Est. reading time: 6 minutes
The Performance Max structure question usually arrives as “should we run one campaign or several,” asked as if it were a matter of style. It isn’t. It’s a volume question with a control question attached, and getting it backwards is expensive in a specific way. Every campaign you add divides your conversion signal, PMax learns from that signal, and five campaigns with ten conversions each will lose to one campaign with fifty essentially every time. Structure should be the simplest arrangement your data supports, expanded only when a real control problem demands it.
One terminology correction before the framework, because the vocabulary in circulation is muddled. There’s no “single asset structure” in Google Ads. The hierarchy is campaign, then asset groups, then assets. The campaign holds the goal, the budget, and the bid target. Asset groups are themed bundles of creative, audience signals, and (for ecommerce) listing groups from your feed, all sharing the campaign’s budget and target. The real decision is which of those two levels your segmentation belongs at.
Performance Max campaign structure: why one campaign usually wins
PMax makes three decisions at once, which channel to serve on across Search, Shopping, YouTube, Display, Discover, Gmail, and Maps, which user to serve, and what to bid. All three run on conversion data, and consolidation is what gives the system enough of it to decide well. An account below roughly 50 to 60 conversions a month, our working threshold, belongs in a single campaign, full stop, and plenty of accounts above it do too.
Consolidation doesn’t mean one undifferentiated blob, because asset groups exist to do the segmentation without splitting the signal. Build one per genuinely distinct theme, meaning a product category, a customer motivation, or a service line that needs its own message and landing page, each with full creative coverage and its own audience signals. Two disciplines keep this layer honest. First, signals are suggestions rather than targeting, a starting point the algorithm expands from, so duplicated asset groups distinguished only by audience signal converge to the same performance and waste your time. Second, more than four or five asset groups in one campaign is usually a sign the account is smuggling in complexity it should either delete or promote to a real campaign split for a real reason.
When multiple campaigns earn their keep
Budgets and bid targets live at the campaign level, and that fact generates the entire legitimate list of reasons to split. Different ROAS or CPA targets, most commonly because margin tiers differ, since a single target asked to serve 60 percent margin products and 10 percent margin products will default to the easy sellers and starve the rest. Budget protection, when a category or service line must be guaranteed its spend rather than left to compete. Different geographies or languages. A dedicated new-customer acquisition campaign using PMax’s new-customer settings, separated from the campaign harvesting existing demand. And genuinely distinct business lines with different economics.
Notice what’s not on the list. Splitting to feel sophisticated, splitting because a case study had seven campaigns, and splitting past the point where each campaign clears the volume floor. Segmentation should solve a control problem you can name, and if you can’t name it, the asset group level almost certainly handles it. When you do split, the volume math travels with you, meaning a split is affordable when each resulting campaign still generates enough monthly conversions to learn from, and premature otherwise.
The controls that matter more than the split
Structure gets the attention, but the inputs decide the outcome, and PMax has grown real controls worth using. For ecommerce, the product feed outranks everything, since titles, attributes, and images determine how products match queries across every surface, and no structure compensates for a thin feed. Brand deserves a deliberate arrangement, meaning a dedicated branded Search campaign for your own terms plus brand exclusions in PMax so the automated campaign stops taking credit for demand you already owned. Campaign-level negative keywords, after years as the most requested feature in the format, are now available and should be built before launch rather than after the first bad search terms report. Search themes give each asset group up to 25 keyword themes of directional input, worth setting for anything niche.
Then there’s the reading. Channel-level reporting now shows where PMax actually spends across Search, Shopping, YouTube, and Display, which turns “the black box is fine, ROAS is fine” into a real diagnosis when budget migrates to a channel that isn’t converting. Asset ratings flag which creative is dragging, and creative supply is a standing obligation rather than a launch task, because the system can only rotate what you’ve fed it, which is the argument we made in why one winning ad can’t carry your account. Landing pages are inside the system too, both because final URL expansion will route traffic where it predicts conversions and because the page is half of every conversion rate you’re optimizing, territory we covered in multivariate testing on landing pages.
Lead generation accounts have one extra obligation, closing the quality loop. PMax optimizing on raw form fills will cheerfully find you cheap, worthless leads, so import offline conversion outcomes from your CRM so the system learns what a qualified lead looks like, and put real nurture infrastructure behind the leads themselves, the kind of ActiveCampaign workflows that turn a form fill into a customer record instead of a spreadsheet row.
The decision, compressed
Count your monthly conversions. Under the volume floor, run one campaign and segment with asset groups. Above it, split only where a campaign-level setting genuinely must differ, meaning targets, budgets, geos, or acquisition goals, and stop adding campaigns the moment the next split would drop any of them below the floor. Then leave the structure alone, because PMax re-stabilizes slowly after structural changes, and spend the recovered attention on the feed, the creative, the negatives, and the channel report. The advertisers winning with Performance Max in 2026 aren’t the ones with the cleverest structure. They’re the ones whose structure is boring enough that the inputs get all the work.
