The most effective way to approach Performance Max is with what I call the "Shopping-First" framework, a model Sam Piliero details on the 2X eCommerce Podcast. It means you treat PMax as a Shopping campaign first and foremost, and your main job is to guide the algorithm to spend your budget there, because that's where the highest-purchase-intent customers live.
The core problem is that PMax is essentially a black box. As Sam Piliero explains, a single PMax campaign pulls from all five of Google’s major platforms: Shopping, Search, Demand Gen, YouTube, and Display. On Ecommerce Playbook, Tony Chopp calls this Google getting "free reign" to place your ads wherever its algorithm sees fit. Without careful management, you can end up spending a lot of money on low-converting YouTube or Display placements when what you really want is to dominate the product-related search results that Shopping owns.
This is where Sam Piliero's framework comes in. He argues that Shopping is the center of the universe for e-commerce on Google, and every other channel should be seen as supportive of it. Your first step is to perfect your product feed, since that’s the fuel for Shopping. Your titles, images, and data must be highly optimized. Then, you structure your campaigns and asset groups to constantly push the algorithm toward winning Shopping placements. You’re actively managing the machine, not just handing over the keys.
A critical next step in a good Google Ads strategy is separating your brand and non-brand campaigns. Tony Chopp and Luke Austin discuss this on Ecommerce Playbook. Your "Brand PMax" campaign will capture people already searching for you. It will look like it has an amazing ROAS, but these sales were likely coming anyway. They estimate branded search is only about 30% incremental. Your "Non-Brand PMax" campaign is your growth engine for finding new customers. Separating them lets you properly evaluate performance and allocate your budget for real growth, not just harvesting existing demand.
So where does this Shopping-first model break down? Its main weakness is that by trying to force PMax into a bottom-of-funnel box, you might be fighting the algorithm