Google's AI Brief Is Replacing Keywords. The Power Imbalance Stays.

The keyword has been declared dead so many times over the past fifteen years that it's basically the internet's version of a soap opera character — killed off in the season finale, back by episode three. Advertisers learned to ignore the obituaries, and fair enough. Every previous prediction was premature because Google had nothing better to replace keywords with.

Here's the thing: they do now.

At Google Marketing Live last month, Google unveiled AI Brief — a Gemini-powered feature that lets advertisers steer campaigns using natural language prompts instead of keyword lists. You describe your business, define what your messaging should and shouldn't say, specify the searches you want to capture or avoid, and outline who you're trying to reach. In practice, it's a conversational brief — an iterative back-and-forth with the model rather than a static spreadsheet of terms uploaded into an interface.

AI Brief isn't happening in isolation. It's the latest step in a systematic process where Google has replaced every structured advertising input with a conversational one. And for ecommerce sellers who've built their competitive advantage on campaign architecture and keyword precision, this goes considerably deeper than a feature launch.

Your Customers Moved First (And Didn't Tell You)

Previous arguments about the keyword's decline were supply-side stories. Google broadened match types, made responsive search ads pick the winning variation, let Smart Bidding set prices, and reduced visibility in search terms reports. Each step was framed as Google taking the keyword away from advertisers — the bully stealing your lunch money, basically.

This time, the pressure is coming from consumers. AI Mode has surpassed one billion monthly users. Queries submitted through it are, on average, three times longer than traditional searches. And Google's search box — redesigned for the first time in twenty-five years — now dynamically expands as users type, actively inviting longer, more conversational inputs. The interface itself is telling consumers that typing two or three words is no longer the expected behaviour.

The spend data tells the same story. Optmyzr's 2026 Match Type Study, based on analysis of 30,000 Google Ads accounts, shows exact match has lost nearly ten percentage points of spend share since 2022, while broad match has climbed steadily to become the dominant match type by budget. Advertisers are growing more comfortable trusting Google's AI with broader targeting — a shift driven by Smart Bidding's maturation rather than exact match suddenly losing its edge.

And the convergence is hard to ignore. OpenAI's ad surface, launched this year, is keyword-optional from day one. When the company that invented keyword advertising and the company most actively disrupting it both ship a keyword-optional product within the same quarter, the direction of travel isn't exactly subtle.

The Great Input Collapse (Or: How Google Ate Your Entire Dashboard)

To understand what AI Brief means for ecommerce sellers specifically, it helps to map the inputs advertisers have traditionally controlled — and notice how many of them Google has already absorbed.

The bid was the first to go. Smart Bidding has been setting prices based on real-time auction signals for years, and most serious advertisers accepted the trade: you define the financial goal, the algorithm executes at a scale no human team can match. Fine. Reasonable.

The ad creative followed. Responsive search ads let the system assemble headlines and descriptions. AI Max for Shopping now reads your Merchant Center feed attributes and generates ad copy dynamically, serving those ads across AI Overviews and AI Mode. Slightly less fine, but manageable.

AI Brief adds the targeting layer. You no longer dictate which queries to match. You describe, in natural language, what your business is and what kinds of searches you want to appear in. The system interprets those instructions and decides how to execute them.

That leaves the product feed as the last structured data input standing in the ecommerce advertising stack. And even that boundary is softening — Google's Merchant Center AI-powered onboarding tool, introduced alongside GML, can now scan a website and generate a product feed automatically. (The quality of that auto-generated feed is a separate conversation, and not a flattering one for the tool, but the trajectory is unmistakable.)

Google is systematically converting every input that used to require structured, manual specification into something the AI handles based on natural language or automated extraction. For sellers who built their Google Ads expertise on campaign architecture — segmenting by match type, device, audience, and product category with meticulous precision — this is a structural shift, not a settings update.

The Comfortable Black Box (Or: Why "More Control" Doesn't Mean What You Think)

AI Brief is being positioned as giving advertisers more steering capability — a richer, more expressive way to communicate intent to the system. And that framing isn't entirely wrong. Expressing your targeting strategy as a conversational brief is, in principle, a higher-bandwidth communication channel than a keyword list. You can convey nuance, context, exclusions, and priorities in ways that match types never allowed.

But it's worth being precise about what has actually changed and what hasn't.

With keywords, an advertiser dictated a specific instruction: show my ad when someone searches for this term, with this level of matching flexibility. The system's discretion was bounded. With AI Brief, the advertiser describes a preference. Google's AI interprets that preference and decides how to act on it.

The brief is a negotiation — but it's a negotiation where one party controls the auction, the matching logic, the creative assembly, and the measurement. Which is a bit like "negotiating" your rent with a landlord who also owns every other building on the street.

Google has reported that campaigns using Smart Bidding Exploration see an average of 27% more unique converting users. That's a meaningful lift. But what that lift represents, on close inspection, is the algorithm finding queries and audiences the advertiser didn't target, bidding at prices the advertiser didn't set, and generating conversions that can't be fully attributed to specific decisions. It works. You just can't entirely explain why. (Comforting.)

