Google pitches Performance Max to online stores the same way every time. Connect your Merchant Center feed, pick a budget, and one campaign puts your products on Search, Shopping, YouTube, Gmail, Discover, Maps and partner sites. The setup looks like a ten-minute job because every hard decision already has a default filled in.
Those defaults are the problem. Three decisions determine whether Performance Max makes a store money, and Google's setup treats all three as optional: which campaign carries your products, how they're organized, and how it bids before it knows who buys from you.
This guide assumes the groundwork is done: a clean product feed, purchase tracking that records real order values, and an account that already sells. We cover the feed in Google Shopping Ads Have No Keywords. Your Product Feed Is the Campaign., and our view on timing is simple: Standard Shopping comes first, and Performance Max joins once the account makes about 30 sales a month. Our ecommerce guide's Performance Max chapter explains why that number. Below it, the campaign has too little to learn from.
Feed-Only Performance Max Lost Its Edge in 2026. Pick Standard Shopping or the Full Version.
For a few years, stores had three ways to put products on Google, and experienced managers often picked the odd one.
- Standard Shopping. Product ads on Google Search and the Shopping tab, nothing else, with the most control.
- Full Performance Max. Your feed plus headlines, images and video, shown across every Google channel.
- Feed-only Performance Max. A Performance Max campaign with your feed and no creative at all.
Feed-only was a workaround. With no images or video, most of the money went to Shopping ads, where stores earn most of their sales. It also had two edges over Standard Shopping: it could start on Maximize conversion value, chasing the most revenue without a fixed return target, and you could add text later to show search ads from the same campaign.
Both advantages have faded. Google's own comparison table now lists Maximize conversion value as an automated strategy for Standard Shopping, alongside Target ROAS, Maximize clicks and manual bidding. And Google has added AI Max for Shopping, which lets a Standard Shopping campaign also serve text ads built from your feed, sent to whichever page fits the search. That gives you Shopping plus search ads without YouTube, Gmail and Display coming along.
There are fine-print limits worth knowing before you plan around it. AI Max for Shopping is a beta, so not every account has it. The text-ad part only works on a Target ROAS strategy, and Google says the text features are limited to English feeds for now.
Feed-only also has a leak that most guides skip. Google says advertisers can launch Performance Max without creative, but "ads may be auto-generated on your behalf when possible," and without uploaded video it will make videos from your images and feed. So a feed-only campaign is not guaranteed to stay in Shopping. Most of the time the leak is small. It is still a leak, and it grows when the campaign has a loose budget and no return target.
So the real choice in 2026 is two options, not three.
Standard Shopping wins when:
- Return on ad spend matters more right now than total sales volume.
- You want to decide what goes to Shopping, Search and video yourself, with a separate campaign and budget for each.
- You want Standard Shopping's own controls, such as campaign priority settings that decide which of several Shopping campaigns bids first on a product.
Full Performance Max wins when:
- You want volume and are willing to accept a lower return while it finds new buyers.
- You already have images and short videos that sell, often from your Meta ads, and want them tested on YouTube and Gmail without building separate campaigns.
- You would rather manage one campaign than four.
If you run feed-only today and it is profitable, leave it alone. Moving costs you a fresh learning period, and that dip can wipe out whatever the cleaner setup would have earned. If you do move, build the new campaign next to the old one and shift the budget in steps, about a quarter at a time, letting each step settle for a week or two before the next. Never switch a working campaign off in one day.
Standard Shopping and Performance Max Can't Share a Product. Ad Rank Picks One Every Time.
Plenty of stores end up running both, and that's fine. The mistake is letting both campaigns sell the same products.
Only one ad from your account can enter a given auction. When a Standard Shopping campaign and a Performance Max campaign both include the same pair of boots, Google has to choose between them. When Performance Max launched, it simply won. That rule is gone. Google's help page now says that when both campaigns target the same products and the same settings, "Ad Rank will determine which campaign serves an ad."
That's fair, and still wasteful. Each search for those boots goes to whichever campaign has the higher Ad Rank, so neither sees all the data, and you pay for two campaigns to learn the same lesson slowly.
The fix is to give every product exactly one home. Put a custom label on each product in Merchant Center saying which campaign it belongs to, then filter each campaign by that label: listing groups in Performance Max, product groups in Standard Shopping.
How you split the catalog is where experienced managers disagree. One camp says never run Standard Shopping beside Performance Max at all, because it only divides your data. The other keeps a Standard Shopping campaign as a catch-all for every product Performance Max doesn't carry: new arrivals, slow sellers, anything still proving itself. Products that start selling move up into Performance Max. We lean toward the second for most small stores, because it gives untested products a cheap, controlled place to earn data. Either way, the rule holds: one product, one campaign.
One Performance Max Campaign Is Enough Until a Product Line Needs Its Own Budget or Target.
