Google Ads Performance Max Best Practices: When PMax Helps, When It Quietly Destroys Your Budget, and How to Actually Control It

Google launched Performance Max with a simple pitch: feed it your assets, set a goal, and let machine learning find conversions everywhere Google runs ads. Sounds great. And for a narrow slice of accounts, it genuinely is.

For everyone else? PMax becomes a black box that burns budget on YouTube bumper ads, Display placements from 2009, and Search queries you’d have added as negatives on day one — all while Google tells you performance is “within targets” because it engineered the attribution to say so.

We’ve managed PMax campaigns across dozens of accounts since the beta. We’ve seen it double revenue for a DTC brand and quietly cannibalize $40K/month in existing Search traffic for a B2B SaaS company. The difference wasn’t luck — it was setup, structure, and knowing exactly which levers actually work. Here’s all of it.

The Honest Case For Performance Max (It’s Shorter Than You Think)

PMax is genuinely good in a specific situation: you have a well-tracked account, clean conversion data, a healthy feed (for ecommerce), and you’ve already maxed out impression share on your core Search campaigns. At that point, PMax is the logical expansion layer — it finds incremental demand you couldn’t reach through keyword targeting alone.

It’s also strong for ecommerce brands running Smart Shopping campaigns before the forced migration. If your Shopping campaigns were performing well, PMax is roughly what that evolved into, with more channels bolted on. For many of those accounts, performance held or improved.

Where it earns its keep: high-volume ecommerce with good ROAS history, lead gen accounts with verified conversion tracking on high-intent actions, and brands expanding into new markets where they lack the keyword research to build traditional campaigns from scratch.

That’s it. If your account doesn’t fit that profile right now, you’re not getting the best of PMax — you’re getting the worst of it, which is a campaign type that optimizes confidently toward the wrong thing.

Why PMax Quietly Cannibalizes Your Best Campaigns (And How to Catch It)

Here’s the mechanic Google doesn’t put in the help docs: when PMax and a standard Search campaign are both eligible for the same query, PMax wins the auction almost every time. Google’s official line is that “existing campaigns take priority for exact match queries on identical keywords.” In practice, that exception is narrower than it sounds.

What this means for your account: PMax will serve on queries your Search campaigns were already winning, claim the conversion, and make your Search campaigns look like they’re declining organically. Your Search ROAS drops. You pause spend. PMax looks like the hero. This is not a conspiracy theory — it’s a structural incentive baked into how PMax reports performance.

How to catch it: Pull your Search campaign impression share and conversion volume for the 60 days before and after PMax launch. If Search impression share dropped without a corresponding CPM increase, PMax ate your lunch. Also look at your branded query costs in the Search terms report — if PMax is bidding on your brand name, you’re paying for traffic you’d have gotten for a fraction of the cost through a branded Search campaign.

The fix is a brand exclusion list applied to PMax (available via Google support ticket for now, though the interface control is being rolled out) and ensuring any high-intent, high-converting keyword themes you own are locked down in exact or phrase match Search campaigns before PMax goes live.

PMax Asset Groups: Stop Treating Them Like Ad Groups

Most accounts we audit have one PMax campaign with one asset group. Everything in the product catalog, one pile of headlines, one pile of images, one audience signal that says “website visitors.” This is the fastest way to get average results from a campaign type that punishes average inputs.

PMax asset groups are your primary lever for telling the algorithm what matters and to whom. Think of them less like ad groups and more like mini-campaigns with their own creative brief and audience thesis.

Here’s how we structure them for a mid-size ecommerce account:

• One asset group per major product category — separate headlines, descriptions, images, and audience signals for each. A hiking boot asset group should have copy about trail grip and waterproofing, not generic “great shoes” language.
• High-margin products get their own asset group — even if they’re in a category group, the SKUs that drive profitability deserve dedicated creative and an audience signal built from purchasers of that product type.
• Seasonal or promotional pushes live in their own asset group — not mixed into evergreen groups where performance data will never be clean.

The practical upside isn’t just creative relevance. When an asset group is tightly themed, the algorithm learns faster because the signal is cleaner — “people who clicked this asset group and converted” is a more coherent audience seed than “people who clicked anything in this campaign.”

Review asset group performance labels weekly. Google will tell you which individual assets are “Low,” “Good,” or “Best.” Cut “Low” assets after 4–6 weeks. Don’t get sentimental about a headline you wrote — if it’s dragging down the group, it goes.

