Performance Max can be profitable for e-commerce stores, but only when you feed it accurate conversion values, a clean product feed and a ROAS target based on your real margins. Left on default settings, it tends to spend heavily on your own brand searches and remarketing, then report results that look better than the new revenue it actually created. The fix is structure, measurement and a few guardrails.

This guide covers how Performance Max works for online stores, how to set it up, and how to keep it honest.

What Performance Max actually does

Performance Max (PMax) is a single Google Ads campaign that can show your ads across Search, Shopping, YouTube, Display, Discover, Gmail and Maps. For e-commerce, the core of it is your Merchant Center feed: most PMax spend for retailers typically goes to Shopping placements, with the rest spread across other channels.

You supply:

  • A Merchant Center feed (your products)
  • Assets: headlines, descriptions, images, logos and videos
  • Audience signals: hints about who your customers are
  • Search themes: hints about what people search for
  • A conversion goal and bid strategy, usually Maximize Conversion Value with or without a target ROAS

Google's automation decides where, when and to whom your ads show. You control the inputs and the guardrails, not the individual auctions.

Smart Shopping, PMax's predecessor, was retired in 2022. If you still see advice about it, it's out of date.

When PMax is a good fit (and when it isn't)

PMax tends to work well when:

  • You have reliable purchase tracking with real order values
  • The account has steady conversion volume. A common rule of thumb is at least 30 conversions a month per campaign, more is better
  • Your feed is in good shape and most products are approved
  • You have a reasonable budget so the campaign isn't starved

PMax tends to struggle when:

  • Conversions are rare or tracking is broken
  • Budget is very small relative to your product prices
  • You need to see exactly which query triggered each sale
  • Most of your sales come from people who already know your brand

If you're in the second group, Standard Shopping is often a better starting point. You can move to PMax once you have the data.

Before you launch: three foundations

1. Conversion tracking with accurate values

PMax optimizes toward whatever you tell it is a conversion. If you count add-to-carts, page views or duplicated purchases as conversions, it will chase them.

Check that:

  • Purchase is the only primary conversion action for sales campaigns
  • Each purchase passes the actual order value and currency
  • Purchases aren't counted twice (a common issue when both a Google Ads tag and a GA4 import are set as primary)
  • Enhanced conversions are set up to improve measurement accuracy

Our guide to Google Ads conversion tracking for e-commerce walks through the setup.

2. A clean, rich product feed

PMax reads your feed to decide which searches your products match. Strong titles, accurate categories, GTINs and good images matter just as much here as in Standard Shopping. Fill in custom labels for margin tier, price band, best sellers and seasonality, because you'll use them to structure asset groups. See product feed optimization for specifics.

3. Know your break-even ROAS

Before you set a target, calculate the ROAS where you stop making money:

Break-even ROAS = 1 / profit margin

If your margin after cost of goods, shipping and fees is 35%, break-even ROAS is 1 / 0.35 = 2.86. Any target below that means you lose money on first orders. Use the break-even ROAS calculator to run your numbers.

How to structure PMax for an online store

Start simple

For most stores, one or two PMax campaigns are enough at first. Splitting too early spreads conversion data thin and slows learning.

A reasonable starting structure:

  • Campaign 1: core products. Your best sellers and healthy-margin products, with a target ROAS based on that margin.
  • Campaign 2: low-margin or long-tail products (optional). A higher target ROAS or a smaller budget, so they don't steal budget from your winners.

Split by margin, not just category

Two products in the same category can have very different margins. Use a custom label for margin tier and assign listing groups accordingly. That lets you give each campaign a ROAS target that matches its economics.

Asset groups

Inside a campaign, asset groups let you pair a set of products with matching creative and signals. Group products that share an audience and message, such as "men's running shoes" vs. "women's trail shoes." Each asset group should have:

  • A listing group filter for the right products
  • Relevant headlines, descriptions, images and ideally a video (if you don't add video, Google may generate one, which is rarely your best work)
  • Audience signals built from your customer lists and site visitors
  • Search themes that describe what buyers search for

Feed-only PMax

Some stores run PMax with only the feed and minimal or no creative assets, to keep spend concentrated on Shopping placements. This can work, but it's not a guarantee: Google can still serve your products in other formats. Test it against a full-asset version and compare profit, not just ROAS.

Settings that protect your profit

Brand exclusions

By default, PMax can bid on searches for your own brand name. Those people were likely to buy anyway, and the sales make PMax's ROAS look inflated.

Use brand exclusions at the campaign level to stop PMax from serving on your brand searches, and run a separate, cheap brand Search campaign instead. This gives you a much clearer picture of what PMax actually adds. Note that brand exclusions apply to Search and Shopping inventory.

Campaign-level negative keywords

PMax now supports campaign-level negative keywords, up to 10,000 per campaign. They apply to Search and Shopping inventory within the campaign. Use them to block:

  • Irrelevant queries you find in the search terms report
  • Terms like "free," "used," "repair," "manual" or "jobs" if they don't fit
  • Competitor names you don't want to target

Account-level negative keyword lists can apply to PMax too. See negative keywords for e-commerce for starter lists.

Customer acquisition goal

PMax lets you choose to bid higher for new customers or show ads only to new customers. If growth matters more than repeat sales, this setting is worth testing. It requires you to define existing customers, usually through a customer list and conversion data. Be honest about your numbers: a new-customer value you set too high will inflate reported conversion value.

