Performance Max: AI Discovery Strategy for 2026

To get Performance Max working for AI agent discovery, you have to be deliberate about campaign structure and how you create your assets, particularly now that AI-driven search and recommendation engines are becoming the main way people find things. The success of digital advertising from here on out is all about how well your campaigns can feed these intelligent systems with clean, high-quality data.

Setting Up Your Performance Max Campaign for AI Discovery

A successful Performance Max (PMax) campaign for AI discovery all comes down to the setup. AI agents are basically just pattern-recognition machines. Giving them a clear, well-organized structure right from the start is everything.

Creating a New Performance Max Campaign

1. Navigate to Campaigns: Inside your Google Ads Manager, find the navigation panel on the left and click on Campaigns.
2. Initiate New Campaign: Hit the big blue + New Campaign button.
3. Select Campaign Goal: It’ll ask you to “Select a campaign goal.” Pick the one that actually matches what you’re trying to do, like Sales, Leads, or Website traffic. PMax is flexible, but AI agents are wired to follow user intent, so a clear goal is absolutely necessary for their learning algorithms to make sense of your campaign.
4. Choose Campaign Type: Pick Performance Max from the campaign types. This is the only choice that gives the AI full access to all the inventory across Google’s entire network.
5. Enter Business Information: Put in your website URL and give the campaign a name that tells you what it is at a glance. Something like “PMax_Q3_ProductLaunch_AI_Discovery” provides instant context. Then click Continue.

Pro Tip: I always start with a single, clear campaign goal. Even though PMax is built to be an all-in-one solution, giving the AI a focused objective stops it from spreading its learning too thin trying to chase five different conversion types. When you’re this specific, you’re telling the AI agent exactly what a ‘win’ looks like for your business.

Structuring Asset Groups for AI Agent Understanding

Asset groups are the basic building blocks of PMax, and how you set them up is probably the single biggest thing you can do to influence AI discovery. Each asset group needs to be its own self-contained theme, product line, or audience segment. This modular setup is what lets AI agents test and serve the most relevant ad combination to a user, no matter where they are.

Defining and Populating Asset Groups

1. Name Your Asset Group: Give it a descriptive name you’ll understand later, like “Luxury_Watches_Mens_Audience” or “Service_A_LeadGen.” It makes tracking performance so much easier.
2. Add Final URL: Point it to the single most relevant landing page for this specific asset group. AI agents will send traffic from these assets to this URL.
3. Upload Creative Assets: This is where you feed the machine its content. Google’s AI will mix and match these to build the actual ads people see.
   • Headlines (up to 5): Write short, punchy headlines (max 30 characters). Pack them with keywords relevant to the asset group’s theme.
   • Long Headlines (up to 5): These are your more descriptive lines (max 90 characters).
   • Descriptions (up to 4): Write unique descriptions (max 90 characters) that sell the benefits and have a clear call to action.
   • Business Name: Your brand.
   • Images (up to 20): Upload a wide variety of high-res images. You absolutely need different aspect ratios (square, field, portrait) so they look good everywhere. For example, have a 1.91:1 field image for display ads, a 1:1 square for feeds, and a 4:5 portrait for Shorts. Visuals still drive engagement, eMarketer’s 2024 report confirmed that again, so image quality is something you can’t compromise on.
   • Logos (up to 5): Provide a few logo versions, like a square and a field one.
   • Videos (up to 5): You *must* include videos. If you don’t, Google will auto-generate some, and they’re almost always terrible compared to something you made yourself. Give the AI agents options by varying the lengths, like a quick 15-second spot for Shorts and a longer 30-second version for other placements.
4. Add Audience Signals: This step is your chance to steer the AI. While PMax is automated, strong audience signals help it learn way faster and find the right people.
   • Custom Segments: Build segments based on what your ideal customers are searching for or the sites and apps they visit. If you sell high-end cameras, a custom segment could be people who searched “professional mirrorless cameras” or who browse photography review sites.
   • Your Data (First-Party Data): Upload your customer lists for remarketing. This is the most valuable signal you can provide, as it allows the AI to find new users who look just like your existing customers.
   • Interests & Detailed Demographics: Layer in relevant interest categories to add more context.

