Yesterday, Ad Age’s chief tech reporter Garett Sloane polled agencies on why they are willing to spend on ChatGPT ads on behalf of their clients even though the ad product has no track record and “limited measurement” capabilities at the outset.
Sloane previewed his article on LinkedIn:
“Why did agencies and brands jump at the chance to run ads in ChatGPT, even though the platform insisted it won’t influence their organic presence and they have no real way to measure yet? Well, Ad Age looked into it.
‘This is a monumental shift,’ [OMD Worldwide’s Chief Media Officer] Ben Hovaness tells us. ‘We don’t see a new media channel springing up every day or every year,’ he said.
This channel, ChatGPT, also happens to be the one that took 20% to 30% of brand referral and discovery behavior online in the past year. A crazy reframing of the consumer mind.
Now, even though ChatGPT says the ads are separate from organic and won’t directly affect things like GEO. Ads will be part of ChatGPT, so they are invariably a part of how brands show up there.
‘It has to at some point,’ said Dan Roberts[, Global SVP of Search] at Assembly Global … there could be ‘this halo effect or this brand recall effect, if it doesn’t convert straight away.’..”
Read more on LinkedIn — including a chart on ChatGPT ad spenders. (February 17)
Read: “Why agencies are rushing into ChatGPT ads despite limited measurement” (February 17) – Ad Age (subscription)
From tipsheet: Sloane quotes an Omnicom rep saying they will use incrementality measurement “from the jump.”
Incrementality is the flashlight inside the AI advertising “black box.”
LLMs & CHATBOTS
Developments
- Introducing Claude Sonnet 4.6 (February 17) – Anthropic
- Mistral AI buys Koyeb in first acquisition to back its cloud ambitions (February 17) – TechCrunch
- Apple Ramps Up Work on Glasses, Pendant, and Camera AirPods for AI Era (February 17) – Bloomberg (subscription)
BRANDS
Unilever partners with Google Cloud
Consumer packaged goods company Unilever announced a new five-year pact yesterday with Google Cloud in an effort to get in front of the agentic opportunity ahead for the marketer.
Marketing Dive distilled the news:
“Unilever’s five-year pact with Google demonstrates that the CPG marketer is taking the rise of agentic commerce seriously. The deal will see the sprawling company migrate its own data platforms and enterprise applications to Google Cloud as it tries to stay on the leading edge of the AI race. Linking with Google is a sign that technology ‘has moved to the core of value creation at Unilever,’ according to Chief Supply Chain and Operations Officer Willem Uijen.” (…)
The news follows Google inking agentic shopping partnerships with top U.S. retailers like Walmart and Target last month. Target separately is among the first companies to test ads in OpenAI’s ChatGPT program, including for advertisers that buy campaigns through its retail media network.”
Read: “Unilever changes its brand discovery calculus with Google Cloud AI pact” (February 17) – Marketing Dive
More:
-
“Google Cloud partnership pioneers next generation of consumer goods technologies” (February 17) – Unilever
-
Unilever Customer Hub – Google Cloud
LLMs & CHATBOTS
ChatGPT ads or Bing ads
On the most recent episode of the Digiday podcast, the reporting team took a deep dive into the latest ChatGPT ads news.
Among the insights shared by the team:
-
Echoing OpenAI’s guidance, the ChatGPT ads launch is not a product launch, it’s a “pilot.” The product is still being figured out and the pilot is one step in that process.
-
Executive editor Tim Peterson said that one agency executive asked clients in January if they would be willing to temporarily move all of their [Microsoft] Bing ad budget to ChatGPT for the pilot. Peterson continued: “And 45 clients said, yes, I would do that. And what’s interesting about — I mean — this is Bing that we’re talking about, so there’s definitely a willingness to move money away from Bing. But these are also search advertisers willing to move money towards ChatGPT. And if they do that with their Bing budgets and they see performance from it, that could open up Google budgets because they’re effectively going to be grading against the same measures.”
Listen: “ChatGPT enters the ad game. Now what?” (February 17) – The Digiday Podcast
Related: “Subscriptions are rising at big news publishers – even as traffic shrinks” (February 17) – Digiday (subscription)
From tipsheet: Ad budgets may be zero-sum for now. But in the long run, better performing ads courtesy of AI will drive overall ad spend higher. Much higher.
SELL-SIDE
Agents to inspire insertion orders
In an article titled, “Why AI May Cause a lot of Ad Tech Companies to Duck and Cover”, “Next in Media” founder Mike Shields spoke to Charles Manning, CEO of omnichannel measurement firm Kochava.
Mr. Manning saw benefits coming to publishers courtesy of agents — he thinks they’ll strip away “the ad tech tax” and create new opportunities.
Manning said that courtesy of agentic frameworks:
“Publishers may begin building exclusive inventory access at ultra-premium rates once they understand their true value. And whereas programmatic was about auctions per impression, we may start to see auctions at the IO level — like programmatic guaranteed at scale.”