Performance Max organizes products into asset groups, each with its own products, headlines, images, videos and search themes. Treat one like an ad group in a Search campaign: one theme, and every piece of it about that theme.
For most stores, the theme is the product category. A store selling outdoor gear might run one asset group for tents, one for sleeping bags and one for cookware, each with photos and copy about that category alone. A store where a few products drive most of the revenue can give those best sellers a group of their own. Two rules keep the structure honest:
- Never build asset groups around audiences. Audience signals are hints, not targeting, and Google reaches past them anyway. Two asset groups with the same products and different audiences just compete with each other.
- Never put a product in more than one group. Listing groups decide which products each group can show. If every group includes "all products," your carefully written tent headlines can end up next to a frying pan.
Each asset group can take up to 50 search themes, which steer Google toward the searches you care about early on, before it has sales data. Write them the way you'd name an ad group, such as "two-person backpacking tent," not as a pile of keywords. On creative, Google allows up to 15 headlines, 5 long headlines, 5 descriptions, 20 images and 15 videos per group. You don't need to fill every slot, but the photos should vary: product on white, product in use, close-up details.
When should you split into a second campaign? Only when a group of products needs something a campaign-level setting controls: a different daily budget, a different ROAS target, or a different country. Each campaign bids on its own conversion history, so every split divides your data. A store that sells a $40 accessory and a $900 flagship may need two targets, which is the margin problem we walk through in You Organized Your Google Ads Account. You Starved It. A store whose categories all earn similar margins usually doesn't.
A Campaign's Average ROAS Hides the Products That Lose You Money.
A ROAS target applies to the campaign's average, not to each product. That means a few strong products can carry a long tail of products that lose money on every sale, and the campaign still looks like it's hitting its number.
Here's an illustration. A store sells at a 40 percent gross margin, so it breaks even at a 250 percent ROAS. Its one campaign reports 400 percent. Underneath, 30 products earn 900 percent on a small slice of the budget, while 120 products take more than half the spend at 180 percent, losing money on every order. The average looks healthy. The account is quietly subsidizing its worst sellers.
The fix is to sort products by what they've actually done. Open the Products tab across all campaigns, set a long date range, 90 days or more, and use two thresholds:
- A return threshold: comfortably above your break-even ROAS. In the example, 300 percent.
- A spend threshold: below it, a product hasn't had a fair test. A useful rule is about one average cost per sale. If you usually pay $30 for an order, a product that has spent less than $30 hasn't had its chance yet.
That gives you four groups:
- Winners. Above both thresholds. They deserve more budget and a slightly lower target, so Google can bid harder for them.
- Money-losers. Enough spend, return below the line. They get a stricter target and less budget, or they come out of the ads.
- Not enough data. Return looks good or bad, but on too little spend to trust. They need more spend to tell you anything.
- Never shown. No impressions at all. Often a feed problem, such as a vague title, missing attributes or an uncompetitive price, rather than a bad product.
Label each group in your feed with a custom label, and the labels become the filters your campaigns use. Refresh them monthly or quarterly, not weekly; products need time in a group to show what they'll do. This is the same split you'd use to separate high-margin and low-margin products, just driven by results instead of a spreadsheet of margins.
A Day-One ROAS Target Leaves Performance Max Guessing. Add It After the Sales Come In.
A ROAS target tells Google to win as much as it can while returning a set amount per dollar. That's the right instruction for a campaign that knows who buys from you, and the wrong one for a brand-new campaign with nothing to base its bids on. Set a high target on day one and the campaign often barely spends.
Google's guidance is to start differently. When you track order values, it recommends Maximize conversion value, and with a limited budget it says to start there rather than with a target. The campaign spends its budget chasing the most revenue it can find, and it builds the history a target will need later.
Add the target once the campaign produces steady sales. Google's floor for a ROAS target on Shopping campaigns is 15 conversions in 30 days. We wait for about 30, for the same reason we wait to launch: a target set on thin data swings around. When you add it, set it near what the campaign is already achieving over the last two weeks, using the return Google expects once delayed conversions arrive, not a number from your business plan. A target far above current results throttles spending overnight.
After that, move it in small steps, about 10 to 15 percent at a time, and only after the last change has settled. Want more volume? Lower the target a step. Need a better return? Raise it a step. Google also says to let a new Performance Max campaign run at least six weeks before judging it, and that holds for target changes too. Judging a target change after three days is judging noise.
Google's New Customer Bonus Inflates the ROAS Your Dashboard Shows.
Performance Max offers to bid harder for new customers. In the most common version, you add a bonus value to every new customer's order, and Google bids as if those orders were worth more. Our article on Demand Gen's new prospects mode covers the three settings and how Google tells new customers from existing ones.
