Google AI Max Is Now GA: The DSA Sunset Playbook for Paid Search

Google AI Max Is Now GA: The DSA Sunset Playbook for Paid Search

On Tuesday, a search marketer at a mid-size DTC brand got an email from her Google rep. The subject line: *Your Dynamic Search Ads will upgrade automatically in September.* She forwarded it to her paid team with one sentence on top — "Do we have a migration plan, or are we just going to find out in Q4?" That question is now sitting in a lot of Slack channels, because AI Max just moved out of beta on April 15, and the countdown to the forced DSA migration is real.

Google's pitch is simple. AI Max runs an intent-based auction instead of a keyword-based one — it reads the query, the landing page, and the creative as a bundle, and ranks them by predicted conversion instead of keyword match. Google says the average AI Max campaign gets 7% more conversions or conversion value at a similar CPA. The catch: if you wait for the automatic September cutover, you'll be learning a new auction model during Q4, which is the worst quarter of the year to be learning a new auction model.

What AI Max actually changes

The thing to understand is that AI Max isn't just DSA-with-more-AI. It replaces three separate campaign primitives in one motion:

• Dynamic Search Ads. The old headline was "let Google pick which queries to match" — AI Max does that plus creative selection.

• Automatically Created Assets (ACA). The old ACA layer auto-generated headlines. AI Max takes that further and lets you set *text guidelines* — brand rules, forbidden phrases, target messaging — that constrain what it's allowed to generate.

• Campaign-level broad match. The old broad-match-as-default setting got quietly folded into AI Max's matching engine.

For a lot of paid teams, this is three tool migrations at once. The reporting also changes — Google is now surfacing query-level, cross-headline, and cross-landing-page reports that weren't available in the old broad-match + DSA stack. That's a real upside. But it means your weekly paid search dashboard needs rebuilding before the migration, not after.

Why voluntary early is the correct call

The case for flipping now, voluntarily, comes down to one calendar fact: you only get one shot at learning this auction in a low-stakes window. Right now — April through August — most brands are coming off Q1, the testing appetite is high, and you can run a 60/40 split (60% legacy, 40% AI Max) on a non-hero campaign and actually read the results. If you wait, the upgrade flips in September, your team learns the new controls while Q4 budgets are already committed, and mistakes show up in revenue.

The other reason is the controls themselves. Brand marketers spent the last three years complaining that Performance Max was a black box. AI Max was built after that feedback, and the control surface is meaningfully better. You get geo settings that actually bite, brand controls with exclusion language, and a text-guideline feature that lets you say *"never generate a headline using the word 'cheap' or 'discount'"* in plain English. These are all features that reward the team that learns them first.

The migration QA checklist

If you're going to pilot AI Max before September, do it with an actual QA plan. A few bullets that should be on that plan:

• Baseline the legacy campaign first. Last 90 days: CTR, CPA, conversion volume, branded-vs-non-branded query mix, top 20 landing-page-to-query pairs. Save it. You'll need this to judge the AI Max results.

• Start with a non-hero campaign. Not your brand-term campaign, not your highest-volume product line. Pick something middle-tier where a 15% drawdown for two weeks won't ruin the quarter.

• Write the text guidelines before you flip. The default is permissive. Spend an hour writing brand rules — banned words, required phrases, tone constraints — and load them in on day one. Otherwise AI Max will generate headlines that make your brand team angry in week two.

• Rebuild the weekly report around query-level data. The old DSA reports surfaced auto-generated final URLs. AI Max gives you query-level and cross-headline attribution. Your Monday paid-search dashboard needs to use the new reports, not the old ones.

• Set a 14-day no-touch window after launch. AI Max needs two weeks of learning before the numbers mean anything. If you start tweaking on day 4, you'll burn the test.

What changes downstream

The DSA sunset is the leading edge of a bigger shift. Microsoft announced its own AI Max equivalent on April 21, pitched at the "agentic web" — same idea, intent-based auction, landing-page-aware matching. The big platforms are all converging on a model where you stop specifying keywords and start specifying outcomes, and the auction figures out the rest. That's a real handoff of control from the search marketer to the platform, and the practitioners who thrive through this transition are the ones who rebuild their skill stack around guiding the AI — writing text guidelines, curating landing pages, reading query-level reports — instead of bidding on keywords.

So here's the specific action you can take this week. Pull last quarter's DSA campaign list, identify the middle-tier campaign you can afford to put on AI Max for 30 days, and get a voluntary migration on the calendar for early May. That gives you two test cycles before the forced September flip — enough to understand the new controls, rebuild your reporting, and walk into Q4 with a working playbook instead of a forwarded email and a panicked Slack message.