Responding to the Manning quote, Mediaocean’s Ramsey McGrory, who is president of his company’s Prisma platform which pulses with insertion orders, said in the article’s comments:
“Yes! It’s called an insertion order, and it’s been used by buyers and sellers for decades to convey dates, dollars, goals and constraints. [Programmatic Guaranteed] is an IO routed through programmatic pipes, often unnecessarily. More than 1/2 of all CTV/OLV dollars transacted are still executed by IO, and it’s been consistent for several years. Look for much greater automation of this without the programmatic fees. That’s the baseline for agentic at scale.”
Read more. (February 17)
LLMs & CHATBOTS
LLMs spending on retail media
Amid all the fuss about online retail media being potentially disintermediated by the AI chatbot experience, here’s a twist…
Marketing intelligence platform Sensor Tower analyzed the 2025 ad spend by GenAI companies such as Anthropic, OpenAI, Microsoft and Google and found that retail media networks appear to be the beneficiary.
A sample insight:
“Eight Generative AI brands were behind the category growth in 2025. Microsoft Copilot dominated with 392M impressions, followed by Google AI (132M) and Adobe (129M).
While Copilot included Costco in their retailer mix, all other brands focused their efforts on Amazon and Best Buy.
For Google AI, the category’s 2nd largest advertiser, messaging highlighted Gemini integrations across their range of hardware lines, including laptops, smartphones, and smart home devices, showcasing AI’s versatility in daily life.”
Read: “How Gen AI Prompted a Retail Media Surge” (February) – Sensor Tower
Related: “Agentic AI poised to disrupt retail, even with 50% of consumers cautious of fully autonomous purchases” (November 2025) – Bain & Company
AGENCIES
‘Heartland’ agencies and AI
A trend is emerging among agencies’ marketer clients — they are concerned their brands are not connecting properly with consumers in the “heartland” of the United States, according to The Wall Street Journal. So… they’re turning to agencies in the “heartland.”
The WSJ’s Katie Deighton observed that larger, coastal ad agencies and their holding company models are, in part, dragged down by this: “Now clients ask their agencies: Can AI help you do the work even faster, and for less money?“
Mollie Rosen, president of member experience for ad agency trade group the 4A’s, noted the small agency trend. Deighton reported:
“But smaller agencies across the board are getting better at refining their own brands and hiring employees jumping ship or laid off from big competitors, Rosen said. Falling costs of technology and data access have further leveled the playing field for smaller ad firms, she added.”
Read: “Madison Avenue Is in Crisis. Midwestern Ad Agencies See an Opportunity” (February 17) – The Wall Street Journal (subscription)
Related: “Omnicom Gears Up to Report Q4 Earnings on Wednesday, 2/18: What’s in the Offing?” (February 17) – Zacks on Yahoo Finance
From tipsheet: Hot take — all brands still want humans around even in an agentic world.
BRANDS
CMO discovers how to market to chatbots
“In late 2024, Stacy Simpson, the chief marketing officer of Athenahealth, a health care software and services provider, started asking artificial intelligence chatbots like ChatGPT about her company. The results were not ideal.
The chatbots did not know about some of Athenahealth’s offerings and did not name the company as an option when asked. They pulled details from insider-y software websites that had not been updated in years.
Ms. Simpson realized that she had to figure out how to market to A.I. chatbots. ‘This is one of the single biggest shifts we’ve seen in decades,” she said in a recent interview.’”
Read: “Chatbots Are the New Influencers Brands Must Woo” (February 17) – The New York Times (subscription)
More: “AI Is Giving CMOs Their Power Back With Sumit Virmani of Infosys” (February 17) – Adweek
Related: For B2B marketers — “Customers Hold The Key To Your New AEO Strategy” (February 2) – Amy Bills, analyst, Forrester Research blog
RESEARCH
Even eMarketer does GEO: AI Visibility Index
eMarketer Chief Content Officer Vladimir Hanzlik provided a graphic on LinkedIn yesterday from his research colleague Daniel Flynn who recently published a dataset from a November pilot called: “eMarketer AI Visibility Index.”
According to Mr. Hanzlik, the dataset tracks “how often brands get recommended when you ask ChatGPT for personal care and beauty products.”
Diving deeper, he observed:
“‘AI shelf space’ clearly doesn’t map neatly to traditional market share. Dermatologist-recommended, ingredient-focused brands punch well above their weight, while mass market giants with huge ad budgets can be almost invisible in AI answers.”
Read more on LinkedIn. (February 17)
Related: “eMarketer’s AI Visibility Index is measuring inclusion. But what about resolution?” (February 16) – AIVO Edge, an AI recommendation intel firm, on Reddit
From tipsheet: It would be interesting to see the volatility of these results — i.e. how much does visibility change week-to-week when a chatbot answer usually never gives the same answer twice? Underpinning answer volatility is large language models’ (LLMs) probabilistic versus deterministic design.
MORE
- Blackstone-Backed Ad Tech Firm Liftoff Withdraws IPO Filing (February 17) – Bloomberg (subscription)
- Google I/O 2026 is May 19-20. (February 17) – Google’s The Keyword blog
- The Series A CMO playbook: My day one strategy (February 17) – Jason Fairchild, CEO, tvScientific on Beehiiv
- Pinterest Calls ‘Code Reds’ to Revive Growth Amid AI Threat (February 17) – The Information (subscription)
- Survey: Content marketers see benefits in rise of AI (February 17) – Chain Store Age