1 OpenAI ships ChatGPT Ads Manager beta — CPC bidding, $100M in 6 weeks
OpenAI rolled out a beta self-serve Ads Manager on May 5, with CPC bidding and a new conversion-measurement suite. Launch partners include Dentsu, Omnicom, Publicis, and WPP as agency early-access partners, plus Adobe, Criteo, Kargo, Pacvue, and StackAdapt on the ad-tech side. The pilot — running on US Free and ChatGPT Go tiers — has already generated $100M in revenue in its first six weeks.
This is the first time a chatbot company has shipped a real self-serve ads platform. The Big 4 holdcos are launch partners, not bystanders — they’ve already integrated. If you’re not budgeting test spend on ChatGPT by Q3, you’re going to be behind on data and CPCs for a search-adjacent layer that’s growing 30%+ month over month.
2 Meta Ads AI Connectors open globally — external AI can now manage your Meta campaigns
Meta launched AI Connectors in open beta to all eligible advertisers globally. The connector enables a secure, direct link between a Meta ad account and external AI tools that support Model Context Protocol (MCP) — ChatGPT and Claude at launch, more on the way. Cross-channel insights and full campaign management — without ever opening Ads Manager.
Meta just stopped requiring you to log in to its own UI. Whoever owns the AI assistant your team types into now owns the workflow — and Meta is the first major platform willing to plug straight in. Same pattern Amazon and Google are racing toward. The agencies who build prompts and playbooks for these connectors first will run leaner teams against bigger books.
3 Amazon stands up a dedicated agentic commerce team — to let ChatGPT plug into checkout
A LinkedIn job listing surfaced on May 8 (spotted by Juozas Kaziukenas) revealing Amazon’s first dedicated agentic commerce team. The mandate: build infrastructure for external AI platforms — ChatGPT, Claude, others — to connect with Amazon services in a controlled, API-mediated way. The move parallels Google + Shopify’s Universal Commerce Protocol and OpenAI’s Agentic Commerce Protocol (with PayPal and Stripe as early backers).
Amazon spent a decade fighting search outside its walled garden. Now it’s building plumbing so chatbots can fulfill orders inside it. McKinsey projects agentic commerce will drive $3–5 trillion globally by 2030. Every product detail page is about to compete for AI-agent visibility, not just SERP rank — which means your PDP copy, structured data, and review hygiene now feed bots, not just buyers.
4 YouTube ships view-through conversion bidding for Demand Gen — measurement gap closed
YouTube enabled Demand Gen campaigns to bid for users who watched an ad and converted later — without clicking. The long-standing measurement gap with Meta and TikTok — which both already optimize for view-through — is now closed.
For three years YouTube has under-credited itself in every multi-touch model. Demand Gen results were structurally underperformed by 20–40% vs. Meta in mixed-channel attribution. With view-through bidding live, expect Demand Gen ROAS reports to jump — and Google Marketing Live on May 20 to lean on this number hard. If you’ve been writing off YouTube’s lower funnel, your media plan needs a fresh look this week.
5 Meta Engage-through attribution replaces engaged-view — your conversion math just changed
Meta replaced engaged-view with a new category, Engage-through attribution. Conversions count when someone likes, comments, shares, saves, or watches a video for 5+ seconds and converts within 1 day. Click-through now strictly requires an actual link click — likes and saves no longer qualify. A new Maximize Interactions performance goal replaces Post Engagement.
If your Meta dashboard suddenly looks different this week, this is why. Engage-through inflates reported conversions compared to the old strict click-only counting; the tightened click-through definition deflates it. The numbers are not directly comparable to last quarter — re-baseline your reporting before your CFO opens the dashboard and asks why “real” performance changed overnight.
6 Meta launches Instants — Instagram’s ephemeral-photo app goes after BeReal
On May 13, Meta launched Instants — a new iPhone app plus Instagram feature for ephemeral photo sharing. Instants disappear for viewers but are saved in the user’s archive for up to a year, with a recap-to-Stories option. Friends can react and reply, and replies route straight to DMs instead of public comments.
BeReal’s “one daily prompt” moat keeps shrinking as Meta surrounds it. For brands: Stories were already a private-feeling channel; Instants doubles down on that. The bigger play is private-mode social — the place ads and brand-built community become harder to plant. If your IG strategy still ends at the feed, you’re working off a 2023 playbook.
7 Meta opens Reels Trending Ads to all advertisers — brand lift up to +20%
Meta announced that Trending Ads on Reels are now available to all advertisers globally — letting brands place ads directly alongside the hottest, culturally relevant Reels of the moment. Early tests show brand awareness lift up to 20%.
This is Meta’s answer to TikTok’s “ride the trend” buys. For brands without an in-house creator team, you can now pay for proximity to trending content instead of trying to manufacture it. Pricing will be premium and inventory will be moment-dependent — test small, expand only when paired with a brand lift study. Don’t let your media team default to this for performance metrics; it’s an awareness buy in performance clothing.
8 Google AI Mode shows inline links + subscription-labeled sources
Google updated AI Mode and AI Overviews to place links directly next to the relevant generated text inside AI responses — not grouped at the bottom of the answer. On desktop, hovering over an inline link now previews the site and page title. Subscription content is explicitly labeled.
Inline links are Google’s answer to a year of publishers screaming about zero-click search. Click-through from AI answers should jump for content that’s tightly cited in-line — meaning short, fact-rich passages with clear data points get cited; vague analyses don’t. The subscription labels are a quieter shift: Google is signaling which content commands a paywall premium and which is “free web,” shaping both reader trust and future licensing leverage.
9 Snap Place Loyalty goes global — gold/silver/bronze badges on Snap Map
Snap rolled out Place Loyalty globally. Users in the top 1% of visitors to a location get a gold badge, top 10% silver, top 25% bronze. For brands and chains, Snap aggregates visits across all locations. Shareable stickers, ranking screens, badge unlocks.
This is loyalty without an app — running on phone location data Snap already has. For QSR, retail, gyms, coffee chains — a free top-of-funnel loyalty layer with social bragging built in. The bigger picture: Snap Map is quietly becoming the “where do people actually go” dataset at consumer scale, and that’s monetizable at the brand-and-roofline level, not just the ad unit. The first three brands to sponsor a “Top Visitor in X City” cohort will own the playbook.
10 GrowthLoop’s 2026 AI and Marketing Performance Index — data quality is the real blocker
GrowthLoop released its 2026 AI and Marketing Performance Index on May 13, surveying 300+ marketers and data leaders across the US and Canada. Headline finding: data issues significantly slow marketing cycles, experimentation, and personalization — most teams are buying AI tools faster than they can clean the data feeding them.
AI doesn’t fix bad data — it surfaces it. Every agentic ad workflow above (Meta Connectors, ChatGPT Ads Manager, YouTube view-through bidding) is leverage that multiplies the quality of your inputs. If your audience taxonomy is held together with VLOOKUPs and tribal knowledge, agentic tools will make your reporting worse, not better. First-party data hygiene is the prerequisite, not next quarter’s project.