Performance Max in 2026: What's Actually Working (And What Google Won't Tell You)

I've got a love-hate relationship with Performance Max, and at this point I think Google does too. PMax is the campaign type they keep pushing, the one that's supposed to "do everything" — Search, Shopping, Display, YouTube, Discovery, Gmail, Maps — all in one tidy little box. Just add assets and let the machine do its thing.

Except it's not that simple. It never was. And after managing PMax campaigns across everything from artisan food brands to industrial security equipment, I can tell you: the gap between a PMax campaign that crushes and one that hemorrhages money comes down to about five decisions. That's it. Five things.

Let's get into it.

First: When to Use PMax vs. Standard Shopping

This is the question I get more than any other, and the answer is annoyingly context-dependent. But here are my general rules after years of testing both:

Use PMax when:

• You have a product catalog with strong margins and clean feed data
• You're willing to give Google real creative assets (not just stock photos and lazy headlines)
• You have enough conversion volume — I want at least 30 conversions/month in the campaign for the algorithm to learn properly
• You need reach beyond just Shopping (which, for most eCommerce brands, you do)

Use Standard Shopping when:

• You need granular control over individual products or product groups
• Your budget is tiny and you can't afford PMax's "learning" phase eating into your spend
• You sell products where the visual matters more than the algorithm — think highly specialized or niche items where Google's automation doesn't have enough data to be smart

Look, Google would love you to put everything in PMax. Of course they would. But I've seen plenty of accounts where Standard Shopping outperforms PMax on specific product lines, and vice versa. We often run both. The key is testing, not dogma.

The Asset Group Structure That Actually Works

This is where most PMax campaigns go wrong, and it usually happens at setup.

Google makes it easy to throw everything into one asset group and call it a day. One group with all your products, a handful of images, some auto-generated headlines, and boom — you're live. That's also how you end up with a PMax campaign that spends 80% of its budget on branded search and Display remarketing while telling you it's "driving conversions."

Here's how I structure asset groups:

Segment by product category or theme, not by audience. Each asset group should map to a distinct product line, with assets (images, videos, headlines, descriptions) that are specific to that line. An outdoor apparel brand shouldn't have gloves and sunglasses in the same asset group. The creative needs to match the product or the whole thing falls apart.

We did this for an eComm apparel client — broke their PMax into product-specific groups. Their gloves-focused PMax campaign hit 3.22x ROAS with 1,095 conversions at $27.52 CPA. Not earth-shattering, but steady and profitable. If we'd lumped all products together, the algorithm would've just thrown budget at whatever converted cheapest, which might not be what you actually want to sell more of.

Keep it to 5-7 asset groups max per campaign. More than that and Google doesn't have enough data per group to optimize. I'd rather have five well-fed asset groups than fifteen starving ones.

Audience Signals: What Matters and What Doesn't

OK so here's where I might get some disagreement from other PPC folks, but I'll say it anyway: audience signals in PMax are suggestions, not targeting. Google has been very clear about this. Your audience signals tell the algorithm where to start looking, but it will happily go way beyond those signals if it finds conversions elsewhere.

That said, good audience signals absolutely help. They give the algorithm a head start instead of making it figure everything out from scratch. Here's what I prioritize:

• Customer match lists: Upload your actual customer email lists. This is the strongest signal you can give PMax. Period.
• Website visitors who converted: Create a list of past converters and use it as a signal. The algorithm will find people who look like them.
• Custom segments based on search terms: Feed in the high-intent keywords your customers are actually searching. This one is underrated.
• In-market audiences: Useful for eCommerce, less useful for niche B2B. Test and see.

What doesn't matter much? Demographic targeting, interest-based audiences for broad categories, and honestly, most of the detailed targeting options Google offers. The algorithm will figure out the demographics on its own — your job is to tell it what a good customer looks like, not what age range they fall into.

The Black Box Problem (and How to Read the Signals)

Here's the thing Google won't tell you: PMax reporting is intentionally vague. You can't see which search terms triggered your ads (unless you dig into the search terms insights report, which is... incomplete at best). You can't see how much budget went to Shopping vs. Search vs. Display vs. YouTube. You can't see individual placement performance.

This is by design. Google wants you to trust the machine. And sometimes the machine deserves that trust — we've got a premium eyewear client hitting 9.89x ROAS at a $23.28 CPA on PMax. I'm not going to argue with those numbers.

But blind trust is a bad strategy. Here's how I keep tabs on what PMax is actually doing:

1. Check the asset performance ratings. Google grades your assets as "Low," "Good," or "Best." If most of your headlines are "Low," that's the algorithm telling you your creative stinks. Fix that before touching anything else.

