Performance Max for E-Commerce: The Truth Nobody Tells You

Performance Max is Google's most profitable product. For Google.

For your e-commerce brand, it might be your biggest blind spot. We've managed PMax campaigns across 50+ e-commerce accounts, and here's what we can tell you with certainty: most brands are running PMax wrong, and Google has zero incentive to tell you that.

We think PMax is overrated for most new e-commerce brands. That's a controversial opinion in the Google Ads world. But we have the data to back it up, and by the end of this post, you'll understand exactly why.

If you suspect your PMax is underperforming but can't prove it, book a free audit call here. We'll pull the network breakdown most brands never see. Otherwise, keep reading.

What Google Tells You vs. What Actually Happens

Google pitches PMax as an AI-powered campaign that finds customers across all networks - Search, Shopping, Display, YouTube, Gmail, Discover, and Maps. One campaign to rule them all. Sounds incredible.

Here's the reality across 50+ accounts we've managed: PMax is primarily an automated Shopping campaign with a side of everything else. For e-commerce, roughly 60-80% of PMax spend goes to Shopping placements. The remaining 20-40% goes to Display, YouTube, Gmail, and Discover.

The part Google doesn't mention? You can't control the network split. You can't say "spend 90% on Shopping and 10% on YouTube." PMax decides. And PMax has an incentive to show ads everywhere because that increases Google's total revenue across its properties.

We learned this the hard way on an early PMax campaign. The team trusted the blended ROAS number. It looked great. It was a lie.

The Hidden Tax Nobody Talks About

We analyzed one e-commerce account where PMax was reporting a 4.5x ROAS overall. The client was thrilled. Then we pulled the network breakdown.

Shopping placements: 7.2x ROAS. Display: 0.8x. YouTube: 1.1x.

The Shopping performance was subsidizing the waste on other networks, and the blended number looked acceptable. That brand was paying a 30% tax for Google to run Display and YouTube ads they never asked for. On $20,000/month in PMax spend, that's $6,000/month going to placements that lost money. Every month.

But the real win came from something we found in the search themes report - more on that below.

How PMax Steals Credit From Your Brand Campaign

This is the biggest issue with PMax for e-commerce, and what nobody tells you about during the sales pitch.

PMax has priority over Standard Shopping campaigns and competes with your Search campaigns. When someone searches for your brand name - someone who was going to buy from you anyway - PMax shows them a Shopping ad, they click, they buy. PMax reports it as a PMax conversion.

Think about that. You just paid for a click from someone who was already your customer. That's not growth. That's a rebate to Google.

The Test That Reveals The Truth

Here's the simple test we run on every account with PMax:

1. Check your brand Search campaign metrics before and after PMax launched. Did brand Search impression share drop?
2. Pull PMax search themes (Insights tab or via the API). How many conversions come from brand queries?
3. Look at total account conversions - not just PMax. Did total conversions actually increase, or did they just shift from Search to PMax?

In our experience across our accounts in the last quarter, 30-50% of PMax conversions in e-commerce accounts come from brand-related queries. That's not incremental growth. That's PMax taking credit for sales your brand campaign would have captured at a lower CPC.

Remember the search themes we mentioned earlier? On a healthy-looking PMax account, a large share of conversions often turn out to be brand queries. Apply brand exclusions and the true non-brand ROAS can fall to barely profitable.

Brand exclusions in PMax are non-negotiable. If your PMax doesn't have them, you're subsidizing Google's revenue report, not yours. Book a free call for help setting this up correctly - it takes 10 minutes but the impact is immediate.

Brand Exclusions: The 30-Second Fix Worth Thousands

Google now allows brand exclusions in PMax. Go to campaign settings, find Brand Exclusions, add your brand name and common misspellings.

Two things will happen. First, your PMax ROAS will drop - sometimes by 40-60%. Don't panic. You just removed the easy, high-converting brand traffic. Second, you'll see the true non-brand performance, which is the metric that actually measures growth.

Give the campaign two weeks to readjust after adding exclusions. The algorithm needs time to reallocate spend toward non-brand queries. If non-brand PMax ROAS is still above your profitability threshold, great. If it's below, you now know your PMax was propped up by brand traffic all along. Painful, but better to know.

