Performance Max for ecommerce: when it works, how to steer

Performance Max looks like a black box from the outside: one budget, one goal, and Google decides the rest. That is why it performs very well in some stores and quietly burns money in others. This article covers the levers that genuinely exist, when Performance Max is the right choice for an ecommerce account, and which mistakes are usually behind a campaign that does not work.

What Performance Max means for an ecommerce account

Performance Max combines Search, Shopping surfaces, Display, YouTube, Discover, Gmail and Maps in one campaign. You supply product data and assets; the automation decides placement, audience and bid per auction.

What you do not get is a campaign you build once and never look at again. Control simply moves: away from keywords and bids, towards data, goals and inputs. Give it nothing to work with and the results are shaped by whatever signals happen to be available, and those are often cheap brand queries and remarketing.

The prerequisites

Performance Max only works on a usable data foundation. Before you launch, all of this has to be true:

• A purchase conversion with a revenue value, counted once, set as a primary goal. Without values the campaign cannot optimize for revenue. If your purchase numbers are shaky, fix that first with GA4 ecommerce tracking not working.
• Merchant Center with approved products, current prices and availability, linked to Google Ads. The full walkthrough is in Google Merchant Center setup.
• Enough volume. As an experience value from practice: 15 to 30 conversions per campaign per month. That is not a Google requirement and not a threshold checked in the account; it is the order of magnitude at which decisions based on the numbers stop being noise. With less data, one campaign covering everything beats three small ones.
• Time. Two to four weeks to a stable learning period, six to twelve weeks before the numbers support firm conclusions.

The seven real levers

1. Conversion goals of the campaign

At campaign level you decide which conversions count. If newsletter signups sit there next to purchases, the campaign will optimize towards the cheaper action. Reduce the goals to the purchase.

If you want to weight new customers more heavily, Google Ads has a new customer acquisition setting with three modes: bid higher for new customers, bid higher for high-value new customers, or bid for new customers only. The last one restricts heavily and only fits if you already reach existing customers reliably through email and organic channels. All three need a value-based bidding strategy, and the first two need a purchase conversion goal.

2. Asset groups

An asset group is your unit of content: text, images, logos, video, plus the product group from the feed. The same asset group covering hiking boots and swimwear produces ads that suit neither.

Rule of thumb: one asset group per product area for which you have your own images and your own claims. Not one per product, because then no group has enough data.

On asset quality:

• Fill the text fields as far as you can and write genuinely different claims: range, price level, shipping, returns, material, use case. Five variations of the same sentence achieve nothing.
• Upload images in all requested formats, so square, portrait and landscape. Missing formats exclude surfaces.
• Upload your own video. If there is none, Google can assemble one from your other assets, which is rarely as good as a real one.
• Watch the asset ratings in the reports and replace weakly rated elements instead of endlessly adding more.
• Check whether automatically created assets are active. If you want control over wording, switch them off.

3. Product groups from the feed

The product group decides which items an asset group advertises. That is how you separate areas with different margins, exclude clearance stock, or give new products their own stage. It requires maintained feed attributes: product type, brand, category, and above all custom labels for things like margin class, season or stock level.

The feed is the single most important steering layer in Performance Max. Title, image, product type and price decide which queries your products can match at all, and that replaces the keyword. Two feed problems that silently cap a campaign are covered in Merchant Center price mismatch and missing or invalid GTIN.

4. Audience signals and search themes

Signals are hints, not a targeting limit. The campaign starts there and goes beyond it if it finds conversions elsewhere. Useful signals:

• Your own customer lists, split into buyers and prospects
• Website visitors, cart abandoners, purchasers from recent months
• Relevant custom segments based on search behavior and websites

Search themes are a different thing. They are not an audience but an input per asset group describing what your products are searched for. Use them for terms where you suspect demand but which do not appear in your titles and product data. Good signals and clear search themes shorten the learning period noticeably. Supplying neither is the most common avoidable oversight at launch.

