Performance Max for Ecommerce: How to Structure Campaigns

Performance Max shows your products across Search and Shopping results, YouTube, Display, Discover, Gmail and Maps. In each auction, it picks the product, the placement and the bid. With hundreds or thousands of products, structure decides which sales each campaign learns from. Count each purchase once, with its real order value, and give each product one place. Split campaigns only where budget, target or market differ, and only if each one keeps enough conversions. GetProfit’s portal treats 30 conversions a month per campaign as the minimum and 50+ as comfortable.

How Performance Max works for an online store

Performance Max is a Google Ads campaign type that buys ads across Google’s inventory from a single campaign. Google lists that inventory as YouTube, Display, Search, Discover, Gmail and Maps in its article About Performance Max campaigns. For an online store, Shopping ads remain an important format inside it, shown in Search and Shopping results, according to Google’s Retailer best practices for AI-powered Performance Max campaigns. Which of these places gets your budget is a separate question, answered in where Performance Max shows your products.

The campaign starts from your product data. That is your product feed: the file of products, prices and availability that your store sends to Google. It lives in Merchant Center, Google’s service for product data. Before Performance Max can advertise those products, you need to link your Google Ads account to Merchant Center (Performance Max for online sales with a product feed (retail)).

How Google Shopping ads work follows one product from the feed into an ad and shows what stops it from showing at all. Every product ad is built from the feed, so feed quality is a lever of its own: see product feed optimisation.

Inside a campaign, asset groups bundle creative, audience signals and a set of products, and one campaign can hold several of them. On top of the feed, the campaign runs on three more inputs from you:

Bids come from Smart Bidding: Google’s automated bidding, which sets bids in real time, auction by auction, towards your conversion or conversion value goal (About Performance Max campaigns). The split of work between you and Google looks like this:

One consequence is easy to miss. Google’s FAQ (Answering Your Top Questions About Performance Max) explains that Performance Max optimises for marginal return across all channels: the extra return from each additional unit of spend. So the average result of one channel inside the campaign can mislead you: judge the campaign by its total first. What has to be ready before a first launch, from tracking to budget, is covered in launching your first Performance Max campaign.

Most stores put almost all their budget into Performance Max

Performance Max runs in 141 of 146 online stores in our data and takes a median 95.7% of their ad budget (GetProfit data, 146 stores, June 2025 – June 2026). Its median ROAS, revenue divided by ad spend, is 476% across the 138 of these stores included in the ROAS calculation. We calculate each share within its own store, so we never add up the five currencies in the sample.

Read the table with two caveats. First, each median comes only from the stores that run that campaign type, and each ROAS median only from the stores counted in brackets. So every row describes a different group of stores, and the table doesn’t rank the channels. Second, our ROAS counts revenue, not profit, because we have no product costs.

These are observations, not an experiment. They show what stores do, not what works better.

In practice, for most stores, the structure of Performance Max is the structure of the whole ad account. And that budget spreads unevenly across the catalogue. In our study of 1.4 million products, 55.0% of products received no ad budget at all over 13 months. In the median store, the top 10% of products took 68.1% of spend.

Structure is how you decide which products compete for that budget at all. Stores with tens of thousands of products face this at the largest scale, covered in Performance Max for large catalogues.

Performance Max, Standard Shopping or both?

Whether your store needs Performance Max, Standard Shopping or both depends on your conversion volume, your budget and how much control you need. We weigh that decision in Performance Max vs Standard Shopping.

Google’s position is clear. It reports that advertisers who moved from Standard Shopping to Performance Max got 25% more conversion value on average at a similar ROAS (Retailer best practices for AI-powered Performance Max campaigns). That figure is Google’s own data about its own product: global, October 2022 – March 2023. Google itself suggests that you check it in your own account with an experiment.

In our data, the two types often run side by side, but not on equal terms: Standard Shopping runs in 74 of the 146 stores and takes a median 6.5% of their budget. When several campaigns in one account could serve the same product or the same search, three rules decide which one does:

1. Performance Max no longer gets automatic priority over Standard Shopping when both target the same products in one account. The campaign with the highest Ad Rank serves, and your bid targets are the main lever. Google’s Ads Product Liaison and its retail product lead explain this in the podcast How to build a retail growth engine in 2026 (April 2026).
2. Campaign priority (Low, Medium, High) still decides between Standard Shopping campaigns that share a product: the higher priority wins (Create a Shopping campaign, Google Ads API documentation).
3. A Search campaign whose exact match keyword is identical to the query is still preferred over Performance Max (About Performance Max campaigns).

