Performance Max for Ecommerce: When It Works and When It Doesn't
PMax is Google's most pushed campaign type — and the one most likely to waste your budget if you don't understand what it's actually doing.
Jakob Sperber
Director
Google Ads
/
Performance Max is Google's most pushed campaign type. Every Google rep will tell you to run it. Every automated recommendation nudges you toward it. And if you're not careful, it'll cannibalise your brand search traffic, inflate your reported ROAS, and leave you thinking you're winning while your actual acquisition efficiency drops.
PMax can work. But only when you understand what it's actually doing and structure it to avoid the traps.
What PMax Actually Is
Performance Max is a single campaign type that runs ads across every Google surface: Search, Shopping, Display, YouTube, Discovery, Gmail, and Maps. You provide assets (images, videos, headlines, descriptions) and Google's algorithm decides where to show them and to whom.
Google positions this as "let the AI do the work." The reality: you're giving Google full control over placement, bidding, and audience targeting with minimal visibility into what's actually happening.
When PMax Works
Broad product catalogues (50+ SKUs). PMax excels at finding buyers across a wide catalogue where manually managing Shopping campaigns would be impractical.
Sufficient conversion data (50+ conversions/month). The algorithm needs data to learn. Below this threshold, it's guessing.
Strong product feed. PMax is only as good as the data you feed it. Titles, descriptions, images, pricing — all need to be optimised.
When you've maxed out manual campaigns. If your standard Shopping and Search campaigns are capped by budget or audience, PMax can find incremental volume.
When PMax Fails
The Brand Search Problem (#1 Issue)
This is the biggest trap. PMax loves to serve ads on your own brand terms — searches from people who already know you and would have bought anyway. This makes PMax ROAS look incredible (8x, 10x, 15x) because brand search converts at extremely high rates.
But those conversions aren't incremental. PMax is claiming credit for sales that would have happened through free organic listings or a cheap brand Search campaign. Your PMax dashboard shows "great performance" while your actual new customer acquisition is flat or declining.
How to detect it: Run a separate brand Search campaign alongside PMax. Compare total brand conversions before and after launching PMax. If brand conversions in PMax grew but total brand conversions stayed flat, PMax is cannibalising, not acquiring.
Small Catalogues
With fewer than 20 SKUs, standard Shopping gives you more control. PMax spreads budget across placements you may not want (Display, YouTube) and you can't opt out of specific surfaces.
Low Conversion Volume
Below 30–50 conversions per month, PMax can't optimise properly. You'll be stuck in learning phase, burning budget with inconsistent results.
How to Structure PMax Properly
Asset Groups by Product Category
Don't dump everything into one asset group. Separate by category, margin tier, or product type. This gives the algorithm clearer signals about which products to push to which audiences.
Audience Signals (Not Targeting)
PMax audience signals are suggestions, not targeting. Add your customer lists, website visitors, and in-market audiences as signals. The algorithm uses them as starting points but will expand beyond them.
Final URL Expansion: Turn It Off
By default, PMax can send traffic to any page on your site. For most ecommerce brands, this means ad clicks landing on blog posts, about pages, or other non-converting pages. Turn off final URL expansion and direct traffic to your product and collection pages.
Feed Quality Is Everything
PMax Shopping performance lives and dies on your product feed.
Titles: Front-load with the most important keywords. "Women's Running Shoes — Lightweight, Size 6-11" beats "Nike Pegasus 41."
Custom labels: Use them for margin-based bidding. Label products as high/medium/low margin so you can set different ROAS targets per group.
Images: Clean, high-quality product shots on white backgrounds. Lifestyle images for Discovery and YouTube asset groups.
Supplemental feeds: Use these to override or enhance your primary feed without touching your store backend.
Bidding: Set Targets From Your Break-Even ROAS
Your tROAS target should come from your unit economics, not Google's suggestions. Calculate your break-even ROAS (1 / CM1%) and set your target above that based on your profit goals.
Google will recommend "maximise conversion value" with no target. This gives the algorithm permission to spend at any efficiency level. Always set a tROAS floor. Start conservative, then relax it as you gather data.
For the full framework on what constitutes a good ROAS by channel, check our benchmarks guide.
PMax vs Standard Shopping
The honest answer: run both.
Standard Shopping gives you control — you can see search terms, adjust bids by product, and segment by performance. PMax gives you reach — it accesses placements Shopping can't and finds audiences you wouldn't have targeted manually.
The common approach: run standard Shopping for your top 20% of SKUs (where you want granular control) and PMax for the long tail (where automation adds efficiency).
Whatever you do, set your total budget from your P&L — not from what Google recommends. And track new customer CAC separately from blended metrics. PMax makes everything look good on the surface. Your job is to look underneath.
The Bottom Line
PMax is a powerful tool in the right hands. It's a budget trap in the wrong ones. Use it when you have sufficient data, a strong feed, and the ability to monitor for brand cannibalisation. Don't use it as a set-and-forget replacement for a properly structured Google Ads account.
