SEM
Performance Max vs Search campaigns for stores
Performance Max vs Search is the decision I get asked after a store already spends on Google Ads and the Shopping tab still looks like a black box. I am a Gold Coast web developer with 18 years on catalogues. I do not run media buying as a day job. I do ship the feed, the landing pages and the conversion tags that decide whether either campaign type can learn. In 2026 that split is not a branding argument. It is a question of query control, asset groups and whether brand Search is quietly funding everything else.
This note is for merchants and developers who sit next to a Google Ads account. If you cannot see search terms, you cannot tell whether Performance Max is stealing brand or finding new customers.
What Performance Max vs Search actually means for a store
Performance Max vs Search means you are choosing how much query control you keep versus how much inventory Google can place. Google's Performance Max documentation describes PMax as a goal-based campaign that serves across Search, Shopping, Display, YouTube, Gmail and Discover from one set of assets. A Search campaign still bids on keywords you chose, with match types, negatives and a search terms report you can read on a Monday morning.
For a store the practical difference is the product. Search can send a "navy merino crew" query to a collection or a PDP you named. PMax will also do that if the feed title and landing page agree, but it will also show the same SKU on YouTube next to a lifestyle video, and on Display next to a banner that may crop the product. If the asset group is a generic brand film and a homepage URL, PMax will hunt for conversions anywhere. That is how you get branded search cannibalisation: PMax claims the cheap brand click, Search looks expensive, and nobody added demand.
I treat Search as the instrumented layer. Brand keywords sit in their own campaign so you can see CPA for people who already knew the name. Generic and product keywords sit in another, with negatives for brand and for competitor names you refuse to buy. PMax then sits beside them, not on top of them, with brand exclusions where the account allows it, and with asset groups that map to product types rather than one bucket called "everything."
The feed is the silent third campaign. If titles are "Product 18442 blue," PMax has nothing to match against a real query. Search can still win that query if you wrote an ad and a landing page. That is why I write Google Shopping feed quality for product ads as a sibling to this piece. PMax vs Search is a budget conversation that is actually a catalogue conversation.
When Performance Max vs Search is the wrong split
Performance Max vs Search is the wrong split when you copy a template from a different vertical and hope the algorithm respects your brand. Fashion, bikes, appliances and food do not share a query graph. A Magento appliance catalogue with thousands of SKUs and a Shopify sleepwear brand with a tight range need different structures even if both use Smart Bidding.
Wrong split one: turning off Search the week you launch PMax. You lose the search terms report for the queries you cared about. PMax insights are not a replacement. They are a summary. If brand volume was 40 percent of clicks, PMax will happily take it and call it efficiency. Your generic CPA looks better because the mix changed. Revenue may be flat.
Wrong split two: running Search on exact brand only and dumping every generic into PMax with a homepage final URL. The landing page then has to do work it was not built for. I write about that failure in landing pages that convert Google Ads traffic. PMax will still spend. The bounce is not PMax's fault. It is yours.
Wrong split three: one asset group for the whole catalogue. Google then mixes a $40 accessory with a $400 hero product. Conversion value bidding will chase the accessory if it converts more often. Merchandisers think PMax "does not understand the brand." The asset group never told it which products matter.
Wrong split four: using PMax as a Display dump because Search "is expensive." Display and YouTube inside PMax will find cheap sessions. If your conversion tag is leaky, those sessions look like gold. Fix measurement first. I will not diagnose a campaign mix on a tag that double-counts.
Campaign type also does not fix a PDP that hides the spec the query asked for. On retail conversion at scale paid search only paid when Magento category pages matched how people shopped appliances. On Brigade Overland a rooftop-tent query has to land on that kit, not on a generic camping story page.
How asset groups and Search campaigns should share the account
Asset groups should map to how the catalogue is merchandised, not to how the Google Ads UI is grouped by default. I want one asset group per product type or per margin band when volume allows: tents, vehicles, apparel, or whatever the store actually sells. Each group needs:
- A final URL that is a collection or a PDP cluster, not always the homepage.
- Headlines that repeat the words in the feed titles.
- Images that are product-first. Lifestyle is allowed. Lifestyle-only is how you get accidental fashion ads for a hardware SKU.
- Logos and a short brand line so YouTube does not look like a random UGC scrape.
- Audience signals as a hint, not as a prison. They seed learning. They do not replace the feed.
Search campaigns need a negative keyword list that includes junk and the brand terms you already cover. Shared lists keep PMax and Search from bidding each other up on the same brand query. Review search terms weekly on Search. Review PMax insights when Google shows them. If a category is all brand, tighten exclusions.
URL expansion is guilty until proven useful. Exclude cart, account, internal search, and Magento layered-nav URLs that duplicate a canonical collection.