Common Mistake: So many advertisers just dump all their assets into one generic group. That’s a huge waste. Each asset group should be laser-focused on a theme. If you sell three different kinds of software, you should have three different asset groups, each with its own tailored creative and landing page. This kind of segmentation explicitly tells the AI, “these assets and headlines are for this audience,” which is far more effective than forcing it to guess and diluting its performance.

Optimizing Bidding and Budget for AI Agent Performance

PMax automates a lot of the bidding and budget work, but your strategic decisions are what make it actually work. The whole point is to give the AI enough data and rope to hit your goals efficiently.

Configuring Bidding and Budget

1. Set Your Budget: Decide on an average daily budget. Just keep in mind that PMax campaigns have permission to spend up to double your daily budget on some days to grab conversion opportunities, though your total monthly spend won’t go over your daily budget times the number of days in the month.
2. Choose a Bidding Strategy:
   • Maximize Conversions: This is the default and usually the best place to start with PMax, especially if you have your conversion actions dialed in. The AI will just try to get as many conversions as it can for your budget.
   • Maximize Conversion Value: If your conversions have different financial values (like a $500 purchase versus a free newsletter signup), this is the strategy to use. You just have to make sure you’ve assigned those values correctly in your conversion tracking.
   • Target CPA (Cost Per Acquisition) or Target ROAS (Return On Ad Spend): After the campaign has some history, usually 30+ conversions in the last 30 days, you can add a tCPA or tROAS target. This gives the AI a specific cost or return goal to aim for.
3. Define Conversion Goals: Double-check that your conversion actions are set up and tracking properly in Google Ads. This is literally the target the AI is optimizing toward. If your conversion data is wrong, the AI is flying blind.

Expected Outcome: With a reasonable budget and a conversion-based bidding strategy, the AI will start its learning phase, looking for patterns in users who are likely to convert, which should be followed by more stable performance. If the AI doesn’t have enough budget, it can’t run enough experiments to learn. A common mistake is setting a budget that’s way too low for the tCPA or tROAS you want which just starves the system of the data it needs to figure anything out.

Monitoring and Iterating Based on AI Insights

PMax campaigns don’t give you the same granular controls you’re used to from traditional campaigns. Your job shifts to monitoring the right metrics and making smart, strategic nudges based on the AI’s feedback.

Analyzing Performance and Making Adjustments

1. Review Asset Group Performance:
   • Go to your PMax campaign in Google Ads and click on Asset groups.
   • There’s a “Performance” column that rates your assets as “Low,” “Good,” or “Best.”
   • Anything rated “Low” is what the AI is ignoring or struggling with. Get it out of there and replace it with new creative variations.
2. Examine Audience Insights:
   • Under your PMax campaign, click on Insights.
   • Check out “Audience Insights” and “Consumer Interests.” This is where the AI tells you what kinds of users it’s finding and what they’re into. You can use this data to refine your audience signals or even inspire new product ideas.
3. Check Placement Reports:
   • Even though PMax is about broad reach, you can still see where your ads are showing up. Look for “Placement Insights” under the Insights tab.
   • This report shows you which channels (like YouTube, Display, Search) are actually driving results. If one channel is always doing poorly, you might have a problem with the creative assets designed for that format.
4. Adjust Budget and Targets: Based on what you’re seeing in the data, you might decide to push more budget to a high-performing campaign or asset group. Or, if the AI is consistently missing your goals, you might need to adjust your Target CPA/ROAS to be more realistic.

I’ve found that small, consistent tweaks based on the AI’s signals work far better than big, sudden changes. The AI needs a stable environment to learn. If you keep ripping everything up every week, you’re just resetting its progress. It’s a feedback loop, which the IAB’s 2024 “AI in Advertising” report also points out is the key to getting these campaigns to fly.

The entire advertising game is getting tangled up with AI. By structuring Performance Max campaigns the right way, feeding them rich and varied creative, and giving the AI smart signals to start with, marketers can find a level of efficiency and discovery for their brands that wasn’t possible before.