What Google's setup screen doesn't spell out is where that bonus goes. Google's help page says the conversion value column "will include both the original purchase conversion value and the additional conversion value you specify." The bonus is added to the revenue your reports show.
An illustration. A store spends $1,000 and makes 40 sales at $80 each: $3,200 in real revenue, a 320 percent ROAS. Twenty-four of those buyers are new, and the store set a $40 new-customer bonus. Google adds $960 to the reported value, so the dashboard shows $4,160 and a 416 percent ROAS. If the campaign's target is 400 percent, it is hitting its target on paper while returning 320 percent in your bank account. Every new-customer bonus works like this: it quietly lowers the real return you are asking for.
So leave it off at launch. Once the campaign has run a month, check how many of its sales come from new customers, using Google's new-versus-returning customer reporting alongside your own store data. Then add a bonus deliberately, sized to what a new customer is worth to you over time. When you judge results, use the "Original conv. value" column, which shows revenue without the bonus. For the setting to work well, Google's Performance Max setup asks for a list of existing customers with at least 1,000 active members, so very small stores may not be ready for it anyway.
Google Turns On Three Performance Max Settings by Default. Each One Sends Your Budget Somewhere You Didn't Pick.
Your own brand name. Left alone, Performance Max bids on people searching for your store by name. They were already coming, and the campaign takes credit for sales it didn't create. Add a brand exclusion list in the campaign settings. Google then shows you a second choice: whether to "allow Shopping ads on searches that mention excluded brands." If no other campaign shows your products to people searching your name, tick it. A branded Search campaign covers the text ad, but nothing else would put your product photos in front of those shoppers.
Final URL expansion. This is on by default. It lets Google send a click to any page on your site it thinks will convert, not just the product or collection pages you picked. For a store that can be useful, because a search for a specific item can land on that item's page. But it can also land on your blog, your About page or your returns policy. Keep it on only with exclusions for pages that don't sell.
Display and Search partners. For years, Performance Max gave you no say over which channels it used. Google is now rolling out channel priorities, starting in October 2026: you can mark the Display Network and Search partners as low priority and cap each one's share of the daily budget, down to zero. For a lead business we cap Display at zero, as we explain in our Performance Max lead generation guide. A store is different. Display carries dynamic remarketing, the ads that show shoppers the exact products they looked at, and those often sell. Our call for most stores is a small cap, around 10 percent, then watch what it returns. Two cautions from Google's page: the cap is best-effort, so spend can occasionally run over, and setting a priority does nothing if the campaign isn't limited by budget, although the cap still applies. YouTube, Gmail, Discover, Maps and Search stay bundled together as "Google channels" and can't be separated.
One more default to know: if you don't upload a video, Google builds one from your images and feed. A plain 20-second clip of your product in someone's hands, filmed on a phone, usually beats what it assembles.
Campaign ROAS Hides Where the Sale Came From. The Performance Max Channel Report Shows It.
A Performance Max campaign reports one return for everything it does, blending a Shopping ad that sold a tent to someone searching for tents with a YouTube ad shown to someone who already had it in their cart. The channel performance report inside the campaign pulls them apart.
Two labels trip people up. Under Google Search, the report splits spend into "ads using product data," which are your Shopping ads, and ads not using product data, which are text ads. Most of a store's spend should land in the first. We walk through the rest of the report in Performance Max Isn't Your Smartest Campaign. It's Google's.
How much spend outside Search and Shopping is too much depends on what you set up. In a full Performance Max campaign where you deliberately uploaded video, you should expect some YouTube spend and judge it on its return. In a campaign built mainly to sell through Shopping, experienced managers start asking questions when more than about 5 to 10 percent goes elsewhere. One test we like when you're unsure whether your creative earns its place: run a feed-only asset group and a full-creative group over the same products. If the non-Shopping channels take more than half the spend in the first week, pause the creative. If they're still short of your target after a month, keep the feed-only group.
Then go one level down. The Products tab shows which items took the money, which is where you'll find the losers and the never-shown products from the earlier section. In order, the questions to ask are:
- Is the campaign spending its budget, or held back by its target?
- Which channels took the money, and what did each return?
- Which products took the money, and did they earn it?
Fix the answer to the first question before touching the second, and the second before the third. Most problems that look like "Performance Max doesn't work" turn out to be a target set too high or a few products eating the budget.
Performance Max's Defaults Fit Google's Goals, Not Your Store's Margins
Performance Max will happily run a store's ads on its own defaults, which aren't set for your margins, your repeat customers or your catalog. Choose which campaign carries each product, give every product one home, launch on Maximize conversion value, keep the new customer bonus out of your numbers until you have a baseline, and read the channel report before you trust the headline ROAS.
We do this for online stores every week, from the feed to the bidding. If you'd rather hand it off, we run Google Ads for ecommerce businesses. And if you'd rather build it yourself, everything above is how we'd do it.