Google has a ready-made experiment for this. According to its retailer best practices, a one-click A/B test builds a new Performance Max campaign from an existing Standard Shopping campaign. Before you compare the two sides, make sure they bid towards the same goal and a comparable target ROAS. Testing Performance Max against Standard Shopping shows how to read the result without fooling yourself.

If you decide to move for good, plan the transition so sales hold steady while the new campaign learns: see moving Standard Shopping to Performance Max. If you keep Standard Shopping, its product groups need their own logic as the catalogue grows, covered in how to subdivide Shopping product groups. Google revises these rules, as the Ad Rank change shows. Performance Max updates worth knowing collects the changes from 2025–2026 that matter for a store.

How many Performance Max campaigns can your store feed?

GetProfit’s portal treats 30 conversions a month per campaign as the minimum and 50+ as comfortable. So the number of product campaigns your store can feed is roughly your monthly conversions divided by 30. The more conversions a campaign learns on, the less its results swing and the closer it stays to its target. That is the core of campaign structure: it depends on how many conversions each campaign gets, not on how many categories you sell.

Google’s own guidance is also to consolidate. Its FAQ (Answering Your Top Questions About Performance Max) recommends separate Performance Max campaigns only when they need different goals, budgets, target ROAS or target CPA for business reasons. For different themes inside one business, it points to asset groups within one campaign. The reason Google gives: the system learns faster and predicts better with more data and performance history.

Example store, not client data.

A tableware store with 3,000 products gets 300 orders a month from Performance Max, and each order is one conversion. In one campaign, all 300 orders feed one bidding model, which is ten times the portal’s minimum.

Now split the store into ten campaigns by category. Plates and bowls get 150 orders, glassware 60, cutlery 30, and the other seven categories share 60 between them. Three campaigns reach 30 a month. The other seven get about 9 orders a month each, far below the portal’s minimum of 30.

Real accounts can hold even more campaigns. In our study of 1.4 million products, the median store had 28 campaigns. The study doesn’t say how many of them were product campaigns, but the arithmetic is simple. Ten product campaigns need 300 conversions a month between them before each one reaches the portal’s minimum, and 28 would need 840. To run this count on your own account, campaign by campaign, see how many campaigns your store supports.

One campaign versus one per category explains what a split by category costs and when you need one. A store with a few hundred products faces the same arithmetic with even less data per product. Performance Max for small catalogues covers when such a store can justify a second or third campaign.

The four rungs the portal uses

The portal fits the structure to each account. First it checks whether your conversion numbers can be trusted. Then it counts how many conversions your product campaigns bring and puts the account on one of four rungs:

The portal’s working rules for changing structure are just as specific. It suggests a split when a campaign gets more than 100 conversions a month, or when its parts differ in ROAS by more than 50%. Even then, it suggests the split only if both new campaigns would get at least 30 conversions a month. In the other direction, it suggests merging asset groups that get fewer than 5 conversions a month.

When you split, leave the products that already sell in the existing campaign, so it keeps its learning history. Move only the uncertain rest into the new campaign. Then only the new campaign has to learn from scratch. Splitting or merging Performance Max campaigns walks through both moves.

Separate campaign, asset group or listing group?

A Performance Max campaign has three levels, and each controls something different. Budget and target belong to the campaign. Creative and audience signals belong to the asset group: Google advises building each asset group around a single theme, such as a product or service category (Multiply conversions with Performance Max). The choice of products belongs to the listing groups inside each asset group, and you can sort products for them with custom labels in your feed.

Google’s list of reasons for separate campaigns is short: different countries or languages, and different budgets or target ROAS for new products, top-selling products, seasonal moments or store locations (Retailer best practices for AI-powered Performance Max campaigns). Google’s FAQ on Performance Max adds a different target ROAS for certain product categories. When products only need different creative, see asset groups versus separate campaigns.

Each way of splitting the catalogue costs you data. We weigh the trade-offs between category, performance and margin splits in how to split Performance Max campaigns.

Bestsellers deserve a separate note, because Google lists top-selling products among the reasons for a separate campaign. Our study of 1.4 million products adds a timing problem. Of the products that became bestsellers, 71% stayed in the top for no more than 3 months out of 13. A stable bestseller, one with two or more months of sales, held that status for a median of 3 months.

If you build a campaign around this quarter’s bestsellers, refresh its product list at least that often. A separate campaign for bestsellers weighs whether such a campaign pays off at all.

How listing groups decide which products a campaign shows

Listing groups sit inside each asset group and decide which products it includes or excludes (Listing groups for retail, Google Ads API documentation). You can subdivide them by category, brand, item ID, condition, product type, channel (online or local) and custom attributes (your custom labels). Each asset group can hold up to 1,000 listing groups (System limits). Google notes that listing groups work best on groups of products rather than single item IDs, and custom labels are one way to build such groups.

