All posts
Google Ads
Performance Max for Ecommerce: A Campaign Structure That Scales Profitably
Performance Max promises something every ecommerce team wants: one campaign that finds customers across Search, Shopping, YouTube, Display, Discover, Gmail, and Maps, bidding automatically toward your goals. For many stores it delivers. For many others it quietly spends on the products and placements that were already selling, reports a healthy return, and leaves the catalog's real growth untouched.
The difference is rarely the algorithm. It is the inputs. Performance Max optimizes toward the conversion data, product data, assets, and structure you give it. Feed it a single campaign with every product, a blended revenue target, and generic creative, and it will do exactly what you asked: chase the easiest revenue.
This is the structure I use to make Performance Max work for ecommerce businesses with real margins, real inventory constraints, and more than a handful of products.
Before structure: fix the inputs
No campaign structure can compensate for bad inputs. Before building anything, check three things.
1. Conversion tracking
Smart Bidding learns from your conversion data. Duplicated purchases, missing purchases, or revenue that includes the wrong values will teach the system the wrong lessons. A deduplicated purchase event with accurate value and currency is non-negotiable. My guide to GA4 ecommerce tracking with Google Tag Manager covers a setup that holds up.
2. The product feed
In ecommerce, Performance Max is largely a Shopping campaign with extra inventory. Product titles, categories, images, and attributes decide which searches products can match. Disapproved products cannot serve at all. If the feed is weak, start with Google Merchant Center feed optimization before changing a single bid.
3. Business goals
Decide what you are actually optimizing for. Revenue at a target return on ad spend, profit, new customers, or clearing specific inventory are different goals and need different settings.
The core idea: segment by what the business values
Google's own Performance Max guidance recommends consolidation, and fewer campaigns usually do perform better because each one gets more data. But consolidation without segmentation means one target applies to every product, whether it carries a 70% margin or a 15% margin.
The answer is not dozens of campaigns. It is a small number of campaigns separated only where the business needs different targets or budgets. Custom labels in the product feed make this possible.
A practical structure for most mid-size catalogs:
Hero products. Best sellers with healthy margins. Their own campaign, so budget is never starved by the rest of the catalog.
Core catalog. Everything else with normal margins, at a standard return target.
Low-margin or price-sensitive products. A higher return target or a tighter budget, so they do not absorb spend that profitable products could use.
Clearance or seasonal. Separate when there is a specific goal, such as moving inventory before a deadline.
Within each campaign, product groups and listing groups can be refined further, but the campaign split itself should map to decisions the business actually makes.
Asset groups: give the system themes, not leftovers
Asset groups hold the headlines, descriptions, images, videos, and audience signals Performance Max uses outside Shopping placements. Many accounts create one asset group with generic brand assets. That limits how well the campaign can match creative to intent.
Build asset groups around themes, usually product categories or customer use cases:
Product-specific headlines that describe what the shopper is buying, not just the brand slogan
Images that show the products in the group, in square, landscape, and portrait ratios
At least one video. If you do not supply one, Google may generate one automatically from your images, and it rarely represents the brand well
Audience signals built from customer lists, site visitors, and relevant search themes
Audience signals are hints, not restrictions. They help the campaign start in the right place, which matters most in the first weeks.
Bidding: start conservative, then let it learn
New Performance Max campaigns need conversion volume to learn. A sensible progression:
Launch with Maximize Conversion Value and no target, or a target set slightly below the historical return, so the campaign can gather data.
Wait through the learning period. Judging a campaign after a few days usually leads to changes that reset learning.
Introduce a target ROAS once there is enough conversion history, then adjust in small steps rather than large jumps.
If acquiring new customers matters more than repeat buyers, use the new customer acquisition goal with customer lists. Otherwise, the campaign will happily retarget people who would have purchased anyway.
Protect brand search and measure incrementality
Performance Max can serve on your own brand searches, where conversions are cheap and would often happen anyway. That inflates reported return and hides how the campaign performs on genuinely new demand.
Use brand exclusions where appropriate, and run a separate brand Search campaign with its own budget.
Review search term insights regularly to see which themes drive conversions.
Compare total business revenue, not just platform-reported revenue, when budgets change.
Where volume allows, run holdout or geo tests to measure what the campaign truly adds.
Reading Performance Max reports without fooling yourself
The campaign reports a return on ad spend. The business cares about profit. Close the gap:
Report by product and custom label, not just at campaign level, to see where spend actually goes.
Use the placement and asset reports to find weak creative and replace it.
Check new versus returning customers. A campaign that only retargets looks efficient and grows nothing.
Track contribution margin, subtracting product cost, shipping, and returns, at least monthly.
Common Performance Max mistakes
One campaign for the whole catalog with a single return target
Launching before conversion tracking is verified
A weak feed with untouched supplier titles
No video, leaving Google to generate one
Changing targets every few days and resetting learning
Ignoring brand search inflation in reported results
Frequently asked questions
How many Performance Max campaigns should an ecommerce store run?
As few as possible while still giving products with different margins or goals their own targets and budgets. For most mid-size stores that means two to four campaigns.
How long does Performance Max take to learn?
Expect a learning period of several weeks, depending on conversion volume. Avoid major changes to targets, budgets, or structure during that time.
Should I keep Standard Shopping campaigns with Performance Max?
It depends on the account. Some teams keep Standard Shopping for tight control over specific products or tests, but the two compete for the same inventory, so the structure needs to be deliberate.
Why is my Performance Max return high but sales flat?
Often because the campaign is capturing brand searches and returning customers who would have purchased anyway. Check new customer share and brand search exclusions.
Where to start
Verify conversion tracking, fix the feed, then add margin-based custom labels. With those three in place, even a simple two-campaign structure usually outperforms a single catch-all campaign.
To see where your own campaigns stand, try my free ROAS calculator. It shows your break-even ROAS and the target ROAS to set for a profit goal.
I plan and manage Google Ads, Merchant Center, and the tracking behind them for ecommerce brands. See the creative side in the IOGEAR campaign case study, browse selected projects, read more about Jeffrey Hakim, or contact me about your account or a growth marketing role.
