# Google AI Max Is Rewriting Paid Search
Google is moving paid search away from a system marketers can inspect line by line. The shift to AI Max is not just a new campaign feature. It changes who makes the decisions, what counts as control, and how much evidence a marketing team needs before trusting the result.
That matters because Google says eligible campaigns using automatically created assets and campaign-level broad match will begin automatic upgrades in September 2026. The Dynamic Search Ads sunset is now scheduled to begin in February 2027, while the other settings move sooner.
The keyword was never the strategy
For years, paid search rewarded marketers who could build a careful relationship between query, keyword, ad, landing page, and conversion. The structure was never perfect, but it gave teams something valuable: a visible chain of reasoning.
AI Max breaks that chain into a wider set of signals. Google describes a system that combines advertiser inputs with broader intent signals, search term matching, text customization, and final URL expansion. The platform can find queries and destinations that the account manager did not explicitly select.
That is useful when the system finds demand your keyword list missed. It is dangerous when the account becomes a black box that produces conversions without explaining which assumptions created them.
The question is no longer, “Did this keyword work?” It is, “Which parts of the machine are making decisions that affect this business, and can we still test them separately?”
A similar measurement problem is already visible in the fight over AI marketing attribution. More automation does not remove the need for evidence. It increases it.
The campaign manager's job is moving from selecting every input to setting the boundaries.
Google is selling scale, not certainty
Google's case for AI Max is straightforward. Search behavior is becoming harder to predict, and machine learning can match ads to more intent than a static keyword list.
Google reports an average of 7 percent more conversions or conversion value at a similar cost per acquisition or return on ad spend when advertisers use the full feature suite compared with search term matching alone. That is Google's internal data for non-retail advertisers, not a universal benchmark.
The distinction matters. A platform can report an average lift and still produce a bad outcome for a specific account. Averages describe the system. They do not explain the account.
Google's own guidance says advertisers can use brand and location controls, text guidelines, URL controls, and experiments to steer the system. Those are meaningful controls, but they are not the same as owning every decision. The new role for the marketer looks more like setting constraints, supplying useful inputs, and judging the output.
That can work. It can also create a new kind of performance theater, where the dashboard looks healthy while the team cannot answer basic questions about audience quality, query mix, or landing-page behavior.
The same mistake appears in AI marketing measurement plans that report activity instead of outcomes. More outputs are not the same thing as more insight.
The account manager becomes a governor
AI Max does not make the paid search manager irrelevant. It changes the job.
The old workflow was production-heavy. Build campaigns, expand keyword lists, write variations, review search terms, adjust bids, and repeat. The new workflow is closer to governance. Define the offer, the audience boundaries, the acceptable destinations, the conversion signal, the exclusions, and the experiment design.
That sounds more strategic, but only if the team actually invests in those inputs. A weak product page, vague conversion event, or polluted first-party data set gives the model weak material to work with. Machine learning does not turn a confused commercial proposition into a clear one.
A practical AI Max review should answer four questions:
- What business outcome is the campaign optimizing for, and is that outcome connected to revenue?
- Which search categories and landing pages are eligible for expansion?
- What exclusions protect the brand, geography, compliance posture, and customer fit?
- What experiment would prove incremental value instead of simply showing platform-reported improvement?
This is where first-party data strategy becomes less of a data-team project and more of a media-quality issue. The cleaner the signals, the more useful the automation can be.
The useful question is not whether the platform found more conversions. It is whether the right customers found the business.
Measurement has to get stricter
The worst response to AI-driven campaign expansion is to measure less. The better response is to separate platform reporting from business evidence.
Start with the conversion event. If the campaign optimizes for a shallow action, AI Max can get very good at finding shallow actions. A form fill, page view, or low-intent call is not a business result just because it appears in a conversion column.
Then look beyond the campaign dashboard. Track qualified lead rate, opportunity rate, revenue per lead, repeat behavior, and the quality of the landing pages receiving expanded traffic. For ecommerce, connect media performance to contribution margin rather than revenue alone. For local businesses, inspect calls, direction requests, booked appointments, and customer quality.
Hold out a test wherever the economics allow it. Google recommends experiments for the transition, but the design still matters. A before-and-after comparison can confuse seasonality, demand changes, promotions, and platform learning with genuine incrementality.
There is also a reporting problem. AI Max can make a campaign look simpler while making the causal story harder to tell. Teams need a change log for settings, assets, exclusions, URLs, conversion definitions, and experiments. Without one, performance reviews become arguments about screenshots.
What marketers should do before September
Do not wait for the automatic upgrade to discover that the account's inputs are weak.
First, inventory every campaign using Dynamic Search Ads, automatically created assets, or campaign-level broad match. Record the current landing-page rules, exclusions, conversion actions, budget logic, and recent search themes. This is the baseline you will need when the interface changes.
Second, separate the good constraints from the accidental ones. Some legacy settings are worth preserving. Others only exist because nobody has reviewed them in two years. Treat the migration as an audit, not a button click.
Third, build a small experiment around one commercial question. For example, can expanded query coverage produce qualified demand in a defined location without lowering conversion quality? The narrower the question, the easier it is to learn something useful.
Fourth, make landing pages and conversion signals do more work. If the model is allowed to discover intent, the site needs to explain the offer clearly enough for both the visitor and the system. This is not a reason to publish more generic pages. It is a reason to make the important pages more specific.
Finally, write down what the platform is not allowed to decide for you. Brand safety, regulatory boundaries, customer fit, and business economics should not be delegated because a setting exists.
The control debate is not over
AI Max is a preview of where paid media is going. Platforms will keep absorbing the manual tasks because scale is their product advantage. Marketers will keep asking for controls because accountability is theirs.
The winning teams will not be the ones that reject automation or accept every recommendation. They will be the ones that know which decisions can be delegated, which decisions need a guardrail, and which decisions still require a human who understands the business.
Google can find more possibilities. It cannot decide whether those possibilities are good for the company.
Frequently asked questions
AI Max is Google's AI-powered expansion layer for Search campaigns. It combines advertiser inputs with broader intent signals and can support search term matching, text customization, and final URL expansion. Google is moving it out of beta and using it to replace or absorb several legacy campaign settings.
Google says campaigns using Dynamic Search Ads are scheduled to begin automatic upgrades in February 2027 after the timeline was extended. Campaigns using automatically created assets or the campaign-level broad match setting are scheduled for automatic upgrades beginning in September 2026.
Not exactly. It reduces the role of manually maintained keyword coverage by using broader signals to match intent and expand reach. Marketers still provide important inputs and controls, but the system makes more of the matching and creative decisions.
There is no universal answer. Google reports an average 7 percent lift in conversions or conversion value at similar cost per acquisition or return on ad spend for a specific comparison, but that is platform-reported internal data. The right test depends on the account's conversion quality, economics, and ability to measure incremental results.
Start with qualified business outcomes, not only platform-reported conversions. Track lead quality, opportunity rate, revenue, contribution margin, repeat behavior, and landing-page performance, then use controlled experiments when possible to estimate incremental impact.
No. Advertisers should first audit conversion signals, landing pages, exclusions, and campaign baselines. Early testing can make sense, but the decision should follow a clear business question rather than Google's default recommendation. The machine will make more of the decisions. That is exactly why the humans need a better record of what happened.