If you came from years of running Search campaigns, the absence of a keyword box feels like flying blind. It is not. PMax is still competing in the same query auctions your Search and Shopping campaigns live in — it just picks the queries itself. Understanding what replaces keywords is the difference between a campaign you can diagnose and one that quietly eats your margin.
So if there are no keywords, how does PMax decide what to show?
Performance Max reads three inputs and infers intent from them: your Merchant Center product feed, the asset group (images, video, headlines, descriptions), and your final landing pages. For retailers, the feed does most of the heavy lifting — smec, which manages ad spend for hundreds of retailers, reports that 74–97% of PMax cost goes to feed-based (Shopping-style) ads. That means your product titles are doing the job keywords used to do.
This is the single most important mental shift. In Shopping and PMax, there is no keyword to match against — the product title is the matching surface. A title that reads "Ceramic Mug" enters far fewer auctions than "Funny Cat Lover Coffee Mug 11oz Ceramic Gift." Front-load product type, defining attributes, and the gift occasion if that is how people search.
Search themes: the closest thing to keywords
Google did add one keyword-adjacent lever. Search themes let you tell PMax which queries you already know your customers use. You can add up to 50 search themes per asset group, and they act as signals that are additive to what PMax already predicts — they do not directly trigger ads the way a keyword does.
Here is the nuance most articles skip: search themes slot into the same prioritization as your phrase and broad match keywords, and exact-match keywords in your Search campaigns still take priority over them (Google Ads Help). So a search theme is a nudge, not a command. Use it to point PMax at category terms it might otherwise miss, not to force it onto a specific query.
Can you add negative keywords to Performance Max?
Yes — and this is the control that matters most for protecting margin. Campaign-level negative keywords for PMax shipped in the 2025 update wave, alongside full search-terms reporting (The Media Image). The old "PMax is a black box" complaint is stale. You can now see the actual queries PMax bought and exclude the junk.
The workflow for an operating store is simple: pull the PMax search-terms report monthly, flag the irrelevant or informational queries ("how to clean a mug," "free svg"), and add them as negatives. We walk through this in detail in our guide to adding negative keywords to Performance Max. This is not set-and-forget; it is a recurring audit.
Brand exclusions: the keyword control nobody mentions
There is a second exclusion that quietly decides whether your PMax ROAS is real. PMax chases the cheapest conversions available, and your own branded searches are the cheapest of all. Left unchecked, it absorbs traffic that was already coming to you, inflates its own reported return, and starves your brand Search campaign.
Account-level brand exclusions for PMax have been available since 2024 and cover brand variants and misspellings (AdNabu). Turn them on, run a dedicated brand Search campaign, and judge PMax on its non-brand performance. smec frames the cost of not doing this bluntly: you end up paying "$1.50 for a click that you could have bought for $0.20" (smec).
Does PMax outrank my keyword campaigns?
Not automatically — and this is where the "no keywords" question gets practical. Google has confirmed that PMax and Standard Shopping in the same account compete on Ad Rank; the higher Ad Rank wins the impression (smec). The old behavior where PMax always beat Shopping is retired.
For Search specifically, Google's own framing is that PMax ads behave like keywords that don't exactly match the query. An exact-match keyword in your Search campaign will still win its own term. So running a tight brand Search campaign isn't redundant — it is how you keep your cheapest conversions out of PMax's hands.
This is why the field no longer treats "PMax vs keywords" as either/or. Optmyzr's study of 24,702 PMax campaigns found 82% of advertisers run PMax alongside Shopping or Search, and split-budget accounts posted the strongest returns. The job is portfolio construction, not replacement — a theme we cover in our guide to optimizing a Performance Max campaign.
What you steer instead of keywords: a worked example
Since you can't bid on keywords, your real levers are feed quality, your tROAS target, and exclusions. Let's put numbers on why the target matters more than any keyword ever did.
Say you run a POD store doing 340 orders a month at a $31 average order value, spending $2,800/month on Google. Your mugs cost $9 to fulfill, shipping eats another $4, and payment processing takes roughly $1.20 per order. That leaves a contribution margin of:
$31 − $9 − $4 − $1.20 = $16.80 per order, or about 54% of the price.
Your break-even ROAS is simply 1 ÷ 0.54 = 1.85x. So a tROAS target of 185% is the line where ads stop making money. If you want a profit buffer, you might set the target to 230%.