AI Brief makes the black box more pleasant to communicate with, but it doesn't make it less of a black box. You're telling Google what you want. Google is deciding how to deliver it. The distinction between dictating instructions and describing preferences is one that advertisers should understand clearly before framing this as a restoration of control.

The practical playbook circulating among experienced practitioners reflects this tension nicely: start new campaigns in phrase match, promote winners to exact, layer broad match with Smart Bidding once you have data, then use AI Brief's matching guidelines to refine from there. That sequence works — but it also reveals that AI Brief is being treated as a refinement layer on top of existing controls, not a replacement for the human judgment that sets the initial structure.

What This Actually Means If You Sell Things Online

For ecommerce operators and Amazon sellers running Google Ads as a meaningful acquisition channel, the implications stack up in layers.

The funnel question comes first. AI Brief is a discovery-layer tool, and the GML material makes the funnel structure explicit. A consumer might start with a conversational prompt — "I just had a baby and I want to remember this period, what are some ideas?" — then move to comparative research, and only then arrive at a short transactional query like "newborn photographer Los Altos." If you're only present at the bottom of that funnel, you're betting your business on being the first click for a query that was shaped by a discovery process you were entirely absent from. That's a bit like showing up to the final round of a job interview when someone else already charmed the panel over coffee.

Then there's the product feed. As every other input in the advertising stack becomes AI-mediated, the quality and depth of your Merchant Center data become disproportionately important. It's the last input where you control the raw material directly. The attributes in your feed — product descriptions, images, pricing, availability, category taxonomy — are what Google's AI reads when it generates your ad copy, decides where to place your ads, and determines which conversational queries your products match. A thin feed in this environment isn't just a missed optimisation opportunity. It's a structural handicap.

And then there's the part that gets the least attention but probably matters most: the scissors gap for smaller sellers. Google's AI-driven campaign types require meaningful conversion data to function — the commonly cited minimum is roughly thirty conversions in thirty days before automated campaigns can optimise effectively. AI Brief doesn't change that threshold. It offers richer targeting expression, but it still depends on the same underlying data volume to learn and improve. Smaller sellers with limited conversion data face the same structural disadvantage they faced before: they need volume to train the algorithm, but they can't build volume without effective campaigns. AI Brief makes the interface more accessible without making the economics more equitable. (Progress! Sort of.)

The Bottom Line

Google is not going to maintain two parallel targeting systems indefinitely. The direction is AI-first campaign types, conversational inputs, algorithmic execution. Short transactional queries will persist — people will continue typing "running shoes" and "protein powder" when they know what they want — but those queries are increasingly the endpoint of an AI-mediated discovery process, not the starting point.

The strategic question for ecommerce sellers isn't whether to adopt AI Brief. It's whether the shift from structured inputs to conversational ones changes the competitive landscape in ways that favour your business — or whether it simply transfers the advantage from advertisers who mastered keyword architecture to those who master the brief.

The feature is young, the feedback loops are short, and the performance data is thin. But the underlying shift — consumers prompting rather than searching, platforms replacing keywords with intent inference, every structured input becoming conversational — isn't a feature cycle. It's an infrastructure transition.

The keyword isn't dead. It's just been told its job is "evolving," which, as anyone who's survived a corporate restructuring knows, is rarely good news.

P.S. If your product feed currently has the depth and richness of a Wikipedia stub, now would be an excellent time to fix that. The algorithm is about to start reading it like a resume.

• These 9 AI-generated ads just won awards. Can you tell what's real?

• Meta launches Business Agent, letting companies of any size automate customer chats across WhatsApp, Messenger and Instagram in local languages. A connected platform plugs into Shopify, Zendesk and more, with paid tiers arriving soon.

• Apple unveils a Google-built Siri AI at WWDC that taps user data to answer complex questions and complete tasks like buying tickets.

• Less than a third of Google searches now send a click, with 68% ending clickless in early 2026. AI Overviews drive the surge, cutting CTR by nearly 60% where they appear. SEO still matters but no longer reliably earns traffic.

• Bots now outnumber humans online at 57.4% of traffic, breaking the web's advertising and subscription economics. As agents intermediate transactions, content becomes low-margin feedstock with little pricing power against compute and search layers.

• OpenAI rolls out product feed campaigns in ChatGPT Ads Manager, letting retailers upload catalogs over SFTP to build ads at scale. Feed ads match products to high-intent conversations and rank among the platform's top performers.

• Canva's Magic Layers now turns ChatGPT-generated images into fully editable designs without leaving the chat, collapsing the gap between AI image generation and hands-on editing.

📹 LTX Studio - End-to-end AI filmmaking from script to storyboard to finished video.

🎶 Artlist - Royalty-free music, footage and AI creative tools for video.

🖼️ Leonardo AI - Image generation with custom model training for consistent characters and assets.

📲 Arcads - Generate UGC-style video ads from a script using 1,000+ AI actors.

😍It was great, all I needed

🙂Good, not great

🙄It sucked