Everything that used to happen around the campaign is moving inside it. Economic impact studies that arrived months late now run mid-flight. Ad creative that took a production cycle is now an API call any AI agent can make. A whole standalone ad product — Local Services Ads — is being folded into an automated campaign type, account by account, starting this week. Meanwhile the platforms funding all this rewiring — Snap, Apple — posted record quarters. The through-line: the stack no longer pauses for your review cycle. It generates, measures, and reallocates continuously, and the human role is compressing into a single checkpoint — approval. The marketers who win from here design that checkpoint deliberately instead of inheriting it.
1 Google folds Local Services Ads into Google Ads
The facts: Google confirmed that Local Services Ads will migrate into Google Ads as Performance Max pay-per-lead campaigns, starting with a small group of US advertisers in early August — plumbing, HVAC, electrical, appliance repair, house cleaning, lawn care, roofing, pest control, and moving first — then expanding through late 2026 and wrapping up with non-US accounts in 2027. Advertisers still pay per qualified lead and ads still appear on Search and Maps; targeting, budgets, schedules, and photos transfer automatically.
So what: Another standalone ad product just got retired into an automated campaign type — LSA follows Dynamic Search Ads (folding into AI Max by February 2027) and Microsoft’s Predictive Matching. If you manage local budgets, document your current LSA settings before your account migrates: bidding, reporting, and Business Profile management all change in the move, and “seamless” migrations have a way of resetting defaults you didn’t know you’d chosen.
2 Snap’s turnaround showed up in the numbers
The facts: Snap’s Q2 revenue hit $1.6 billion, up 19% year over year — advertising grew 9% to $1.28 billion while “other revenue,” mostly Snapchat+ subscriptions, surged 85% to $316 million. Adjusted EBITDA jumped to $250 million from $42 million a year ago, free cash flow was positive for the eighth straight quarter, daily actives reached 493 million, and the stock popped roughly 10% after hours.
So what: The diversification story deserves as much attention as the beat: subscriptions grew nine times faster than ads and now fund the roadmap. Evan Spiegel credited improving momentum with large North American advertisers — the cohort that quietly left in 2023–24. When a platform’s ad products are improving faster than its ad prices, that’s where underpriced reach lives. Worth re-testing before the Q4 auction heats up.
3 Apple quietly posted a record ad quarter
The facts: Apple’s fiscal Q3 delivered $109.4 billion in total revenue, with services setting a record at $30.7 billion, up 12%. Inside that, Apple called out June-quarter records in advertising, the App Store, AppleCare, Apple Music, and Apple TV+ — the advertising milestone landing just months after Apple began selling ads in Apple Maps.
So what: Apple doesn’t break out ad revenue, which makes the callout itself the signal: the company is choosing to talk about its ad business now. Between App Store search ads and fresh Maps inventory, Apple is building an owned-surface ad platform the way Amazon did — quietly, on first-party intent data, while everyone watched the bigger players. Early auctions on new inventory are usually cheap. That window doesn’t stay open.
4 StackAdapt moved economic measurement into the buy
The facts: StackAdapt launched Economic Impact Report, a self-serve measurement product that connects media exposure to more than $4 trillion in consumer card-transaction data from Affinity Solutions — and reports results while campaigns are still running. Advertisers configure it inside the buying platform in minutes; the initial release targets travel marketers, tourism boards, and brands with location-based objectives.
So what: Economic impact analysis used to be a third-party study that arrived after the budget was spent. Moving it into the buying interface turns it into a mid-flight reallocation tool — a structurally different product. One caveat before treating it as ground truth: card panels miss cash, most B2B, and anything that closes off-panel. Ask for coverage in your category first, then decide which number governs your budget.
5 Political ads got cross-channel receipts
The facts: XR (Extreme Reach) and AdImpact announced a partnership combining AdImpact’s ad intelligence — covering all 210 US DMAs and more than 41,000 ZIP codes — with XR’s MRC-accredited digital ad delivery data. Political campaigns get unified local TV and digital reach-and-frequency reporting they can act on mid-flight, and the companies say the same intelligence extends to automotive, telecom, retail, healthcare, and financial services.
