TLDR: AI Max is Google’s latest evolution of the Search campaign and it is the direction Google is heading, whether you opt in or not. It moves away from keyword-based bidding towards search themes and audience signals, sitting somewhere between a traditional Search campaign and Performance Max.
What is AI Max (and what it’s not)?
AI Max is a new campaign type within Google Search, but it is worth being clear from the start about what that means, because it is easy to assume it works like a traditional Search campaign. It does not.
A standard Search campaign gives you tight control over exactly who sees your ads, matching specific keywords to specific queries with a level of granularity that lets you get very precise, right down to exact match terms. AI Max does not work that way. You cannot set up an AI Max campaign around an exact keyword like “black handbag” and expect it to simply serve your ad to everyone searching for that term. Instead, AI Max operates on audience signals and search themes, using Google’s AI to match your ads to relevant intent rather than specific keywords.
The closest comparison is PMax, which launched as Google’s previous step towards broader automation and cross-channel reach. The difference is that AI Max stays within the Search environment. It prioritises overall search coverage over the pinpoint keyword matching of a traditional campaign, and it uses search themes as its targeting mechanism rather than keyword lists. There is an element of crossover with both standard Search and PMax; it lives in the Search space but leans heavily on the automation and signal-based approach that PMax introduced.
The other important thing to understand is that AI Max is not staying optional for long. Any Dynamic Search Ads campaigns currently running will be automatically rolled into AI Max. Campaigns running on broad match settings are also moving in this direction. This is where Google is going, less keyword-based bidding, more AI-driven automation.
That said, AI Max is still relatively new, and like PMax before it, it sits in a “black box” phase where you do not have the full level of insight you would have with a standard Search campaign.
How to enable AI Max in your Search campaigns
Setting up AI Max follows a similar process to creating any standard Search campaign in Google Ads, with one key addition.
Start by clicking the plus button to create a new campaign and selecting your objective. For most accounts this will be a conversion-based goal: purchases, form fills, or whatever conversion action is most relevant to the business. You then select Search as your campaign type, AI Max is still a Search campaign at its core.
The key step is in the campaign settings, where there is a toggle to turn AI Max on. Enabling that toggle is what switches the campaign from a standard Search setup into AI Max mode. Once enabled, you will see a range of additional settings within the campaign that can be turned on or off depending on how much control you want to retain.
Google will typically recommend enabling everything from the start to maximise the learning data available to the campaign. That is a reasonable approach, and it will give you more data to work with faster. However, there are specific settings you should configure regardless of how broadly you open up the campaign, particularly around URL expansions and exclusions.
It is also worth checking whether an experiment setup is available for your account before going live. This allows you to run AI Max alongside a base campaign with a split budget, which gives you a cleaner read on performance without committing your entire spend to the new campaign type from day one.
How URL expansion works and when to restrict it
URL expansion is one of the most important settings to understand within AI Max. With it enabled, Google can serve an ad pointing to a different page on your website than the one you originally specified.If its AI determines that another page is more relevant to the user’s search.
The logic behind this is sound. If a user’s search signals suggest they are interested in a specific product that lives on a different URL to the one in your original ad, serving them the more relevant landing page is likely to perform better. Used well, URL expansion gives Google the flexibility to match users more precisely to the content on your site.
The risk is that without proper exclusions in place, Google may start sending traffic to pages that are not useful for conversion and where spend will be wasted.
Pages to exclude as a priority:
- Order confirmation pages. Users who land here have already purchased. There is no value in serving ads that send people to a confirmation page.
- Login pages. If someone is navigating to a login page, they are an existing user, not a new prospect. Spending budget on that click is unlikely to generate a new conversion. For example, if a member-based client has a login page that existing members use to access their account, you absolutely do not want to be paying to send ad traffic there.
- Careers pages. Traffic landing on your careers page is not in buying mode. Exclude the full URL or use a URL pattern exclusion, anything containing “/careers”, to prevent the campaign from sending budget there.
- Any page without conversion tracking. If Google cannot see a conversion happening on a page, it cannot optimise towards one. Pages with no tracking in place should be excluded.
You can add URL exclusions either as a full URL or using a pattern, for example, excluding any URL containing “/careers” or “/login” will cover all variations of those pages without needing to list every oneindividually.
Reviewing the URL expansion report regularly as the campaign runs will also show you which pages are receiving traffic. If a page is pulling in significant clicks but generating zero conversions, it is a signal to add it as an exclusion and redirect that budget towards pages that convert.
How to review and manage automatically created assets
When AI Max is running, Google will generate assets, headlines and description lines, automatically, based on your website content and the signals available to the campaign. These appear in the campaign under the assets tab, which is separate from the standard ads view you would use in a traditional Search campaign.
To find them: go into the campaign, click on Ads, then navigate to the Assets tab. From there you can filter to see only the automatically created assets, the ones Google has generated rather than the ones you wrote yourself.
Each asset will be rated by Google, typically as Best, Good, Low, or Learning. Here is how to interpret those ratings and what to do with each:
Best and Good
These assets are performing well relative to others in the campaign. If you did not write them yourself, check that they are on-brand and that the messaging is accurate before leaving them to run. Google optimising towards a piece of copy that misrepresents your product or uses language that does not fit your brand is a problem even if the CTR is strong.
Low
A low rating does not automatically mean the asset is bad, it may simply be competing against stronger copy, or it may not yet have accumulated enough data to be fairly rated. If the campaign has only been running for a couple of weeks, do not make changes yet. Wait until you have around 30 days of data before drawing conclusions. Once you do have enough data and an asset is consistently rated low, look at the metrics, click-through rate, impressions, conversion rate, and compare it against your best-performing assets. If it is clearly underperforming and the copy itself is weak or off-brand, you can remove it. There is a checkbox next to each asset and an edit button that lets you opt out of serving that asset.
