tighten asset-group content and negative keywords and review AI Max campaigns closely rather than trusting defaults. Source
The July 18-19 volatility drove a familiar thread: sharp movement across trackers, no confirmation from Google. The community consensus landed on patience, watch trend lines across two to three weeks and resist rewriting a strategy over a single volatile day. Source
Marketers shared data showing ChatGPT's share of AI web referrals sliding from roughly 76% to about 53%, while Gemini and Claude picked up ground. The takeaway: AI visibility is no longer a single-assistant game, so build cross-platform measurement and earn citations across the assistants your audience actually uses. Source: cross-platform AI-referral roundup
July trade discussion reframed deliverability basics as table stakes: Gmail and Yahoo now enforce SPF, DKIM, and DMARC plus one-click unsubscribe, and non-compliant sends get filtered regardless of content quality. The message to senders: authentication and list hygiene are the floor, not an optimization you get to later. Source: MarketScale
Creators reported Instagram scanning for duplicate and near-duplicate content and expanding its "Your Algorithm" controls into the main feed. The upshot for repost-heavy strategies: plan original edits per placement rather than posting the same asset everywhere, and treat native variation as part of the workflow. Source: social-platform roundup
Creative & Brand
On July 15, Canva launched Code 2.0, turning natural-language prompts into editable websites and apps inside Canva, with about 75% faster code generation and 50-plus templates.
It rolled out to all 265M-plus users, including free accounts. The gap between idea and shippable asset keeps shrinking, which raises the premium on taste, direction, and brand consistency over production time.
On July 7, Figma added parallel AI image edits from the toolbar, letting you generate multiple variations at once.
On July 9, it made GPT-5.6 selectable in Figma Make across plans. Together the updates speed up batch creative variations, useful for testing multiple directions before committing to a single hero.
On July 16, Google Vids added Gemini Omni AI avatars and conversational editing, with SynthID watermarking on generated output.
On July 24, Midjourney set V8.2 as its default image model. Both land as a useful backdrop while Sora's consumer app wound down, giving teams more reliable production options and clearer provenance signals.
On July 1, Sporting CP revealed its first identity overhaul in 25 years for its 120th anniversary, a simplified badge drawing on the 1945 crest and tuned for digital.
It is a clean example for legacy brands weighing a careful refresh over a full rebuild: honor the heritage, sharpen it for modern surfaces, and keep the equity recognizable.
Social Media Updates
On July 21, Instagram launched Replace Audio, letting creators swap the music or sound on live feed posts and carousels without losing existing engagement or reach.
It is a small but useful lever: refresh an evergreen post with trending audio, or fix a licensing issue, without starting the post's performance over from zero.
On July 24, YouTube added custom thumbnails to Shorts, desktop first, giving creators more control over the first impression in feeds and on channel pages.
Separately, YouTube clarified in mid-July that generic, distressing, or AI-persona content is ineligible for ads. The two moves point the same way: invest in distinct, human-credible creative if you want reach and monetization.
On July 13, X reworked its algorithm to boost posts from accounts you mutually follow within reply sections.
The change rewards reciprocal relationships over broadcast reach in conversations. For brands, that favors genuine, niche community engagement, replying and building mutual follows, over chasing viral one-offs.
On July 30, LinkedIn added a button to report AI-generated "slop," with flagged posts seeing reduced reach, and replaced its own AI post-writer with a lighter proofreader.
It is a clear signal that generic AI output is a reach liability on the platform. Prioritize a distinct human voice and point of view, and use AI to sharpen rather than to generate wholesale.
On July 2, Reddit added native ad Split Testing in Ads Manager, declaring a winner at 65% confidence, with a $1,000 per day minimum.
Built-in experimentation makes it easier to learn what resonates with Reddit's community-driven audiences. If Reddit is in your mix, use the tool to test creative and messaging rather than guessing.
Email & Deliverability
Disclosed July 1 and patched by July 21, Apple's Hide My Email flaw could leak a user's real address via bounce messages.
The fix is on Apple's side, but it is a timely reminder to audit how your program handles bounces and rejections, and to keep list hygiene tight so you are not exposing or retaining data you should not.
In July, HubSpot began showing predicted engagement (High, Medium, or Low) on segments before send, and brought Brand Kit custom fonts, with fallbacks, into the drag-and-drop editor.
Both are available across Marketing Hub tiers. Predicted engagement can help prioritize your most responsive segments, while brand fonts tighten visual consistency, just remember to test fallback rendering across major clients.
On July 9, Infobip acquired SocketLabs, continuing the consolidation among email infrastructure and deliverability vendors.
Consolidation can mean fewer, more integrated options over time, but it is worth watching for tool overlap and roadmap changes if either vendor sits in your stack.
On July 28, Certera began issuing VMC and CMC certificates, with the CMC requiring no trademark, which lowers the barrier to displaying your logo in the inbox via BIMI.
It arrives as July trade coverage treats Gmail and Yahoo authentication as non-negotiable. With SPF, DKIM, and DMARC in place, BIMI is now a more accessible way to add brand recognition and trust to the inbox.
Campaign Spotlight + Cool Stat
July Campaign Spotlight
Not an official FIFA sponsor, Nike built one of the tournament's biggest brand moments. It seeded 42 raw, Polaroid-style teasers on social, then released a six-minute film featuring more than 30 athletes on a set that literally falls apart.
It became the most-shared social post in Nike's history and captured 52.4% of all World Cup brand-endorsement conversation, outpacing the official sponsors. The lesson: athlete-led storytelling and social-first sequencing can out-compete paid sponsorship real estate.
DASH made the biggest media investment in the brand's history, spanning TV, digital, out-of-home, and radio, and reached more than 203 million UK impacts, about 1 in 7 UK adults seeing it roughly 25 times.
The TV spot scored in the top 6% of ads tested. It is a reminder that a bold, well-produced brand push across channels can still drive standout recognition, even in an AI-saturated media landscape.
A ChatGPT World Cup stunt, "#MessiMode," let fans recolor Messi's hair in their national colors. It drew criticism after fans noted Messi had previously said he does not use AI tools.
The lesson: a clever viral mechanic still has to hold up against the spokesperson's own credibility. Match the activation to what the talent has actually endorsed, or the mismatch becomes the story.