So what: The timing isn’t subtle: midterm spend is ramping, and political buyers have historically flown blind across the linear-digital divide, over-frequencing the same households in one channel while missing them in another. Political adtech has a habit of becoming everyone’s adtech about a cycle later — if your category lives on local TV plus digital, this is the measurement pattern to watch.
6 ADA bought Algonomy — decisioning “stops advising and starts acting”
The facts: Singapore-based ADA completed its acquisition of Algonomy, the retail decisioning provider built from the Manthan and RichRelevance businesses. The combined company spans 34 markets across APAC, the US, MENA, and Europe, with 1,300 staff serving 1,500 clients; Algonomy brings more than 400 retail brands. ADA CEO Srinivas Gattamneni said he expects every customer touchpoint to be AI-agent driven.
So what: Algonomy CEO Atul Jalan supplied the quote of the week — decisioning “stops advising and starts acting.” Personalization, merchandising, and supply-chain intelligence now consolidate onto one CDP foundation, which is exactly the stack an autonomous agent needs to act on. If you’re a client of either company, get the written migration path now: an acquisition that closes is not an integration that ships, and you don’t want to discover which platform survives from a roadmap slide.
7 Omneky made ad creative an API call
The facts: Omneky launched a public API and MCP server for its creative generation engine. Send a URL and its agent reads the brand’s positioning, colors, and fonts, writes the headline and CTA, and returns finished ads in every major aspect ratio — including UGC-style and short-form video variants — with automatic regeneration when performance decays. The MCP server means any AI agent that speaks the protocol, including Claude, can order creative directly.
So what: Creative production just became infrastructure. Commerce platforms can white-label it for thousands of merchants; a founder can get launch ads from a chat assistant without opening a design tool. When execution is a commodity API, the billable layer moves up: brand strategy, taste, and the judgment to know which ad shouldn’t run. If your agency’s pitch is production capacity, this is your deprecation notice.
8 Solitics put agentic marketing inside bank compliance
The facts: Solitics released Saai, an agentic AI product for marketing teams in retail banking. Saai takes a plain-language objective — improve digital onboarding, reduce customer inactivity — and generates a fully configured customer journey: audience, triggers, channels, messaging, and logic, presented for review before deployment. It runs inside each bank’s own environment, within existing approval workflows and compliance controls, with integration guaranteed in 45 days.
So what: The design choice that matters is where the human sits: the approval gate lands between generation and deployment, and the generated journey — not the model — is the auditable artifact. That’s the template regulated industries have been waiting for, and it’s a preview of everyone’s workflow: the machine drafts the whole campaign; you approve or you don’t. Design that gate deliberately.
9 The holiday season already started
The facts: Attentive’s new Consumer Pulse research found 83% of US consumers begin planning holiday purchases before Black Friday, 71% plan to start shopping before the event — up 12 points year over year — and 46% start purchasing before November. Four in five say a discount of 40% or less is enough to trigger a purchase, and 90% respond to price-drop alerts on products they already want.
So what: A promotional calendar that opens in November now misses most of the planning window. Move engagement spend into September and October — that’s when the consideration set gets locked — and let the 40% figure set your margin floor, because deeper discounts are buying sales you would have captured anyway. The brands that own the wishlist in October collect the revenue in November.
10 Meta made smart glasses a content format
The facts: Instagram added Stories formats exclusive to Meta’s smart glasses: Spin View, which lets viewers rotate their phone to pan through the glasses’ full wide-angle capture, and Multi-cam, which syncs glasses POV footage with a phone camera angle in a single story. Three new editing tools — Expand for reframing, audio noise reduction, and speed controls — round out the glasses-native kit.
So what: Hardware-exclusive formats are how Meta seeds glasses-native content before the ad load arrives — the same playbook Stories and Reels ran: format first, creator adoption second, ad product third. POV content is about to get a native format and, if history holds, an algorithmic tailwind for early adopters. Brands with demonstrable products — food, beauty, crafts, fitness — should get glasses on a creator this quarter.
The pattern across all ten: the loop is closing. Creative generates itself, measurement runs mid-flight, decisioning acts instead of advising, and entire campaign types migrate without asking. None of this removes the marketer — it relocates you to the approval gate. The difference between teams that compound and teams that drown will be how well that gate is designed: what gets auto-approved, what gets escalated, and what number decides. The stack stopped waiting. Your review process should stop assuming it will.