For twenty years, running a Google Ads search campaign meant one thing above all: building and grooming a keyword list. You picked the exact phrases you wanted to show up for, bid on them, and bolted on negatives to keep the wrong traffic out. That era is ending — and not as an option you get to weigh at your leisure. Google's AI is quietly turning the keyword from the boundary of your targeting into a mere hint, and bidding without keywords is becoming the default rather than the experiment.
The engine behind that shift is AI Max for Search, and if your team is still debating whether to switch it on, the conversation is already out of date. This piece covers what AI Max actually does, why it's happening now, the honest picture of its results (which are more mixed than Google's slides suggest), and — most importantly — how to keep control when the keyword is no longer the thing you steer with.
What AI Max actually is
Strip away the branding and AI Max for Search is a bundle of features that let Google match your ads to searches your keyword list never named. It learns from what you already have — your keywords, your ad copy, your landing pages — and uses that to find and serve ads on related queries. Three pieces do the heavy lifting.
Search-term matching — combines broad match with "keywordless" technology to show your ads on relevant queries you never added, learning from your assets and URLs.
Text customization — automatically generates headlines and descriptions (the feature formerly called "automatically created assets") tailored to the landing page and the query.
Final URL expansion — sends a searcher to the most relevant page on your site rather than a single fixed landing page.
Say you sell running shoes and only bid on "trail running shoes." A shopper searches "best cushioned shoes for a first marathon." Your old keyword list misses them entirely; AI Max can recognise the intent, adapt a headline to it, and route the click to your marathon-training collection page. That's the pitch: reach into demand you were previously blind to. It builds on the automated bidding foundations any advertiser should already understand — the same Smart Bidding logic we cover in structuring a first profitable Google Ads campaign.
Why this is happening now
AI Max isn't an isolated ad product; it's Google's response to Search itself changing. As AI Overviews and the conversational AI Mode reshape how people search — longer, messier, more exploratory questions asked in natural language — a fixed list of two- and three-word keywords simply can't keep up with how queries now look. Google has openly reframed its goal from matching a user's query to predicting what they need next, and ads have to follow the same path. If you want the paid context for what's happening to organic results, it's the mirror image of how AI Overviews are reshaping SEO.
The shift The keyword hasn't been deleted. It's been demoted — from the boundary of your targeting to one of many signals an AI uses to guess intent.
It's becoming the default, not the option
Here's the part most "should you try AI Max?" articles haven't caught up to. AI Max has moved out of beta, and Google is scheduled to begin automatically migrating legacy campaign types — Dynamic Search Ads, automatically created assets, and campaign-level broad match — onto AI Max from around September 2026. In other words, for a large slice of advertisers, keywordless matching is arriving whether they opt in or not. The practitioner's question has changed from "is it worth enabling?" to "how do I stay in control once it's on?" It's the same hand-over-the-wheel dynamic we've tracked across the industry in agentic AI taking over campaign execution, now arriving in the one place advertisers guarded most jealously: search targeting.
AI Max is the on-ramp to ads inside AI Search
There's a second reason to care beyond reach. AI Max is Google's primary mechanism for placing ads inside the new AI search surfaces. At Google Marketing Live in 2026, Google introduced ad formats built for the AI era — conversational, Gemini-powered ad experiences, "highlighted answer" placements, and AI-driven shopping ads that surface as people research inside AI Mode. The route to being eligible for those placements runs through AI Max and Performance Max. So if appearing in AI-generated answers matters to you — and given where search behaviour is heading, it will — this is the on-ramp, much as being surfaced organically is the goal of generative engine optimisation. It's the paid cousin of the same trend we flagged for Meta in Meta's AI assistant in Ads Manager: every big platform is folding campaign creation and placement into an AI layer.
The honest scorecard
Now the part the promotional material skips. Does bidding without keywords actually work better? The truthful answer is: sometimes, and it depends heavily on your account.
Google's own figures are encouraging — it reports that advertisers using the full AI Max suite tend to see more conversions or conversion value at a similar cost, with larger lifts for campaigns that were previously reliant on exact and phrase keywords, and it points to brand case studies with higher conversion rates at lower cost. Those numbers are real, but they're Google's, drawn from the scenarios where the tool shines. Independent testing tells a more sober story: a large majority of advertisers in one body of testing saw neutral or negative results at the account level, and the gains that did appear were often smaller measured across the whole account than they looked at the individual campaign level. The pattern that holds up is that using all three features together beats using one in isolation, and that automated text assets can genuinely lift relevance.
The fair reading is that AI Max is a real tool with real upside and real variance — not a magic button. Which is exactly why you don't take Google's word or a sceptic's; you test it against your own baseline, especially now that a GA4 attribution-model change earlier in 2026 has shifted how conversions are credited, a wrinkle that connects straight to the broader problem of why attribution is getting harder and the case for cross-channel measurement approaches like marketing mix modelling.
