If you already run Performance Max against a live feed, you know the problem. The campaign spends where the algorithm sees cheap conversions, not where you make money. Excluding products is how you take some of that control back.
The ranking guides walk you through the clicks. Almost none of them tell you which SKUs deserve the cut, or what the exclusion does to your profit. This article does both.
The three ways to exclude products from Performance Max
There is no single "exclude" button. You have three mechanisms, and they do different jobs.
Listing-group exclusions inside the campaign
Open your PMax campaign, go to the products or listing-groups view, and subdivide. You can split the group by brand, product category, product type, item ID, or custom label, then flip any branch to "Excluded."
This is the fastest reversible lever. Use it to pull a seasonal collection, a low-margin line, or a single dog SKU out of a running campaign without touching your feed.
Custom labels make this scale. Tag SKUs in your feed as, say, margin_low or clearance, and you can exclude or isolate whole profit tiers in one move instead of clicking through items one at a time.
Merchant Center feed rules
Feed rules work upstream, before the product ever reaches the campaign. In Merchant Center you create a rule that sets the excluded_destination attribute (or filters the SKU out of a supplemental feed) based on a condition like price, availability, or a custom label.
This is the cleaner choice for permanent or rule-based exclusions — out-of-stock items, products below a price floor, anything that should never serve regardless of which campaign picks it up. The exclusion follows the product everywhere.
Because Performance Max is overwhelmingly a feed-driven campaign — smec, which manages feeds for hundreds of retailers, reports that 74–97% of PMax cost goes to feed-based ads — your feed is the real control surface. Clean the feed and you have cleaned most of PMax.
Brand exclusions
This one is not about products at all. Account-level brand exclusions stop PMax from serving on searches for your own brand name, including variants and misspellings, and have been available since 2024.
We will come back to why this is the exclusion that protects margin the most.
Which products should you actually exclude?
Here is the part the SERP skips. The decision is not "what is losing money in the ads tab" — it is "what is below break-even once I subtract every variable cost."
Break-even ROAS is pure arithmetic:
Break-even ROAS = 1 ÷ contribution margin.
Contribution margin for a print-on-demand SKU is price minus base cost, minus the shipping share you eat, minus payment processing. A mug that sells for $24 with a $12 base cost, $2 shipping absorbed, and roughly $1 in fees nets $9 — a 37.5% margin. That mug's break-even is 1 ÷ 0.375 = 2.67x.
Now screen your feed against that number, not against a blended ROAS target. Any SKU whose realistic ROAS sits below its own break-even is a candidate for exclusion — it is buying orders you lose money on, and PMax will happily scale it if its conversion rate looks good to the algorithm.
A worked example
Say your store does 340 orders a month at a $31 average order value, with $2,800 a month going into Google. Blended that looks like roughly $10,540 in revenue on $2,800 spend, a tidy 3.76x.
Now segment. Suppose 18% of that spend — about $504 — sits on a tote-bag line with a 28% contribution margin. That line's break-even is 1 ÷ 0.28 = 3.57x, but it is actually returning 2.9x.
At 2.9x on $504, the totes bring in about $1,462 in revenue and roughly $409 in contribution — less than the $504 you spent. You are losing about $95 a month on that line while the blended number hides it. Excluding it, or quarantining it, stops the bleed and frees the budget for SKUs that clear their own bar.
Zombie products: exclude, or rescue?
Not every low-impression SKU should be cut. Performance Max starves products with no conversion history — they look risky, get no impressions, generate no data, and stay invisible. smec defines these "zombie" products as SKUs under roughly ten clicks a month.
A zombie is not unprofitable — it is untested. Excluding it permanently throws away products that might convert if they ever got shown.
The better move is to exclude zombies from PMax and route them into a separate Standard Shopping campaign with click-oriented bidding to force data collection, then graduate the proven ones back. That rescue campaign usually runs at a worse ROAS on purpose. You are buying information, so budget it like research, not like a profit center. Our Performance Max case study walks through how that split plays out on a real account.
Brand terms: the exclusion that protects margin most
Left alone, Performance Max drifts toward your brand searches because they are the cheapest conversions in the account. It "wins" orders that your organic listings and brand Search campaign were already going to get, then reports a spectacular ROAS for doing it.
The cost is real even when the ROAS looks great. You end up paying more for a branded click than that same shopper would have cost through your own brand Search campaign. That inflated ROAS is why so many operators think PMax is crushing it while total orders stay flat.
Add the account-level brand exclusion, run a dedicated brand Search campaign, and judge PMax on its non-brand numbers. Now that PMax exposes search-terms reporting, which shipped in 2025, you can audit exactly which branded queries were leaking before you excluded them. The deeper mechanics of steering PMax without starving it live in our guide to Performance Max audience signals.
How to verify the exclusion worked
Exclusions are not fire-and-forget. Two checks keep you honest.
First, watch the budget, not just the SKU. When you exclude a product, PMax reallocates that spend somewhere — ideally to better performers, sometimes to your brand terms. Confirm the money moved where you wanted.
Second, expect a short wobble. Any material change re-enters smart bidding's learning phase for a cycle or two, so give it time before you judge. These costs matter more every year — WordStream's benchmark data shows about 87% of industries saw year-over-year CPC increases, so every dollar you stop wasting is a dollar that does not have to fight a more expensive auction. Retail is not cheap either; LocaliQ's 2026 data pegs the average retail search CPC at $4.14.
Where Victor fits
The clicks above take a few minutes. Knowing which SKUs to exclude takes true per-order profit for every product — which most sellers do not have, because ROAS ignores base cost, shipping, and fees.
PodVector AI gives you Victor, an AI employee that operates your Google Ads and Shopify together and computes true per-order profit from your live data. Victor can flag the SKUs serving below their own break-even and propose the exclusions — and because every write action is approval-gated, nothing changes in your account until you say go. Put Victor to work on your store and stop excluding products on a hunch.
If you would rather hand the whole feed-and-exclusion layer to specialists, see how we think about that in our breakdown of working with a Google Shopping ads agency, or get the full picture in the Google Ads economics guide.
FAQs
What is the difference between excluding a product and pausing it?
There is no product-level pause in Performance Max the way there is in Standard Shopping. Excluding — via listing groups or feed rules — is how you take a product out of serving. The listing-group exclusion is instantly reversible; the feed rule is cleaner for anything permanent.
Will excluding products from Performance Max hurt the algorithm?
A single material change re-enters the learning phase for a conversion cycle or two, so expect brief volatility, not permanent damage. Excluding chronic money-losers usually helps, because it stops the system from scaling unprofitable conversions. The risk is excluding untested "zombie" SKUs that simply never got a fair shot — rescue those in Standard Shopping instead of cutting them.
Should I exclude out-of-stock products?
Yes, and a feed rule is the right tool because it follows the product across every campaign. Serving ads for items you cannot ship wastes spend and risks a Merchant Center availability mismatch. Most sellers automate this with an availability-based rule rather than clicking SKUs off by hand.
Can I exclude products by profit margin directly?
Not natively — Google does not know your costs. The workaround is to tag SKUs with a custom label that encodes a margin tier, then exclude or isolate by that label. Victor computes true per-order profit from your live data, so it can tell you which tier each SKU actually belongs in before you build the label rule.
Is a brand exclusion the same as a product exclusion?
No. A product exclusion removes a SKU from the feed-driven side of PMax. A brand exclusion stops PMax from serving on searches for your brand name. Most operating stores need both: product exclusions to control what sells, and a brand exclusion to keep PMax from taking credit for orders you were already winning.