For twenty-five years, paid search ran on one idea. You picked the words a customer might type, you wrote an ad for those words, and you paid when someone clicked. The keyword was the unit of the whole system: how you bid, how you targeted, how you reported, and how you argued with your boss about budget. Google built a company on it.
That model is being dismantled, and Google has put a date on the demolition. In April 2026 it confirmed that Dynamic Search Ads are being retired and folded into a newer system called AI Max for Search. Starting in September 2026, advertisers can no longer create DSA campaigns, and existing ones get upgraded automatically. That sounds like a routine product sunset. It is not. It is the clearest signal yet that Google is rebuilding paid search around intent and automation instead of keyword lists, and the change is worth understanding before it arrives in your account.
This is part one of a two-part series. Here we cover what AI Max is, what the death of DSA means, and what advertisers gain and lose. Part two is the migration playbook.
Origin: the long-tail problem keywords could never solve
To see why this is happening, start with a number Google has quoted for more than a decade. Roughly 15 percent of the searches Google handles each day have never been seen before. No keyword list, however large, can cover queries that do not exist yet. This is the long-tail problem, and it has shadowed paid search since the beginning.
Google's first real answer was Dynamic Search Ads, launched in beta in October 2011. The pitch was simple. Instead of asking you for keywords, DSA crawled your website, matched live search queries to the most relevant page on your site, and generated the headline and landing page automatically. You supplied the description line and the budget. For an ecommerce site with thousands of products that changed weekly, DSA was the only sane way to keep ad coverage in sync with inventory. It ran quietly in millions of accounts as a catch-all for the queries keywords missed.
DSA was, in other words, the first keywordless ad product. What is happening now is not the arrival of keywordless advertising. It is keywordless advertising eating the rest of the system.
The bridge between 2011 and now is a shift in what the automation can do. DSA matched queries to pages with a crawler and some classifiers. AI Max runs on Gemini, Google's large language model, sitting inside the ads quality stack. Ginny Marvin, Google's ads liaison, has described the core change as exactly that: Gemini now reads intent across longer, messier, more conversational queries that keyword matching was never built to handle. The model does not need your keyword list to understand what a searcher wants. That technical fact is what makes the keyword optional, and an optional keyword is a keyword on the way out.
Present: what AI Max actually is
AI Max is not a new campaign type. That distinction matters. It is a setting, a bundle of AI features you switch on inside a standard Search campaign. Google first rolled it out in beta in May 2025 and spent the following year expanding it. Four parts make up the bundle.
Search term matching is the keywordless engine. It uses broad match plus what Google calls keywordless technology to find converting queries beyond the keywords in your campaign. It learns from your existing keywords, your ad copy, and your landing pages, then matches your ads to searches you never listed. This is the feature that turns a keyword from a target into a hint.
Text customization writes and adapts the ad copy. Rather than serving a fixed responsive search ad, generative AI assembles headlines and descriptions on the fly, drawing from your landing pages, your existing assets, and the specific query in front of it. The ad a searcher sees may contain text no human in your account ever wrote.
Final URL expansion picks the landing page. When switched on, AI Max can route a click to whichever page on your domain it predicts will convert best, overriding the final URL you set. It depends on text customization being enabled, because the ad copy has to match wherever the click lands.
The AI Brief is the steering wheel, and it is the newest piece. Announced in April 2026, the Brief lets you instruct AI Max in plain English across three areas. Messaging guidelines tell it what ads should and should not say, for example "never mention prices." Matching guidelines set which searches to chase or avoid. Audience guidelines tailor the message to a segment. The Brief is Google's acknowledgement that handing the system a keyword list was always a crude way to communicate intent, and that a paragraph of context works better. It is also a Gemini feature talking to a Gemini system, which is the shape of where this is going.
Put together, AI Max does the three jobs that used to define a search marketer's week. It decides what queries to show on, it writes the ad, and it chooses the landing page.
The death of Dynamic Search Ads
That is the context for the September 2026 deadline. DSA is the obvious thing for AI Max to absorb, because they do the same job and AI Max does it with a better engine.
The timeline is set. Through August 2026, advertisers can upgrade voluntarily using Google's one-click tools, which convert dynamic ad groups into standard ad groups with AI Max features switched on. From September, no new DSA campaigns can be created through the Google Ads interface, Google Ads Editor, or the API, and remaining eligible campaigns are upgraded automatically. Campaigns using automatically created assets and the campaign-level broad match setting are swept into the same migration. Google expects all of it done by the end of September. For auto-upgrades, AI Max settings are configured to mirror the campaign's old setup, so nothing breaks on day one.
DSA is the headline casualty, but it is not really the story. The story is what its disappearance signals. Google is not retiring a niche tool. It is removing the option to run the keywordless workflow as a separate, contained, opt-in thing and making it the default surface for all of standard Search. Once your DSA campaign becomes an AI Max campaign, the same keywordless matching, generated copy, and URL expansion are available to every other campaign in the account through a setting. The fence around automation is coming down.
What advertisers gain and what they give up
This is where an honest account has to hold two things at once.