5. Brand exclusions and negative keywords

Brand exclusions stop the campaign bidding on your own brand name, which separates cheap brand conversions from real new customer acquisition. The same mechanism works for other brands you do not want to serve.

Negative keywords can be added to an individual Performance Max campaign, and account-level negative keywords apply to it as well. The Google Ads help names one limitation explicitly: Performance Max negative keywords apply to Search and Shopping inventory only, so they do not filter Display, YouTube or Discover placements. Maintain an account-level list with the classics regardless: free, used, repair, tutorial, jobs.

6. The search terms report

Performance Max has its own search terms and landing pages report. It shows the actual queries that triggered your ads together with the landing pages they led to, and the Ad format segment tells you whether a Shopping ad or a text ad was served. That makes two routines possible that used to be guesswork: promoting queries that convert into your search campaign as their own keywords, and adding queries that cost without converting as negatives.

Read it at least every two weeks. If it shows a large share of brand queries, your ROAS is flattering you.

7. Final URL expansion

With expansion active, Google also sends visitors to pages you did not specify, often blog articles or the homepage. That can add reach and cost conversion rate. Switch it off, or exclude specific URLs, if you want only category and product pages advertised.

Performance Max or a search campaign?

In almost every store both types run in parallel. What matters is the arrangement between them: brand queries in the search campaign, brand exclusions in Performance Max, and a regular transfer of good queries from the automated reports into search. How that fits the whole account is described in Google Ads for ecommerce.

A four-week steering rhythm

1. Week 1, check the data foundation. Merchant Center: approved products, price and availability errors. One disapproved bestseller matters more than any bid adjustment.
2. Week 1, read the product report. Which items make the revenue, which get clicks without orders? Remove the latter from the product group, or pull them into their own campaign with their own target.
3. Week 2, assess asset groups. Compare performance per group, replace weakly rated assets, add missing formats.
4. Week 2, read search terms and insights. Use the search terms report for new keywords in your search campaigns and new negatives, and the audience and demand insights for context.
5. Week 3, adjust target values. Change target ROAS in steps of 10 to 20 percent, no more. Then leave it alone for two weeks.
6. Week 3, check landing pages. Spot check: do clicks land on products that are in stock? Is the page fast on mobile?
7. Week 4, reconcile the result. Put Google Ads revenue against contribution margin in your store. The method is in how to calculate ROAS.
8. Week 4, keep notes. Record changes with dates, otherwise you cannot attribute any effect later.

In ecomando the Google Ads and Shopping modules sit on the same daily timeline as your store revenue, which is what makes step 7 a look rather than a spreadsheet exercise.

The most common mistakes

• Too little data, too many campaigns. Three campaigns with eight conversions each learn nothing. Merge them.
• Mixed margins. The campaign optimizes for revenue. Products at 10 percent and at 60 percent margin in one campaign means ROAS looks fine while profit falls.
• Brand queries counted in. Without brand exclusions, every Performance Max campaign looks good.
• Secondary goals as primary conversions. Newsletter signups are cheaper than purchases, so the automation steers towards them.
• An over-ambitious target ROAS from day one. The campaign never spends its budget and never collects enough data. Start without a target, then steer.
• Constant changes. Every larger adjustment resets the learning period.
• Treating the feed as an afterthought. Weak titles and missing attributes limit the campaign harder than any bid.
• Ignoring final URL expansion. Traffic to blog articles looks like reach and rarely produces orders.

When to switch Performance Max off

There are cases where the campaign type does not fit: very small catalogs with few orders a month, products under strict advertising restrictions, stores with no usable image or video material, and stores where the feed cannot be kept reliably current for technical reasons. In those cases tightly managed search campaigns, or standard Shopping with control over product groups, is the more honest solution.

Before you switch anything off, though, check the foundations: measurement, feed, goals, brand exclusions. In practice the reason for poor results sits in those four far more often than in the campaign type itself. If the symptom is spend without sales, the ordered diagnosis in Google Ads: clicks but no conversions is the faster route than rebuilding the campaign. The official description of every setting is in the Google Ads Help Center.