The new ad unit is a brand story. What Google’s AI Brief signals about ads’ future
The shape of the ad unit is changing across Google, ChatGPT, and every AI assistant in between, and Google just made it official.
Last week, Google added a feature called AI Brief to its ads platform. It lets advertisers describe their brand in plain English, what their ads should and shouldn’t say, who they’re for, what kind of search queries to chase and uses that as the input to the matching engine. No keywords. No bids. Just a description.
Paid media has just become closer to AI visibility. In the same way ChatGPT and Claude recommend products by understanding brands and products' story as it is built across media assets, Google will now make its’ ads synthesize a brief. Whether it’s ChatGPT picking which products to recommend or Google deciding which ads to show, the online answer increasingly depends on the story your brand tells across the web.
The numbers
Adobe’s Q1 2026 numbers a pause. AI traffic to US retail sites grew 393% year-over-year, after a 693% surge during the 2025 holiday season.
The conversion data is sharper. In March 2026, AI traffic converted 42% better than non-AI traffic. Twelve months earlier, the same AI traffic converted 38% worse. An 80-point swing in a year.
AI-referred shoppers spend 48% more time on site, and 66% of consumers now say they trust AI tools for shopping.
Ethan Smith’s Graphite data adds the why: ChatGPT traffic converts 6x better than Google search. In Google, ranking #1 is the goal: one slot, one winner. In LLMs, you need to be cited as many times as possible across many sources.
Which is a real change in how brands compete. Google built a $260B+ ad business by being the place shoppers go to find products. Answer engines are taking on that role, except they don’t return ten blue links for the user to choose from. They return one recommendation, with or without reasoning.
The story problem
The old SEO world wasn’t really built for stories. You had 60 characters for a title tag, 155 for a meta description, a keyword you had to repeat a few times without sounding insane. The whole thing was optimized for phrases, not ideas you were trying to land in a slot, not explain why you existed.
There wasn’t much room for the why. Google didn’t really care why your product existed, it just cared whether your H1 matched the query. So brands learned to write like robots, because robots were doing the finding. The story got compressed out of the page.
AI conversation flips that. When a shopper asks ChatGPT “what’s a good non-toxic cookware brand for someone with kids,” the model isn’t matching keywords, it’s reasoning over a narrative. It needs to know who you are, who you’re for, what trade-offs you’ve made, what other people have said about you. The unit you’re optimizing for isn’t a phrase anymore. It’s a coherent, repeatable, citable story.
What Google actually shipped
A year ago Google launched AI Max for Search. The pitch was simple: stop manually picking keywords. Hand Google your landing pages, your product feed, and a few guardrails, and let the model figure out which queries you should show up for. It became the fastest-growing AI Search ads product Google has ever shipped.
The April 30 announcement had three parts. The first is that AI Max is coming to Shopping. Until now, Shopping ads have run on exact product queries, you search for a thing, you see ads for that thing. AI Max for Shopping changes that. It uses your Merchant Center feed, things like fabric and durability and fit, to match products to broader, more conversational questions like “what are the best high-quality clothes for lounging?” The model reads your product the way a salesperson would, and decides which conversations it belongs in.
The second is that the same logic is rolling out for Travel. And the third, the one that’s most interesting, is a new feature called AI Brief.
What AI Brief actually is
AI Brief is a Gemini-powered input box where advertisers, in plain English, tell Google three kinds of things:
- What your ads should and shouldn’t say. (“Never mention prices.”)
- Which searches to capture or avoid. (“Prioritize searches for healthy pantry staples.”)
- Who you want to reach and how to talk to them. (“For people who are health conscious, highlight our clean products.”)
If you read those three again, what you’re actually looking at isn’t really an ad targeting tool. It’s closer to a brand brief, the kind of thing a marketer would write up for an agency.
For the last twenty years or so, Google’s ad system has asked you for keywords, bids, and match types. There hasn’t really been a place to tell it what your brand actually stood for. Now there is, and it’s probably going to end up being one of the more important fields on the page.
Which is why this matters beyond Google. The thing you optimize for has shifted from a keyword to a brief, from a phrase to a story. And once you start looking, it’s the same shift happening in AI search. ChatGPT, Claude, all of them are doing a version of the same thing: reasoning over a narrative instead of matching against keywords. So the brands that get the story right end up winning in both places. The ones still organized around the keyword era are paying more for less in Google, and slowly disappearing from the AI assistants altogether.
Why this is bigger than one feature launch
If you step back, the actual ad unit is changing shape. The old one was basically a slot you bought your way into, on a system that ran on keywords and bids. The new one is starting to look more like a paragraph the model reads about you on Google, in ChatGPT, in Claude, and uses to figure out whether to put you in front of someone who’s about to buy.
And it’s probably worth saying, OpenAI is heading in the same direction. They started testing ads in ChatGPT back in February. There’s no public ad manager yet, no keyword auction, no bidding interface. Ads are picked by “relevance to your current conversation,” which is a polite way of saying the model decides. When OpenAI eventually opens that up to advertisers, and they will, it’s hard to imagine the input being a list of keywords. It’ll be something a lot closer to AI Brief: tell the model who you are, who you’re for, what you stand for, and let it figure out where you belong. Same direction, same shift. The ad unit is becoming a story.
.webp)