The gains are real. AI Max reaches queries no keyword research would have surfaced. Google's launch examples make the point: L'Oreal used AI Max to win net-new searches like "what is the best cream for facial dark spots," queries too specific and too varied to ever sit in a keyword list. There is also less manual work. Building, sorting, and pruning keyword lists, writing endless ad variants, and maintaining match-type structures is a large part of what SEM labor has been. AI Max removes a chunk of it. Google reports that campaigns using the full feature suite see on average 7 percent more conversions or conversion value at a similar cost per acquisition compared with using search term matching alone. That is a modest, plausible number, and it is the figure Google itself stands behind rather than the larger beta numbers it stopped quoting.
The losses are equally real, and they cluster around one word: control.
You lose control over matching. Industry testing has been blunt about this. An analysis by Ezra Sackett of Monks, shared in August 2025, looked at roughly 30,000 AI Max search terms and found that 99 percent of impressions produced zero conversions, that fewer than half of search terms were matched to an actual keyword, and that the campaigns behaved as keywordless ads more than keyword-triggered ones. His verdict on early client results was, in his word, "meh." Practitioners also report a recurring pattern of irrelevant query expansion and aggressive bidding on competitor terms, enough that some now call the feature "competitor Max."
You lose control over the ad. Text customization generates copy without pre-approval. For a regulated advertiser, or any brand with a legal review process, an ad that writes itself is a compliance question, not a convenience. You lose control over the landing page too: with final URL expansion on, AI Max can send a click to any indexable page on your domain, including an out-of-stock product or a stale page you forgot existed.
And you lose visibility into where the money goes. This is the quietest loss and maybe the most damaging. An analysis by Adalysis found that AI Max search terms often show up with no associated keyword and no way to learn why an ad served on them. Worse, when there is no matching broad match keyword, Google attributes AI Max traffic to your phrase or exact keywords, conflating data across match types. AI Max can therefore claim credit for traffic your existing keywords were already winning, which makes its reports look better than its real contribution. If that sounds familiar, it is the same transparency complaint that has dogged Performance Max for years.
The controls Google does offer have names that oversell them. There is an AI Brief, brand inclusion and exclusion settings, location controls, URL exclusion lists, and negative keywords. But Google's "brand safety" controls govern which landing page an ad points to, not which queries trigger it. Negative keywords still apply, yet practitioners report that URL expansion can match ads to queries tied to pages you never specified, and negatives do not always catch those. Every one of these controls is opt-in and off by default. The automation is the default. The guardrails are homework.
Future and impact: why Google is really doing this
None of this is happening in a vacuum, and the reason is the search box itself.
AI is reshaping how people search. Google's AI Overviews now sit above results for a large share of queries, often answering the question outright. Seer Interactive's analysis of data through September 2025, covering 3,119 search terms across 42 client organizations, found that paid click-through rate on queries with no AI Overview ran at 13.9 percent, but on queries showing an AI Overview it dropped to 4.1 percent when the brand was not cited in the overview, recovering only to 7.9 percent when it was. A growing share of searches end with no click at all. We cover that shift in detail in our piece on AI Overviews and paid search.
When the search results page is being rebuilt around AI answers and conversational queries, the keyword-and-exact-match machinery starts to look like the wrong tool. People do not type keywords at an answer engine. They ask questions, in full sentences, the way they would ask a person. A system that matches on a curated word list cannot keep up with that, but a large language model reading intent can. AI Max is Google rebuilding the ad system to match the search system. The deprecation of DSA is one visible brick in a much larger wall.
For a paid search decision-maker, the honest read is that this is neither a gift nor a betrayal. It is a trade, and the terms are not optional. You gain reach into a long tail you could never have targeted by hand, and you give up the granular control that made paid search feel like a craft. The open questions are not yet answered. Whether brand safety holds when an AI writes the ad, whether irrelevant query expansion gets reined in, whether Google ever makes reporting honest enough to see where spend goes: all of that is still live. Anyone who tells you AI Max is simply better, or simply worse, has not looked closely.
What is not in question is the direction. The keyword is becoming a signal rather than a target, manual SEM labor is being automated away, and the deadline for the first big piece is September 2026. The practical question is how to move without losing the control that still matters. That is the subject of part two: AI Max Migration, a playbook for the new SEM.
Council summary
This post argues that the retirement of Dynamic Search Ads is not a routine product sunset but the moment Google stops treating keywordless, automated search advertising as a contained opt-in feature and makes it the default surface for all of Search. It explains AI Max clearly as a four-part setting, search term matching, text customization, final URL expansion, and the new AI Brief, and it stays honest about the trade: real reach into long-tail intent and less manual labor, set against measured losses of control over matching, ad copy, landing pages, and reporting transparency, with independent tests and attribution complaints cited rather than glossed over. The reader's takeaway is that the September 2026 deadline is fixed and the shift is not optional, so the work is not deciding whether to adopt AI Max but learning to keep the guardrails, which are off by default, switched on. The piece is accurate against Google's own announcements and third-party testing, balanced in tone, and earns its length. It passes.
Comments