What replaces the keyword
Losing keyword-level targeting doesn't mean losing control — it means the controls move. The skill shifts from writing a keyword list to setting the guardrails an AI operates inside. These are the levers that now do the job keywords used to.
- Brand controls — specify brands your ads should appear alongside, or block ones they shouldn't, at campaign and ad-group level.
- Locations of interest — target the geography a searcher is interested in, not just where they physically are (someone in Dallas searching "plumber in Austin" can be matched to your Austin ad).
- Negative keywords — still here, and more important than ever as your first line of defence against irrelevant expansion.
- The search terms report — your single most valuable habit now; the improved report shows the actual queries, headlines, and URLs so you can prune waste weekly.
- URL controls and asset removals — steer where traffic lands and remove generated assets that don't fit your brand.
- Landing pages and product feeds — increasingly a control in themselves, because poor-quality pages can disqualify you from AI placements altogether.
That last point deserves weight. When the AI decides which query to match and which page to send it to, the quality and structure of your site becomes part of your targeting, not a downstream detail. Clear, question-focused pages and complete, accurate feeds now influence whether you're even eligible to appear — which is why a well-built, well-structured website matters more in the keywordless era, and why the principles in designing websites for AI search now apply to paid as much as organic.
How to run it without losing control
A practical way to adopt keywordless bidding without handing over the account blindly.
| Move | Why it matters |
|---|---|
| Run it as an experiment first | Google's built-in 50/50 split measures the real lift on your account, not Google's average |
| Set tight guardrails before scaling | Brand controls and negatives keep expansion from wandering into junk traffic |
| Read the search terms report weekly | It's where waste and brand-safety issues surface — and where your negatives come from |
| Protect brand and competitor terms | Keywordless matching can drift onto queries you'd never have bid on |
| Invest in landing pages and feeds | They're now eligibility criteria for AI placements, not an afterthought |
| Re-baseline your measurement | Attribution shifts mean last year's benchmarks may not be comparable |
None of this is exotic; it's the same discipline good advertisers always applied, pointed at a new control surface. The teams getting the most from AI Max aren't the ones who trust it most — they're the ones who test it hardest and fence it tightest.
The bottom line
Bidding without keywords isn't a glimpse of the future anymore; it's the direction Google is actively defaulting advertisers toward, and the gateway to advertising inside AI Overviews and AI Mode. AI Max can genuinely find demand your keyword list missed, but the results vary enough that Google's headline numbers shouldn't be taken as your own — you have to prove it on your account. The winning move is neither to resist the shift nor to surrender to it, but to change what you manage: fewer keywords, more guardrails, relentless attention to the search terms report, and real investment in the pages and feeds the AI now leans on. Keywords built the last twenty years of search advertising. Judgement about where to point the AI, and how tightly to fence it, will build the next few.
AI Max is about to be your default — make sure it works for you, not against you.
We test, fence, and manage keywordless Google Ads campaigns so the AI drives profit, not waste.
Explore Performance Marketing →Frequently asked questions
What is Google AI Max for Search?
A suite of AI features that lets Google match your ads to relevant searches without relying only on your keyword list. It combines keywordless search-term matching, automatically generated text assets, and final URL expansion, learning from your existing keywords, creative, and landing pages to find queries you never explicitly targeted. It's also Google's main mechanism for connecting Search ads to AI Overviews and AI Mode.
Does Google Ads still use keywords?
Keywords still exist, but they're becoming an input rather than the boundary of your targeting. With AI Max and broad match, Google increasingly matches ads to intent rather than exact strings, and it's moving campaign types like Dynamic Search Ads and campaign-level broad match onto AI Max by default. The job shifts from choosing keywords to steering an AI with guardrails, assets, and negatives.
Does AI Max actually improve performance?
The results are genuinely mixed. Google reports meaningful gains and some brands see strong lifts, but independent testing has found a large share of advertisers see neutral or negative outcomes at the account level, with gains often smaller than campaign-level numbers suggest. It's a real tool with real upside and real variance, so testing against your own baseline matters.
How do you control targeting without keywords?
You trade keyword-level control for other levers: brand controls to include or exclude brands, locations of interest for geographic intent, negative keywords to block irrelevant queries, aggressive monitoring of the search terms report, URL controls, and removing weak generated assets. Increasingly your landing pages and feeds act as controls too, since poor pages can disqualify you from AI surfaces.
Should I turn AI Max on now?
Test it rather than flip it on blindly — and soon, because it's becoming the default. Google has built an experiment flow that splits traffic between AI Max on and off so you can measure the true difference on your account. Run that experiment, set tight guardrails, watch the search terms report, and protect your brand terms, then scale where it wins and hold back where it doesn't.