Here is the trap keyword-trained sellers fall into. You see a "good ROAS" blog benchmark of 500% and set tROAS to 500%. Google's own documentation warns that a target far above what the account has historically achieved "may limit the amount of traffic your ads may get" — the system simply declines auctions it predicts won't clear the bar. Your spend collapses, and you blame PMax. In reality, raising the target is a volume decision disguised as an efficiency decision.
The feed is your keyword list
Because there is no keyword box, feed quality is load-bearing. Complete, accurate attributes — descriptive titles, valid GTINs, correct product types, price parity with the landing page — determine which auctions your products enter at all (Google Merchant Center Help). A disapproved or thin-titled bestseller is invisible in PMax no matter how much you spend.
For context on retail click costs, LocaliQ's 2026 benchmarks put the average retail-category Search CPC at $4.14 — directional context, not a number to target. At a $31 AOV and a 3% conversion rate, that CPC is survivable; at a $12 AOV it is ruinous. Your break-even math decides, not the benchmark.
How this compares to Shopping and marketplace ads
If keyword control is what you miss, Standard Shopping gives more of it back: product-group bids, campaign priorities, and full search-term control. Many operating stores run a hybrid — PMax for scale, Standard Shopping for the SKUs and queries they want to steer by hand. If you also sell on marketplaces, the matching logic differs again, which we compare in our breakdown of Amazon versus Google Shopping ads.
For the full picture of how clicks, conversions, and margin fit together across every Google campaign type, start with our Google Ads economics guide.
Where Victor fits
Reading the PMax search-terms report, reconciling Google-reported conversion value against your actual Shopify orders, and catching a brand-cannibalization drift before it eats a month of budget is exactly the recurring work that gets skipped when you're running a store solo.
Victor is the AI employee inside PodVector AI. Victor connects to your Shopify store and Google Ads, computes your true per-order profit — not just ROAS — and flags when a campaign's reported return isn't matching real orders. Every write action Victor takes is approval-gated: Victor drafts the move, you approve before anything executes. Put Victor to work on your store.
If you're leaning toward handing Google Ads to a specialist instead, our take on when a Google Shopping ads agency is worth it walks through the tradeoffs.
FAQs
Does Performance Max use keywords at all?
No. There is no keyword list, match type, or keyword bidding in a PMax campaign. PMax infers which queries to enter from your product feed, asset group, and landing pages. The closest lever is search themes, which act as optional signals rather than keywords that directly trigger ads (Google Ads Help).
What are search themes and how are they different from keywords?
Search themes are hints you give PMax about queries you know your customers use — up to 50 per asset group. Unlike keywords, they don't directly trigger ads; they're additive signals on top of what PMax already predicts, and they carry the same priority as phrase and broad match (Google Ads Help). Exact-match keywords in your Search campaigns still win their own terms.
Can I add negative keywords to Performance Max?
Yes. Campaign-level negative keywords for PMax shipped in 2025, along with full search-terms reporting (The Media Image). Pull the search-terms report monthly and exclude irrelevant or informational queries so you stop paying for clicks that never convert.
Will Performance Max steal traffic from my keyword campaigns?
It can. PMax chases the cheapest conversions, and your branded searches are the cheapest of all, so it tends to absorb them and inflate its own ROAS. The fix is account-level brand exclusions, available since 2024 and covering variants and misspellings (AdNabu), plus a dedicated brand Search campaign. PMax and Shopping otherwise compete on equal Ad Rank footing (smec).
If there are no keywords, how do I control what PMax targets?
You steer it through feed quality (titles, GTINs, product types), your tROAS target, search themes, negative keywords, and brand exclusions. For retailers the feed is the dominant input, since the majority of PMax spend runs through feed-based ads (smec). Treat your product titles as the keyword list you no longer get to write directly.
Should I use PMax or stick with keyword-based Search and Shopping?
Most operating stores run both. Optmyzr's study of nearly twenty-five thousand campaigns found the large majority of advertisers use PMax alongside Shopping or Search, with split-budget accounts performing best (Optmyzr). Use Standard Shopping for the SKUs and queries you want hands-on control over, and PMax for scale — then judge PMax on non-brand results and true per-order profit, not headline ROAS.