Conversion goals must match the business. Add-to-cart as a primary goal trains PMax to find browsers. Purchase with a sensible value should be primary. Newsletter signups as primary will buy cheap emails and call it ROAS.
Budget: fund Search enough to capture brand and known generic terms, then give PMax enough to exit learning. Starving PMax for three days is how people conclude it does not work. Overfunding it while Search sits on leftover coins is how brand gets absorbed.
Production checklist before you move budget
A store is ready to compare Performance Max and Search when the account, the feed and the landing pages can be read in the same afternoon.
- One primary purchase conversion. Value and currency match the gateway in direction if not in exact dollars.
- Merchant Center feed approved for the SKUs you care about. Disapproved items will not save PMax.
- Brand Search in its own campaign with exact and phrase on the names people actually type.
- Generic Search with match types you can still police. Broad needs negatives, not faith.
- PMax asset groups aligned to product types. No single "all products" group unless the catalogue is tiny.
- Final URLs that resolve to 200s, indexable pages, and the same offer as the ad.
- URL expansion off until you have looked at the pages it would choose.
- Brand exclusions on PMax where the interface still offers them, plus a shared negative list.
- Listing groups or product filters so low-margin clearance is not the whole campaign.
- A two-week reading period after a structure change. Do not rebuild weekly.
If the store is mid-migration, wait. A Shopify cutover with new URLs will poison PMax learning. Redirects help SEO. They do not make last week's asset group smarter.
What breaks when PMax and Search fight
They fight when both can bid on the same query and you only look at blended ROAS. Brand terms are the usual battlefield. PMax reports a beautiful CPA. Search reports a worse CPA on leftover generic. Finance sees one number in Ads and another in the gateway. Nobody notices that branded revenue never needed PMax.
They also fight on product titles. Search ads with a tight headline lose to a Shopping card whose feed title is a keyword dump. That is not a reason to abandon Search. It is a reason to rewrite the feed. Conversely, a beautiful Search ad to a slow landing page loses to a Shopping card that loads. Core Web Vitals belong in this conversation even if the media buyer never opens Lighthouse.
Seasonality breaks both. A Magento promo calendar that changes price daily will train PMax on last week's margin. Search can pause a keyword. PMax will keep finding the SKU until you exclude it. Have a process to exclude out-of-stock and gift-with-purchase SKUs that convert as $1.
Creative fatigue looks like a PMax problem and is often a feed problem. The same five images rotate because the feed only has five. Add images at the SKU level. Do not upload a brand reel and hope.
How to measure which campaign type earned the order
Measure at the order, then at the campaign, then at the query. I start with gateway revenue by day. I then look at Google Ads conversion value by campaign for the same lag-adjusted window. I do not treat GA4 last-click as the judge of PMax. PMax is built to assist. If you only credit last click, Search brand will always look like the hero.
Use GA4 data-driven attribution as a second view, not the only view. Hold a merchandiser meeting with three slides: Ads campaign value, GA4 purchase source, gateway. When they disagree, fix tags before you cut PMax.
Search terms and PMax insights should be exported monthly, not stared at daily. You want patterns: rising competitor queries, rising brand share inside PMax, a product type that never spends. Daily tinkering resets learning.
I do not publish a universal "PMax should be X percent of spend" rule. A new store with no query history may need PMax to find anything. A mature brand with a clean Search account may keep PMax smaller and still grow Shopping. The feed and the landing page decide more than the percentage.
Related work on this site
Retail conversion at scale is the Magento operations case: year-on-year sales up $2.5 million over three years, with measurement and merchandising in the same job. Campaign mix sat on top of category pages that could actually convert. Brigade Overland is a custom Shopify catalogue after a platform move. Paid traffic to a migration that still has old URLs in asset groups is how you buy 404s. Get the storefront honest, then argue Performance Max vs Search with numbers that mean something.
FAQ
Is Performance Max vs Search a choice of one campaign type?
No. Performance Max vs Search is usually a portfolio. Search holds the queries you must see. PMax covers Shopping and the other surfaces with the same feed. Turning one off to "simplify" hides the problem instead of solving it.
Will Performance Max vs Search always show PMax stealing brand?
Often, if you let it. Brand exclusions, a dedicated brand Search campaign and a weekly look at insights are how you see the steal. If you only look at blended ROAS you will congratulate PMax for capturing demand you already owned.
Can Performance Max vs Search work with a messy product feed?
Not for long. PMax is hungry for titles, GTINs and availability. Search can limp on ad copy you wrote by hand. If the feed is messy, fix the feed before you add budget to PMax. The comparison is unfair until both systems see the same catalogue.
How soon can I judge Performance Max vs Search after a restructure?
Give learning enough conversion volume, which on a small store may be two to four weeks. If you changed conversion actions in the same window, you cannot judge the campaign type at all. Stabilise tracking first.
Keep reading
Contact if you want this kind of work on a live store.