Your PMax “Remarketing” Isn’t Actually Remarketing

Most advertisers, when they paste a Customer Match list or a 30-day add-to-cart audience into Performance Max, think they've built a remarketing strategy. They haven't. What they've built is a signal, and the gap between a signal and actual targeting is where a lot of money quietly leaks out.

Signal vs. targeting: the distinction nobody explains clearly

In Performance Max, audience signals are hints. Google uses your remarketing list as a seed: the algorithm studies the people on it, figures out who they look like, and then goes out and spends most of the budget on people Google decides behave similarly. You cannot disable this expansion. There's no checkbox, no advanced toggle, no premium tier. Practitioners have been asking Google for hard audience targeting in PMax since the product launched in 2021, and Google has been clear it isn't coming.

In a Standard Shopping campaign, the same remarketing list works very differently. You set it as Targeting, not Observation, and Targeting is a hard filter. Ads serve to people on the list, period. That's the behavior most advertisers think they're getting when they upload a list to PMax. They're not.

Why a dedicated Shopping remarketing campaign wins

Different ROAS economics

Say your add-to-cart 30-day pool runs at 10-12x ROAS. In a dedicated remarketing campaign you can set tROAS at 800-1,000% and Smart Bidding will chase it. In PMax that same warm pool gets averaged into one campaign-wide tROAS together with cold prospecting traffic. Warm gets systematically under-bid. Cold gets systematically over-bid. You don't see it because the campaign-level number looks fine.

Budget guarantee

PMax allocates spend across goal types and audiences however the algorithm decides. You cannot reserve $X for warm traffic. A dedicated remarketing campaign guarantees every dollar goes to people who already added to cart, browsed your category pages, or bought before.

Real reporting

A dedicated campaign tells you exactly what your warm cohort did: impressions, clicks, conversions, ROAS, frequency capping. PMax gives you one aggregate "audience signal contribution" number and the rest is a black box.

Bid strategy flexibility

Standard Shopping supports Manual CPC, Maximize Conversions, Target ROAS, and Maximize Conversion Value. Pick the right tool for the job. PMax is locked into whatever goal type you picked at setup.

It avoids the "warm funds cold" trap

This is the most common complaint about PMax with remarketing inside, and it's a real one. Smart Bidding sees high conversion rates on list users and uses that profit to fund aggressive bids on cold lookalikes. You end up paying more for cold acquisition than you would have with clean cold prospecting, and the PMax campaign-level ROAS still looks healthy because the warm cohort is propping the average up. The only way to see the leak is to slice it apart by audience, which is exactly what PMax doesn't let you do.

When keeping it inside PMax is actually fine

There are real reasons to not break remarketing into its own campaign:

Your remarketing pool is small (under about 1,000 active users a month) and a dedicated campaign would starve Smart Bidding of training data. Or you don't have enough total conversion volume to train two separate Smart Bidding campaigns properly. Or you specifically want cross-channel reach (Search, Shopping, YouTube, Display, Gmail, and Discover all from one campaign) and you're willing to give up control to get it.

For small accounts or accounts still warming up, PMax with an audience signal is a fine starting point. For an account with real volume and a clear high-ROAS warm cohort, it's not. You're leaving the most valuable lever on the table.

A practical structure to test

Take your existing Shopping (or PMax) campaign and treat it as prospecting. Use campaign-level audience exclusions to keep the warm audiences out of it. Then spin up a small dedicated Standard Shopping campaign with the remarketing audiences set as Targeting, typically Add-to-Cart 30-day and All-Site-Visitors 30-day to start. Set tROAS to 700-800%. Give it 30 days. Compare segment-level ROAS to the period before the split.

The math almost always favors the split when there's enough warm traffic to support it. Even modest reallocation, say 10-15% of total Shopping spend moved into the dedicated remarketing campaign, typically nets a meaningful blended ROAS lift, because every dollar in that bucket converts at a multiple of the campaign average.

The bigger point

PMax is a powerful tool. It is not, however, a remarketing tool. Treating it like one is one of those quiet structural mistakes that doesn't show up in your dashboards until you look at the account from the right angle. If you have a warm audience worth talking to, give it a campaign of its own.

I've done this split on accounts across a range of spend levels. The pattern holds: once there's enough warm volume to train Smart Bidding on its own, the dedicated remarketing campaign almost always outperforms the mixed-PMax setup for the same warm cohort.