Performance Max for Ecommerce: How to Structure PMax Campaigns That Actually Deliver ROAS

14 min read · Ecommerce Growth · Last updated July 2026

Quick answer: PMax underdelivers when set up as one catch-all campaign. The stores getting strong ROAS structure by product category, provide audience signals from real customer data, restrict URL expansion, and maintain tight negative keyword lists at the account level.

Introduction

Performance Max launched in 2021. It promised automated, cross-channel reach — Search, Shopping, Display, YouTube, Gmail, Maps — from a single campaign. For many ecommerce stores, the promise has not matched the reality.

The most common outcome for a poorly configured PMax campaign: strong impression volume, mediocre conversion rates, inflated ROAS claims from brand cannibalization, and very little understanding of where the budget actually went.

But PMax is not inherently broken. Stores that structure it correctly — by product category, with real audience signals, with brand exclusions, and with account-level negative keywords — see it perform alongside or better than Standard Shopping at scale.

This guide covers how to set up PMax correctly, what not to do, and how to monitor performance when Google hides most of the placement data.

What you’ll learn:
– Why most PMax campaigns underperform (and the exact setup errors)
– Asset group structure that matches how buyers shop
– How audience signals work (and what happens without them)
– Brand exclusions and negative keywords in PMax
– How to monitor PMax performance without full placement visibility

Table of Contents

1. What PMax Actually Is (And Is Not)
2. The 5 Most Common PMax Setup Mistakes
3. Asset Group Structure by Product Category
4. Audience Signals: Why They Matter More Than Most Think
5. URL Expansion: Turn It Off or Restrict It
6. Brand Exclusions: Non-Negotiable
7. Negative Keywords in PMax: Limited but Critical
8. Creative Assets: What You Must Provide
9. Budget and ROAS Target Calibration
10. How to Monitor PMax Without Placement Data
11. PMax Setup Checklist
12. ROAS Target Calculator by Product Margin
13. FAQ

What PMax Actually Is (And Is Not)

PMax is Google’s fully automated campaign type. You provide:
– A product feed (via Google Merchant Center)
– Creative assets (headlines, descriptions, images, videos)
– Audience signals (suggested audiences for the algorithm to learn from)
– A budget and ROAS target

Google decides:
– Which products to show
– On which channels (Search, Shopping, Display, YouTube, Gmail, Maps, Discover)
– At which bids
– To which audiences

What PMax is not: A replacement for a well-managed Standard Shopping campaign. PMax trades control for reach. Standard Shopping gives you search term data, manual bid control, and clean campaign structure. PMax gives you more placements and a theoretically smarter algorithm — at the cost of visibility.

When PMax is appropriate:
– You have 300+ purchases per month (algorithm has sufficient data)
– Your product feed is fully optimized
– You have customer data (email list, purchaser list) for audience signals
– You have a well-established negative keyword list to port to account level

When to hold off on PMax:
– New store (under 100 purchases/month)
– No customer data for audience signals
– Unoptimized product feed

Key takeaway: PMax works best as an amplifier on a strong foundation — not as the foundation itself.

The 5 Most Common PMax Setup Mistakes

Mistake 1: One catch-all campaign with one asset group

A single PMax campaign covering all products gives Google no structural guidance. The algorithm does not know that your $180 premium product should be shown to a different audience than your $25 accessory.

Fix: Create separate campaigns (or at minimum, separate asset groups within a campaign) by product category.

Mistake 2: No audience signals

Without audience signals, PMax starts cold — it must learn from scratch who buys from you. This learning period wastes budget on irrelevant audiences.

Fix: Always provide audience signals from your own data: customer email list, recent purchasers, website visitors.

Mistake 3: URL expansion enabled without restriction

URL expansion allows Google to send traffic to any page on your site — including pages you would never advertise to, like your careers page, blog articles, or login page.

Fix: Turn URL expansion off, or set expansion to “only URLs in your feed or specified pages.”

Mistake 4: No brand exclusions

PMax will happily serve ads on branded search terms — meaning when someone searches your brand name, they see a PMax Shopping ad instead of (or in addition to) your branded Search campaign. This inflates PMax ROAS by claiming credit for conversions that would have happened anyway.

Fix: Exclude your brand name from PMax using the brand exclusion feature (available in PMax settings).

Mistake 5: No account-level negative keywords

PMax has limited negative keyword functionality at the campaign level. But account-level negative keyword lists apply to PMax campaigns.

Fix: Before launching PMax, build a comprehensive negative keyword list at the account level — especially irrelevant queries, competitor brand names (if you do not want to bid on them), and content/DIY queries.

Asset Group Structure by Product Category

The most important structural decision in PMax is how to organize your asset groups.