Final URL expansion

For e-commerce, final URL expansion can send traffic to pages you didn't intend, such as blog posts or the "about" page. Many stores turn it off or exclude URLs that aren't product or category pages.

Location and other exclusions

Target only where you ship. Review device, age and placement exclusions that are available in your account. Google has been adding more of these controls to PMax, so check the campaign settings periodically.

Search themes and audience signals

Neither is targeting in the strict sense. Both are signals that help the campaign learn faster.

Search themes: you can add up to 50 per asset group. Use them for categories and product terms where you want to make sure you appear, especially new products without history. Don't paste your entire keyword list.

Audience signals: start with your strongest data:

  1. Customer match lists (past purchasers)
  2. Website visitors and cart abandoners
  3. Custom segments built from search terms your buyers use
  4. Interests and demographics, if relevant

PMax will go beyond these signals, which is by design.

Bidding: picking a target ROAS

Most e-commerce PMax campaigns use Maximize Conversion Value, with or without a target ROAS.

A practical approach:

  1. Launch without a target, or with a target slightly below your recent actual ROAS, so the campaign isn't choked.
  2. After two to three weeks of stable data, set a target close to the ROAS it's delivering.
  3. Move the target gradually, often in steps of around 10 to 20 percent, waiting a week or two between changes.

A worked example

Your campaign spent $3,000 last month and reported $10,500 in conversion value.

  • Reported ROAS = $10,500 / $3,000 = 3.5
  • Your margin is 35%, so break-even ROAS = 2.86
  • Gross profit = $10,500 x 0.35 = $3,675
  • Profit after ad spend = $3,675 - $3,000 = $675

Profitable on paper. Now check how much of that $10,500 came from branded searches or returning customers who would have bought anyway. If brand exclusions weren't on and a large chunk was brand traffic, the true incremental profit could be much lower, or negative. That's why the brand exclusion and a separate brand campaign matter so much.

Run your scenarios through the ROAS calculator and read what is a good ROAS for how to set targets by margin.

The reports you should check

PMax used to be a black box. It's much less so now. Review these regularly:

  • Channel performance report: shows how spend and conversions split across Search, Shopping, YouTube, Display and other channels. If a big share of spend goes somewhere with weak results, that's a signal to adjust assets or structure.
  • Search terms report: see the actual queries PMax matched. Use it to find negatives and to confirm brand traffic is being excluded.
  • Product (listing group) performance: find products with spend but no sales, and products that never get impressions.
  • Asset performance: see which headlines, images and videos are pulling their weight and replace weak ones.
  • Insights tab: search category trends, audience insights and diagnostics.

A monthly PMax review checklist

  • Is tracking still accurate? Any sudden change in conversion count?
  • What share of conversions came from brand or returning customers?
  • Which channels got spend, and did they convert?
  • Which products spent the most without sales?
  • Any new irrelevant search terms to add as negatives?
  • Is the campaign limited by budget or by its ROAS target?

Common PMax mistakes

  • Optimizing to the wrong conversion. Add-to-cart as a primary goal produces campaigns that are great at generating carts.
  • No brand exclusion. Reported ROAS looks great while new-customer growth stalls.
  • Too many campaigns and asset groups. Thin data means slow learning.
  • Changing the target every few days. Let each change settle.
  • Judging by ROAS alone. Compare total store revenue and profit over time, not just what PMax claims credit for.
  • Overlapping PMax and Standard Shopping without a plan. PMax generally takes priority for the same products, so split products deliberately.

PMax doesn't have to be your whole account. A common, sensible setup for stores:

  • Brand Search campaign to protect your name cheaply
  • PMax for your main catalog, with brand excluded
  • Standard Shopping for tests, new product lines or products PMax ignores
  • Non-brand Search for high-intent terms where you want tight control over copy and landing pages (see our Google Search Ads service)

The right mix depends on your catalog, margins and budget. The Google Ads budget calculator can help you plan spend across campaigns.

If you want PMax set up with proper tracking, margin-based targets and the guardrails above, Mercantads manages Performance Max campaigns for e-commerce stores as a Google Partner agency. Contact us for an account review.

Frequently asked questions

Is Performance Max good for e-commerce?

It can be, especially for stores with reliable purchase tracking, a healthy feed and steady conversion volume. It's less suited to new accounts with little data or very small budgets, where Standard Shopping is often a better start.

Can you add negative keywords to Performance Max?

Yes. PMax supports campaign-level negative keywords, up to 10,000 per campaign, and account-level negative lists. They apply to Search and Shopping inventory, not to YouTube or Display placements.

How do I stop Performance Max from bidding on my brand name?

Add a brand exclusion in the campaign settings and run a separate brand Search campaign. This keeps PMax focused on non-brand demand and makes its reported results more honest.

How long does Performance Max take to learn?

Plan on at least two to six weeks before judging results, depending on conversion volume. Major changes to targets, budgets or assets can restart part of that learning, so make them gradually.

Should I run Performance Max and Standard Shopping together?

Many stores do, but split products deliberately. When both campaigns contain the same product, PMax generally takes priority, which can starve the Standard Shopping campaign and muddy reporting.