The simple split
When it comes to Shopping vs. Performance Max (PMax), our approach is straightforward: Shopping is the foundation, PMax is the complement. The goal isn’t to let PMax dominate, but to let Shopping cover the bulk of demand while PMax brings in additional wins where it makes sense.
Shopping: your foundation
We treat Shopping with the same mindset you’d apply to SKAGs in search campaigns (not literally, since Shopping doesn’t work that way). The principle is to segment clearly so you get both control and visibility.
That means grouping products in a way that mirrors how people shop: by category, sub-category, and key variants. On top of that, we introduce performance tiers so high performers, mid-range products, and long-tail items aren’t forced into the same bucket.
The result is not a single campaign where the algorithm makes all the decisions, but a clean and structured account that gives you a clear line of sight on performance. We typically run Shopping on a lower target ROAS (tROAS), allowing it to capture the largest share of demand and deliver steady, predictable revenue. In other words, it’s your dependable base layer – reliable, transparent, and easy to manage.
PMax: the controlled layer
PMax has its place, but it needs guardrails. Left unchecked, it will soak up the budget without necessarily driving incremental value. That’s why we keep it constrained.
We run feed-only PMax campaigns, apply inventory filters, set a higher tROAS than Shopping, and limit it to products or categories that have already proven they can perform. The aim is to let PMax operate with precision, adding incremental revenue rather than simply cannibalising Shopping.
If results are strong, we gradually expand its scope, one step at a time, not all at once. This way, it earns its place in the account rather than being handed the keys from the start.
How we test
When possible, we split the catalogue into two comparable sets:
- Set A: everything in PMax (the “all in” approach).
- Set B: Shopping as the base + constrained PMax.
If splitting the catalogue isn’t realistic, we use a geo-split across similar regions. Either way, budgets remain comparable, promotions stay aligned, and we run the test long enough to gather useful data.
What we measure:
- Incremental revenue and profit – is the split actually driving more value?
- Coverage – are the right products showing where they should?
- Incrementality of PMax – is it adding new sales, or just shifting them?
If PMax is simply moving sales around, we tighten its restrictions or scale it back. If it’s driving genuine incremental growth, we allow it more room.
Why this approach works
This split keeps both control and opportunity in play:
- Shopping gives you structure, visibility, and a stable flow of revenue.
- PMax provides additional reach, but only where it proves effective and within efficiency thresholds you’re comfortable with.
Together, they deliver a balance of stability and growth, without leaving you at the mercy of a single “black box” setup.
The takeaway
You don’t need to choose between full automation and full control. Use Shopping to anchor your account and capture demand. Use PMax selectively, with clear boundaries, and let it earn more responsibility as results justify.
That’s how you test account structures properly, scaling efficiently, while keeping performance measurable and sustainable.