Asset groups serve two purposes:
1. They organize which products are shown together (via product feed filters)
2. They house the creative assets (headlines, images, videos) relevant to those products

Recommended structure:

Option A — Separate campaigns by category (most control):
– PMax Campaign: Running Footwear
– PMax Campaign: Casual Footwear
– PMax Campaign: Accessories

Each campaign has its own budget and ROAS target. High-performing categories can have budgets scaled independently.

Option B — One campaign, multiple asset groups (simpler):
– PMax Campaign: All Products
– Asset Group 1: Running Shoes (product filter: category = Running, assets = running-specific)
– Asset Group 2: Casual Shoes (product filter: category = Casual, assets = lifestyle)
– Asset Group 3: Accessories (product filter: category = Accessories, assets = accessory-focused)

Option A gives more control. Option B is easier to manage for smaller stores. For stores with significant revenue differences between categories, use Option A.

What each asset group needs:
– Product group filter (limiting which products appear in this asset group)
– Category-specific creative assets
– Category-specific audience signals

Audience Signals: Why They Matter More Than Most Think

Audience signals in PMax are not targeting — they are suggestions. You tell Google: “Buyers who look like these people are more likely to convert.” Google uses this to prioritize who to show ads to in the early learning phase.

The three best audience signals for ecommerce:

1. Customer Match List (your email list)

Upload your customer email list to Google Ads (Tools → Audience Manager → Customer Match). Use this as your primary audience signal. Google matches emails to Google accounts and uses the profile of your customers to find similar buyers.

Minimum list size: 1,000 matched emails. Larger lists (5,000+) give the algorithm better signal.

2. Remarketing Audiences from GA4

Your GA4 audiences (cart abandoners, product viewers, past purchasers) are the highest-intent signals you can provide. These audiences reflect people who have already engaged with your store.

Which to include: Past purchasers (strongest positive signal), cart abandoners, all visitors in last 90 days.

3. In-Market Audiences

If your own data is limited, supplement with Google’s in-market audiences relevant to your category (e.g., “In-market: Athletic Footwear” for a shoe store, or “In-market: Nutritional Supplements” for a supplement brand).

In-market audiences are weaker signals than first-party data — use them to supplement, not replace, your own customer data.

URL Expansion: Turn It Off or Restrict It

URL expansion allows PMax to send traffic to URLs beyond your product and collection pages. Without restriction, Google may send buyers to your About page, a blog post, or your returns policy page.

None of those pages convert buyers at an acceptable rate.

How to control URL expansion in PMax:

1. In your PMax campaign settings, find “URL expansion”
2. Options:
   – Off: Only URLs in your product feed (recommended for most stores)
   – On, all URLs: Google can send to any page (do not use this)
   – On, with URL exclusions: Keep expansion on but add exclusion rules for non-commercial pages

Recommended setting: Off for most ecommerce stores. If you want expansion for a specific reason (e.g., you have high-value landing pages not in the feed), use “On with URL exclusions” and exclude /blog/, /about/, /contact/, /careers/, /account/, and any other non-commercial paths.

Brand Exclusions: Non-Negotiable

Brand exclusions prevent your PMax campaign from serving on searches that include your brand name.

Why this matters: When someone searches “YourBrand running shoes,” they were already going to buy from you. If PMax claims credit for that conversion, it inflates your PMax ROAS artificially — and masks the true performance of the campaign on net-new customers.

How to add brand exclusions in PMax:

1. Campaign settings → Brand exclusions
2. Search for your brand name and add it
3. Also add common misspellings and abbreviations of your brand name

Run a separate branded Search campaign targeting your own brand name with high bids. This campaign will typically show 10–20× ROAS (because these are warm buyers). Keep it separate so its performance does not inflate or subsidize your prospecting campaigns.

Negative Keywords in PMax: Limited but Critical

PMax does not support campaign-level negative keywords the same way Search campaigns do. The two methods available:

Method 1: Account-Level Negative Keyword Lists

In Google Ads, navigate to Tools → Shared Library → Negative Keyword Lists. Create a list and add it to your PMax campaigns.

This is the primary method. Build this list before launching PMax, then update it monthly based on the (limited) search term data available.

Method 2: Search Themes (Positive direction only)

PMax allows you to add “search themes” — topics or keywords you want the campaign to focus on. This is not negative keyword targeting, but it does influence the direction of the algorithm toward relevant searches.

How to view search terms in PMax: PMax does show some search term data under Insights → Search Terms. The data is filtered (Google only shows terms that meet a minimum threshold) but it provides enough visibility to identify obvious irrelevant queries.

Creative Assets: What You Must Provide

PMax requires creative assets across all channels. Google will generate assets automatically if you do not provide them — the auto-generated assets are typically low quality and should not be relied on.

Required assets per asset group:

On video: If you do not provide a video, Google generates one from your images and text. The auto-generated video is rarely good. Provide at least one 15–30 second product showcase video. Stores that provide video see 15–25% higher PMax ROAS than those relying on auto-generated video.

