How Google Performance Max Campaigns Work and How to Scale Them

Strip away the branding and Performance Max is a bidding model with six shop fronts. You hand it creative and a definition of value. It goes shopping across Search, Shopping, YouTube, Display, Discover and the Gmail and Maps inventory, and it settles every auction on its own prediction of where the next conversion comes cheapest.

The unit of work is the asset group: headlines, long headlines, descriptions, images, logos, and video. Give it no video and Google assembles one from your images and text, which almost always looks like a template because it is one. Upload real video even if it is a plain product demo shot on a phone. Each asset group also takes an audience signal, and the word signal is doing heavy lifting there. It points the model at a starting population. It never fences the campaign in the way a Display audience does.

Retail accounts add a Merchant Center feed and split it into listing groups, which is the closest thing to real structural control the campaign type offers. Without a feed the campaign leans harder on Display and video, because those are the surfaces it can still fill.

Bidding is Smart Bidding only: Maximize conversions or Maximize conversion value, with an optional Target CPA or Target ROAS. There is no manual bid, no keyword bid adjustment, and no channel level cap. If that trade sounds unacceptable, the honest answer is that a standard Search campaign is still the right tool, and our breakdown of marketing funnels explains which stage each campaign type genuinely serves.

Performance Max harvests demand better than it creates it. That is a compliment and a warning at once. Every branded search, every direct visit, every person who already knows your name arrives cheap and lands in the campaign report, which is exactly why the offline half of a plan changes what these numbers mean.

Two practical uses come out of that. The first is creating the demand the campaign will later collect. A mailed offer, a card handed over at a counter, or a leaflet dropped in a delivery box puts a brand in front of someone who was not searching yet. Standard postcards start at $16.48 and are the cheapest physical way to reach a named list, while standard business cards at $17.57 still convert conversations into searches better than any impression does. Point each piece at its own landing URL or its own scannable code so the resulting visit is attributable rather than mysterious. Our walkthrough on building a QR code for Google reviews covers the same tracking mechanics.

The second use is keeping the report honest. If a mail drop or an event pushes branded search up, and Performance Max is allowed to buy branded queries, the campaign will take credit for the print spend. Switch on brand exclusions, keep a separate brand Search campaign, and compare the account total across the weeks a print campaign runs rather than the campaign row. The uplift is real. It just did not come from where the dashboard says.

Where print pulls its weight hardest is the moment after a digital conversion: a printed insert or a folded brochure from $57.11 in the box turns a first order into a second one, and it does it without paying a second click. Flyers from $39.54 do the same job for local footfall. Browse the marketing materials collection or start with postcard printing, and the postcard marketing ideas roundup has offers worth testing.

Most Performance Max accounts do not fail at launch. They fail two months later, when someone doubles the budget on a Monday and the cost per acquisition falls apart by Thursday.

Google already spends flexibly: a campaign can use up to twice its average daily budget on a strong day and balance out across the month. So the model is not being held back by small daily variance. It is held back by conversion volume and by creative. Raise the budget in modest steps, wait for the campaign to settle before judging the change, and never move budget and target in the same week. When you do want to tighten a Target ROAS, move it in small increments rather than one jump, because a target the model cannot hit makes it withdraw from auctions instead of trying harder.

Two changes move the needle more than budget ever does. The first is new creative: replace the assets rated Low, add a second video, and give the campaign a genuinely different angle rather than a reworded headline. The second is structure: split a bloated asset group so high margin lines stop subsidizing low margin ones, or run a separate campaign for a product family with its own economics.

Before you scale anything, use the built-in experiment tool to test Performance Max against your existing Shopping or Search setup rather than trusting the campaign to grade its own work. And check the bottleneck downstream: at triple the spend, the constraint is usually the landing page, stock, or the people answering the phone. The same discipline runs through our guide to scaling with YouTube ads, and there is more of it in the marketing strategy guides library.