Google’s product category is only one of these options. Product type, brand, item ID and custom labels come from your own product data, so you control how they group your catalogue. Custom labels are up to five free-form fields in your product data, numbered 0 to 4 (Filter products, Google Ads API documentation). Google’s retailer best practices give “bestseller”, “trending” and “holiday product” as example values, which you can use to move such products into separate campaigns or asset groups when needed.

The rule that saves the most trouble: one product, one place. Google’s API guide for retail recommends that each asset group target different products, for example items A–L in one asset group and M–Z in another. When the same product sits in two asset groups or two campaigns, its sales can be split between them, so neither place learns from every sale of that product. The portal’s structure check looks for exactly this: the same product serving from two campaigns at once.

The same product in two campaigns shows how to find and remove such overlaps. To decide which attribute to build listing groups on, from product type to item ID, see choosing an attribute for listing groups.

Feed only or full assets: what goes into an asset group

With an online sales goal, Google uses your product feed as the starting point for ads. It strongly recommends adding a variety of text, image and video assets so the campaign can show ads in more places (Retailer best practices for AI-powered Performance Max campaigns). Without a video from you, Google may generate one from your text, images and feed. According to the Google Ads API guide, Google can generate assets automatically for a campaign linked to Merchant Center, “but reach and performance will be limited”.

Running Performance Max on the feed alone, without your own text, images or video, is a setup in its own right. Google offers a way to test it: asset experiments in Performance Max can measure the impact of adding new assets to a campaign. Feed-only Performance Max covers when this setup makes sense and when added assets start to pay off. The opposite case, a campaign with no products attached, behaves differently, and you shouldn’t judge it like a Shopping campaign: see Performance Max without a product feed.

Two more settings at this level shape where traffic goes:

• Audience signals. They are inputs that help the campaign ramp up, not limits: Performance Max doesn’t restrict your ads to the audiences you add (Multiply conversions with Performance Max). Google’s retailer best practices name one type of audience signal as the most valuable: data about past buyers who gave consent, shared through Customer Match. Audience signals for Performance Max tests which signals change anything.
• Final URL expansion. Google’s guide Multiply conversions with Performance Max says it is on by default and recommends keeping it on. When it is on, Google may replace your final URL with another page on your website that better matches the search. To keep pages such as a blog or FAQ out, Google offers URL exclusions. Final URL expansion for online stores weighs whether a store should keep it on.

What should Performance Max learn on?

A campaign learns only on the conversions you give it. For a store, that means purchases, each with its real order value. In its retailer best practices, Google strongly recommends passing transaction-specific values, the real value of each order, when customers buy items at different prices. You can set conversion goals for the whole account or override them for one campaign, as the Google Ads API retail guide shows.

If a sale counts twice, or every order carries the same fixed value instead of its real one, the campaign learns on the wrong signal. On the portal’s structure ladder, that is rung 0, “No trust”: fix the conversion data before you restructure anything.

When you track order values, Google recommends Maximize conversion value, with an optional target ROAS if you have a specific return goal (Multiply conversions with Performance Max). If you’re unsure what target to set, the same guide suggests starting from the targets of your other performance campaigns, such as Search. Then adjust it once the campaign has ramped up.

Google’s retailer best practices suggest basing the target on profit, taking cost of goods sold and customer lifetime value into account. The portal adds a rule for changing it: no more than 15% at a time, and no more often than once every 1–2 weeks. Bidding strategies for online stores covers which strategy fits your current number of conversions a month.

Google also offers a new customer acquisition goal, described in its retailer best practices. In New Customer Value mode, the campaign bids higher for new customers while still selling to existing ones. New Customer Only mode optimises only for new customers, and Google says it makes sense in some cases, such as when you have a budget set aside for acquisition.

How long to wait before you judge a Performance Max campaign

Google’s API guide advises giving a new Performance Max campaign at least 6 weeks before you compare it with existing ones (Compare performance with an existing campaign). That gives the system time to ramp up and gather enough data. A new campaign may go through a learning period while it collects data.

Also allow for the delay between click and purchase. If buyers typically convert within 7 days of the ad, leave the most recent week out of your evaluation, as Google’s guide Multiply conversions with Performance Max advises.

In the portal’s methodology, a rebuild restarts learning only for the campaigns you change. That is why the portal’s approach to splitting keeps the core in place and moves only the uncertain part. It is also why you adjust structure for the season or every few months, when demand shifts, rather than every week.