Learning
Google has not yet gathered enough impressions to rate this asset. Leave it alone. Any changes made during the learning phase will reset the process and slow down the campaign’s ability to gather useful data.
The same review logic applies to the URL expansion report. Check it regularly for pages receiving high traffic volumes with low or zero conversions. When you spot them, add them as exclusions so the budget can be directed towards pages that are actually working.
How to add customer intent signals
Customer intent signals tell Google who your ideal customer is, so the AI can go and find more people like them. There are three types to work with, each with a different level of specificity.
CRM and customer lists
This is the most valuable signal you can provide. Export your customer data, email addresses, phone numbers, from your CRM and upload the list to Google Ads. Google will match these to signed-in users and use their actual behaviour and purchase history as the strongest possible guide to who your ideal customer is. Always prioritise getting a CRM list uploaded before launching an AI Max campaign.
Remarketing audiences
These are people who have already interacted with your brand in some way, visited your website, engaged with a YouTube channel, interacted with a social campaign. They are still high-intent signals, just one step removed from your actual customers. Upload your website remarketing lists and any other interaction-based audiences you have available.
Interest-based audiences: in-market and affinity
These audiences cast a wider net. In-market audiences are people who have shown active purchase intent in a relevant category, they are researching or comparing products like yours. Affinity audiences are broader, based on general interests rather than active purchase behaviour. In-market is the more useful of the two because it carries stronger purchase intent, but it is worth uploading both. The combination gives Google a richer picture of the type of person you are trying to reach, even when they have had no direct interaction with your brand yet.
The key principle is the same as with any campaign: the richer and more accurate the signal you provide, the better the AI can perform. A campaign running without audience signals is leaving Google to work with significantly less information than it needs.
How to monitor AI Max performance week by week
Once AI Max is live, the approach to performance monitoring is different from how you would manage a traditional Search campaign. Google has moved decisively away from daily and weekly optimisations, making frequent changes disrupts the learning process and prevents the AI from building up the data it needs to perform.
The recommended approach is to give a brand new AI Max campaign a solid four weeks before making any significant changes. Weekly check-ins are still valuable for spotting problems early, but they should be diagnostic rather than reactive.
Here is what to look at in your weekly review:
Automatically created assets
Check the asset ratings and confirm that any assets rated as Best or Good are on-brand and accurate. Google may surface copy that performs well in terms of engagement but does not represent your brand correctly. If an automatically generated asset is not approved, remove it even if it is rated well.
URL expansion report
Look for any URLs that are receiving significant traffic but were not intentional targets. If you spot a high-traffic page that you would not have chosen to send users to, add it as a URL exclusion. Also flag any pages receiving high traffic with zero conversions, these suggest a broken user journey and should be excluded so the budget can be redirected.
Search terms
Review the search terms triggering your ads. Even within AI Max, you can see which queries are driving traffic. Look for terms that are clearly irrelevant to the brand, driving spend without converting. These should be added as negative keywords. If a term is generating impressions, clicks, and no conversions consistently, that is a signal the audience signal targeting needs reviewing or that the term needs to be excluded.
For larger optimisations, refreshing assets, revisiting audience signals, reviewing the overall campaign structure, a monthly cadence makes more sense than weekly. The goal in those monthly reviews is to assess whether the overall campaign is moving in the right direction and make strategic adjustments, rather than reacting to short-term fluctuations in the weekly data.
How to test whether AI Max is helping or hurting
The cleanest way to test AI Max against your existing Search activity is through Google’s built-in Experiments tool, which allows you to run a controlled A/B test with a split budget, so you are not committing your full spend to AI Max until you have evidence that it performs.
Note: the availability of the experiment setup for AI Max specifically may vary depending on your account and Google’s rollout at the time of reading, it is worth checking the Experiments tab in your account to confirm what is available.
If you are able to set up an experiment, here is how to approach it:
Choose a mid-level campaign as your base
Not your highest-converting campaign, and not one with very low volumes. You want a stable, consistent level of data, enough to draw meaningful conclusions without the noise that comes from very high or very low volume.
Split the budget 50/50
An equal split between the base campaign and the AI Max test gives you a fair comparison. If the split is uneven, the results will be harder to interpret.
Set an initial end date of four weeks
This gives both campaigns enough time to move through the learning phase and generate meaningful data. In practice, four weeks will sometimes not be enough for a clear winner to emerge, particularly if conversion volumes are moderate. If that is the case, extend the experiment for another month and report back to stakeholders with the interim data and a clear explanation of why more time is needed.
Use the Experiments tab to monitor results
Google Ads provides a comparison table within the Experiments view that shows you what the base campaign and the AI Max test are doing side by side, impressions, clicks, conversions, cost per conversion, and ROAS. This is the clearest picture you will get of whether AI Max is genuinely outperforming your existing setup.
In a testing run for a client, AI Max outperformed the base Search campaign in a head-to-head experiment, which is a meaningful data point given the campaign environment. That does not mean it will outperform in every account and every context, but it does suggest the performance upside is real when the campaign is set up correctly and given enough time to learn.
The honest assessment: AI Max will not suit every campaign or every business right now. But given that Google is actively moving all Dynamic Search Ads and broad match campaigns towards it, building familiarity with how it works, and testing it in a controlled way before it is imposed, is the more prepared position to be in.
Need help setting up or managing AI Max?
If you are navigating the shift to AI Max and want to make sure your Search campaigns are set up to perform in Google’s evolving landscape, our Paid Search team can help, from initial setup and audience signal configuration through to ongoing performance monitoring and experiment design. Get in touch with the Modo25 Paid Search team to find out how we can help.