Asset strength signal: In your PMax campaign, Google shows an “asset strength” rating (Poor / Good / Excellent) per asset group. Target “Excellent” — it correlates with better auction eligibility.

Budget and ROAS Target Calibration

Starting budget: PMax needs enough budget to generate 50+ conversions in its first 30 days to exit the learning phase efficiently. For a store with a $75 AOV and a target CVR of 2.5%, you need approximately 2,000 clicks at $1.50 CPC = $3,000/month minimum. Scale from there.

Initial ROAS target: Set your initial Target ROAS at your break-even ROAS × 1.2 — slightly above break-even to give the algorithm room to find conversions without being overly restrictive.

Example: 50% margin → break-even ROAS = 2.0 → initial PMax Target ROAS = 2.4.

Scaling ROAS targets:
– After 30 days with consistent conversions, increase Target ROAS by 10%
– Wait 2 weeks and review volume vs. efficiency trade-off
– Increase again if volume is maintained at higher ROAS
– If volume drops significantly (>30%), lower back to the previous target

Do not set Target ROAS too aggressively early. A PMax campaign with a 6× Target ROAS for a new campaign with no conversion history will not spend — the algorithm cannot find enough conversions at that threshold to use its budget.

How to Monitor PMax Without Full Placement Data

PMax hides most placement data. Here is how to monitor performance despite limited visibility:

1. Asset Group performance
Navigate to Asset Groups. Check the “Performance” column for each asset group. Groups labeled “Low” need creative refresh or audience signal updates.

2. Insights tab
The Insights tab in PMax shows:
– Search themes driving conversions (limited but useful)
– Audience segments that are converting
– Top-performing creative assets

Review this weekly during the first 90 days.

3. Placement report (partial)
Navigate to Placements report in your PMax campaign. This shows some (not all) placement data — primarily Display and YouTube placements. Exclude irrelevant placements you see here (non-commerce websites, YouTube channels unrelated to your niche).

4. Compare PMax to Standard Shopping in the same period
Run a Standard Shopping campaign alongside PMax. Compare new customer acquisition rate, blended ROAS, and CVR. PMax that only performs well by retargeting warm audiences is not finding new customers — it is just poaching them from your other channels.

5. Use GA4 to validate
Compare purchase counts in GA4 against PMax attributed conversions in Google Ads. A significant gap suggests PMax is claiming credit for conversions that were not truly driven by PMax ad interactions.

PMax Setup Checklist

ROAS Target Calculator by Product Margin

FAQ

Should PMax replace Standard Shopping?
No — run them alongside each other. Standard Shopping gives you search term data, manual control, and clean ROAS measurement. PMax gives you broader reach and automated bidding. Together, they complement each other. If PMax is the only Shopping campaign, you lose the visibility you need to manage the product feed effectively.

How do I know if PMax is cannibalizing my other campaigns?
Compare the number of conversions in your Standard Shopping and branded Search campaigns before and after launching PMax. If branded Search conversions drop while PMax conversions rise, PMax is taking credit for already-warm buyers. Add brand exclusions and compare again.

What happens during the PMax learning period?
During the first 2–6 weeks, expect higher CPCs, lower ROAS, and inconsistent volume. The algorithm is testing audiences and placements. Do not make major changes (budget adjustments over 20%, ROAS target changes) during this period — it resets learning.

Can I run PMax without a product feed?
Yes, PMax can run without a Merchant Center feed — but for ecommerce, always connect a product feed. Without the feed, PMax cannot show Shopping ads, which are typically the highest-converting PMax placements.

What if I cannot provide a video?
Provide images and Google will auto-generate a video. The auto-generated video will be below average quality but better than nothing. Prioritize getting a real video into your account within the first 90 days — even a well-lit smartphone product video outperforms Google’s auto-generation.

How often should I refresh PMax creative assets?
Review asset performance monthly. Replace assets labeled “Low” with new creative. The overall creative set should be fully refreshed every 60–90 days to prevent ad fatigue across display and YouTube placements.

Conclusion

PMax is a powerful tool — when built correctly. The stores that get strong ROAS from it treat setup seriously: category-segmented asset groups, real audience signals from customer data, brand exclusions, and account-level negative keywords all in place before the first dollar is spent.

The stores that get poor results from PMax treated it as a plug-and-play solution. One campaign, no audience signals, URL expansion enabled, brand terms unexcluded. The algorithm does its best, but it works with what you give it.

Give it the right structure and data, and PMax scales reach efficiently. Give it nothing, and it spends your budget learning things you should have told it at the start.

Ready to Grow Your Ecommerce Store?

Our ecommerce growth team handles SEO, email, Shopping, and CRO as one connected system. → Talk to Our Ecommerce Team

Written by the Ignited Nepal ecommerce team. ignitednepal.com