2. Use the Insights tab religiously. It won't tell you everything, but it'll show you the top search categories driving spend, the audience segments converting, and shifts in performance. I check this weekly.

3. Look at placement reports. You can find where PMax is showing your ads — and you'll often find junk placements. Kids' game apps. Random parked domains. If you see garbage, add placement exclusions at the account level.

4. Compare PMax conversions against your actual backend data. PMax is notorious for claiming conversions that your CRM can't verify. Cross-reference everything.

Brand Exclusions: Non-Negotiable

If you take one thing from this entire post, let it be this: set up brand exclusions in your PMax campaigns.

By default, PMax will happily spend a chunk of your budget on branded search terms — people who are already searching for your company name. Those people were going to find you anyway. PMax claims credit for those conversions, and suddenly your ROAS looks incredible, but you haven't actually acquired a single new customer.

Google finally made brand exclusions available in PMax (it took them long enough). Use them. Force PMax to earn its keep by going after net-new customers instead of double-dipping on your existing brand awareness.

For context: one of our DTC food brand clients runs PMax non-brand and pulls a 4.07x ROAS on $85K in spend with 2,914 conversions. That's real, incremental revenue. Their branded Shopping campaign, run separately, does 17.27x ROAS at a $5.57 CPA. Both are great — but they're doing very different jobs, and if you combine them, you'll never know which is which.

The PMax Campaigns That Crush

Let me show you what good looks like across different verticals, because PMax isn't one-size-fits-all and the results vary wildly depending on execution.

Artisan eCommerce brand: PMax Shopping at 7.2x ROAS on sub-$9K spend. Small budget, focused product catalog, beautiful product photography. The algorithm had great assets to work with and a clean feed. That's the recipe.

Security equipment eCommerce: PMax for their gate operator product line — 4.46x ROAS on $833K spend, generating $3.7M in conversion value. This is what PMax looks like at scale when the fundamentals are right. Big catalog, good margins, and an asset group structure that maps cleanly to product categories.

Premium eyewear: Already mentioned this one, but 9.89x ROAS at $23.28 CPA. High-end products with strong creative and a customer match list that gave the algorithm exactly what it needed.

These aren't lucky breaks. They're the result of clean product feeds, intentional asset group structure, proper brand exclusions, and ongoing optimization. The boring stuff.

The PMax Campaigns That Waste Money

And then there's the other side. I've audited accounts where PMax was doing more harm than good, and the patterns are almost always the same:

• No brand exclusions. PMax spending 40-60% of budget on branded terms, claiming great ROAS, while the standard brand campaign it cannibalized was doing the same job for half the cost.
• One giant asset group. Everything jammed in together. Google defaults to the cheapest conversions, which usually means remarketing to people who already had items in their cart.
• Bad creative assets. Stock photos, generic headlines, no video. PMax needs fuel to work. If your assets are bland, the algorithm has nothing to test against.
• Product feed issues. Missing GTINs, bad product titles, wrong categorization. The feed is the foundation for Shopping-based PMax. Garbage in, garbage out.
• Not enough conversion data. If a PMax campaign gets fewer than 15-20 conversions a month, the algorithm is basically guessing. It needs volume to learn.

I audited an account last quarter where PMax was the only campaign type running. No standard Search, no standard Shopping, nothing. Just PMax. The business owner thought they were being "cutting edge." What they were being was blind. They had zero visibility into what was actually working, and their actual customer acquisition cost (when we mapped conversions back to the CRM) was nearly double what Google was reporting.

My PMax Setup Checklist for 2026

Whether you're launching a new PMax campaign or auditing an existing one, run through this:

1. Product feed audit. Titles optimized for search, correct GTINs, clean categorization, high-quality images. This is step zero.
2. Brand exclusions enabled. Run brand search separately. Don't let PMax claim credit it didn't earn.
3. Asset groups segmented by product line. Not one big bucket. Each group gets its own creative, its own products, its own landing page URL.
4. Real audience signals. Customer match lists, past converter lists, custom search term segments. Give the algorithm a running start.
5. Video assets. Even a simple 15-30 second product video. If you don't provide one, Google will auto-generate something awful from your images. Trust me on this.
6. Conversion tracking verified. Make sure you're tracking the right actions at the right value. PMax optimizes toward whatever you tell it is a conversion. If that's wrong, everything downstream is wrong.
7. Account-level placement exclusions. Block the junk placements proactively. There are published exclusion lists floating around — use them.
8. A standard Shopping or Search campaign running alongside. Never rely solely on PMax. You need something with real reporting to benchmark against.