The Right Way to Structure PMax for E-Commerce

Most brands create one PMax campaign with one asset group and throw all their products in. This is the "one Shopping campaign for everything" mistake, but worse - because PMax also controls your Search and Display presence.

Asset Group Strategy That Actually Works

Create separate asset groups for each major product category. If you sell men's shoes, women's shoes, and accessories, those need three different asset groups with:

• Category-specific product listings using listing groups
• Category-specific creative assets - lifestyle images featuring that specific category, not generic brand shots
• Category-specific headlines and descriptions - not "Shop Our Collection" copy that says nothing
• Category-specific audience signals - custom segments based on relevant search behavior and competitor URLs for that category

Audience signals don't restrict who sees your ads - PMax shows ads to anyone it wants. But signals tell the algorithm where to start looking. Strong signals across 3 accounts we tested cut the learning period from 4 weeks to about 10 days.

The PMax + Standard Shopping Combination

Here's the structure we use for most e-commerce accounts that works consistently well:

Standard Shopping for your top performers. Your best-selling products where you want full control over bids, search terms, and budget. Target your top 20-30% of products by revenue here.

PMax for product discovery and the long tail. PMax is genuinely good at finding new customers for products that don't have enough individual data for Standard Shopping to optimize. Let it work the catalog you wouldn't have targeted manually.

This combination gives you control where it matters most and lets PMax do what it actually does well.

Creative Assets: Don't Give Google an Excuse to Make Garbage

If you only upload product images, PMax will create auto-generated ads for Display and YouTube. They will be ugly. They will perform poorly. PMax will still spend money showing them.

Minimum requirements to avoid auto-generated nightmares:

• 5 landscape + 5 square images - lifestyle shots, not product-on-white
• At least one video - even a simple 30-second product showcase. Google's auto-generated videos from your images are almost always terrible.
• 15 headlines and 5 descriptions - max them out. Include benefit-driven copy. PMax tests combinations automatically, so give it good raw material.

How to Honestly Evaluate Your PMax

Here's our framework, measured across every PMax account we manage:

1. Total account performance, not just PMax metrics. Did overall revenue and ROAS improve after PMax launched? Or did conversions just shift from other campaigns?
2. New customer acquisition. If PMax primarily converts existing customers and brand searchers, it's not doing its job.
3. Non-brand PMax ROAS vs. Standard Shopping ROAS. They should be in the same ballpark. If PMax non-brand is dramatically lower, the algorithm is struggling.
4. Placement quality. Pull the placement report. If you see your luxury handbag ads on kids' puzzle games, your PMax needs intervention.

The Bottom Line

PMax is a powerful tool when you understand what it's actually doing. It's not a replacement for a thoughtful campaign strategy. The brands getting real results from PMax use brand exclusions from day one, maintain Standard Shopping for top products, build category-specific asset groups, and evaluate on incremental contribution - not blended ROAS.

The brands handing everything to PMax and hoping for the best? They're the ones consistently subsidizing Google's bottom line instead of their own.

We've restructured PMax campaigns for brands spending $5,000 to $100,000/month. The pattern is always the same: add brand exclusions, check the network breakdown, separate your top products into Standard Shopping. Book a free PMax audit call and we'll show you exactly what your PMax is hiding.

Frequently asked questions

Is Performance Max good for e-commerce?

Yes, when structured and policed, and expensive when left alone. PMax is the biggest revenue driver in most product accounts, but by default it takes credit for brand searches you would have won anyway and spends where you cannot see. The difference between the two outcomes is setup and monitoring, not luck.

Does Performance Max take credit for branded sales?

By default, yes. PMax serves on searches for your own brand name and reports those near-guaranteed sales as its own performance, which inflates its ROAS and hides how the campaign really performs on new customers. Adding brand exclusions takes minutes and is the single most clarifying change you can make.

How should Performance Max be structured for a store?

Not as one campaign for the whole catalog. Group products so that margins and price points are consistent within a campaign or asset group, feed it real audience signals, and supply proper creative assets so Google does not assemble poor ads on your behalf. Structure is what turns PMax from a black box into a tool.

How do I know if my PMax is actually performing?

Evaluate it with brand traffic excluded and against your break-even ROAS. Check where spend actually went in the channel breakdown, compare non-brand revenue before and after changes, and be suspicious of a ROAS that looks too good, because it usually contains your brand demand repackaged.