If you also sell from a showroom

A store with a physical showroom has extra options in the same campaign type. Adding a local product feed turns on local inventory ads, which promote products available in your showroom. If you optimise for store goals, Google requires every type of creative asset. Final URL expansion isn’t available to campaigns that optimise only for store goals (Retailer best practices for AI-powered Performance Max campaigns). Performance Max with a showroom covers whether to use store goals at all and how to keep store visits and calls from distorting online results.

Check your Performance Max structure in eight steps

1. Check what your campaigns learn on. Purchases only, each order counted once, with the real order value. What to check after: the portal flags a product campaign whose average conversion value falls below 0.3× or above 8× the account’s median order value, once it has at least 10 conversions.
2. Count conversions for each product campaign. Take the last 90 days and turn them into a monthly rate for each campaign. The portal’s campaign structure section does this count over a 90-day window and shows how many campaigns your volume can feed. In the portal’s account score, from 0 to 100 with a letter from A to F, structure is worth 30 of the 100 points.
3. Find your rung. If product campaigns bring fewer than 30 conversions a month in total, consolidate. From 30 a month, you can split under the conditions in step 6.
4. Merge what can’t learn. A campaign far below 30 a month, with the same budget logic and target as a neighbour, is a candidate to join it. Asset groups with fewer than 5 conversions a month are merge candidates too.
5. Give every product one place. Check the listing groups: every product you want to sell sits in exactly one asset group, in one campaign. What to check after: the product counts across asset groups add up to the number of products you mean to advertise.
6. Split only for a business reason. A different budget, target, country, language or season. Split only if each part keeps 30 conversions a month, and leave the products that already sell in the existing campaign.
7. Change targets in small steps. Move a target ROAS by no more than 15% at a time, and no more often than once every 1–2 weeks.
8. Wait before you judge. Give a new campaign at least 6 weeks before you compare it, and leave out the latest days that are still missing conversions. Review the structure for the season, not every week.

If you run ads for a client, agree each structural change with the account owner before you make it.

See how many campaigns your account can carry. The portal counts how many conversions your ads bring, says how many campaigns that is enough for, and goes through each of the ones already running. The portal changes nothing without your consent.

Analyse your campaigns →

Sources

• About Performance Max campaigns (Google Ads, September 2025) — the channels Performance Max serves on, Smart Bidding, exact match keywords preferred over Performance Max. Checked 2 October 2026.
• Answering Your Top Questions About Performance Max (Google Ads Help, August 2025) — consolidate campaigns; separate ones only for different goals, budgets or targets; different ROAS targets for product categories; optimisation for marginal return. Checked 2 October 2026.
• Retailer best practices for AI-powered Performance Max campaigns (Google Ads Help) — Shopping ads inside Performance Max, Google’s 25% figure for moving from Standard Shopping, one-click experiments, reasons for separate campaigns, custom labels, transaction-specific values, Customer Match as a signal, new customer acquisition, local inventory and store goals. Checked 2 October 2026.
• Multiply conversions with Performance Max (Google Ads Help) — one theme per asset group, audience signals as inputs rather than limits, value-based bidding and starting targets, Final URL expansion on by default, URL exclusions, conversion delays. Checked 2 October 2026.
• How to build a retail growth engine in 2026 (Google Ads Decoded podcast, April 2026) — Ad Rank decides between Performance Max and Standard Shopping. Checked 2 October 2026.
• Scale results with new asset experiments in Performance Max (Google Ads announcement) — experiments that measure the impact of adding assets. Checked 2 October 2026.
• Performance Max for online sales with a product feed (retail) (Google Ads API, last updated 30 September 2026) — the Merchant Center link, assets generated automatically with limited reach, different products per asset group, campaign-level goals. Checked 2 October 2026.
• Listing groups for retail (Google Ads API, last updated 30 September 2026) — what listing groups do and the attributes they can use. Checked 2 October 2026.
• System limits (Google Ads API, last updated 30 September 2026) — up to 1,000 listing group filters per asset group. Checked 2 October 2026.
• Filter products (Google Ads API, last updated 30 September 2026) — custom labels 0 to 4. Checked 2 October 2026.
• Create a Shopping campaign (Google Ads API, last updated 30 September 2026) — campaign priority between Standard Shopping campaigns. Checked 2 October 2026.
• Compare performance with an existing campaign (Google Ads API, last updated 30 September 2026) — at least 6 weeks before comparing a new Performance Max campaign. Checked 2 October 2026.
• GetProfit data: 146 online stores, June 2025 – June 2026 — share of ad budget and median ROAS by campaign type.
• GetProfit study of 1,404,808 products in 130+ stores over 13 months — products with no budget, budget concentration, campaigns per store, how long bestsellers last.
• GetProfit portal methodology — 30 and 50+ conversions a month per campaign, the four rungs, split and merge rules, target ROAS steps, conversion value checks.