Audience Signals: The One Input That Actually Shapes Where PMax Goes

PMax doesn’t let you target. You probably know this. What you might not realize is that audience signals are not targeting — they’re more like a directional nudge — but a well-constructed signal can dramatically change how the algorithm allocates spend in the early weeks before it has enough data to optimize on its own.

The signals that actually move the needle, ranked by impact:

1. Customer match lists from your CRM — your actual past buyers. Upload a list of customers who converted in the last 90–180 days. PMax will model lookalikes from this and it’s consistently the strongest signal we’ve tested.
2. Website visitors who converted — narrow, high-intent. Apply this to your highest-margin asset groups.
3. Custom intent audiences built from competitor URLs and relevant search terms — underused and effective, especially for B2B lead gen where in-market segments are broad and noisy.

What doesn’t work: dumping your entire “all website visitors” list as the signal on every asset group. That’s telling the algorithm “optimize toward anyone who ever came to our site,” which is so broad it might as well be no signal at all. Be specific. The tighter the signal, the faster PMax finds a productive lane.

The PMax Optimization Levers Google Actually Gives You (Use All of Them)

People complain PMax is a black box with no controls. That’s half-true. The controls are real — they’re just not surfaced prominently because using them well means spending less, which isn’t in Google’s interest.

URL expansion controls: By default, PMax will send traffic to whatever landing page it decides is most relevant. Turn off URL expansion unless you’ve audited every page on your domain and are confident PMax won’t send paid traffic to your careers page or a 404. Apply URL exclusions aggressively. This alone can clean up a significant amount of wasted spend.

Placement exclusions: PMax inherits account-level placement exclusions. Go to your account-level settings and add a placement exclusion list that blocks mobile app categories, games, and parked domains. This won’t eliminate all garbage Display placements, but it cuts the worst offenders.

Bidding targets: If you’re running Target ROAS, set it at least 10–15% above your actual acceptable floor — not your aspirational target. PMax will optimize to the number you give it. If you set tROAS at 200% but would be happy with 150%, you’re telling the algorithm to leave conversions on the table. Conversely, if you set it too low, it’ll buy volume at the expense of margin. Know your actual floor and work from there.

Seasonality adjustments: Use these before major promotions. If you know your conversion rate spikes 40% on Black Friday, tell the algorithm. Without a seasonality adjustment, PMax will spend the first day of your sale learning that conversion rates are higher than expected — and you’ll have wasted your best hours.

Budget and campaign priority: Run PMax on its own dedicated budget. Don’t share it with Search campaigns in a shared budget. PMax will eat whatever it’s given, and that pooled budget setup makes it impossible to evaluate where efficiency lives.

When to Turn PMax Off (Seriously)

This is the section most agencies won’t write because they don’t want to talk clients into spending less on a campaign type Google actively promotes. But here it is.

Turn PMax off — or don’t launch it in the first place — if:

• Your account gets fewer than 30–50 conversions per month total. PMax needs conversion data to optimize. Below that threshold, it’s making decisions on statistical noise. A tightly structured Search campaign will outperform it every time.
• Your conversion tracking is unreliable. If you’re firing conversion events on page views, or your form submission tracking has a known duplication issue, PMax will optimize toward those phantom conversions enthusiastically.
• You sell a complex B2B product with a 90-day sales cycle and you’re only tracking form fills. PMax will find form fills. Many of them will be unqualified. It has no way to know what happened after the click.
• Your brand campaigns are your primary revenue driver and you haven’t set up brand exclusions yet. Launch PMax without brand protection and you’re essentially paying premium rates to serve ads to people who were already searching your name.

Pausing PMax isn’t failure. It’s an allocation decision. For the accounts where Search and Shopping campaigns are performing efficiently and your budget is limited, staying focused on what you can control is often the higher-ROI move.

Frequently Asked Questions

How is Performance Max different from Smart Shopping?

Smart Shopping was essentially Google Shopping plus Display remarketing, automated. Performance Max includes all of that, plus Search, YouTube, Discover, Gmail, and Maps. It also gives you more creative inputs through asset groups and more audience signal options. The core automation logic is similar — Google optimizes toward your conversion goal — but PMax has a much wider reach and, accordingly, more places to waste budget if it’s not structured well.

Should I run PMax and Search campaigns at the same time?

Yes, but with guardrails. Run your highest-intent, highest-value keyword themes in exact and phrase match Search campaigns and let them take priority. Use PMax as an expansion layer for incremental reach, not as a replacement for Search. Monitor impression share on your Search campaigns after PMax launches — if it drops, PMax is cannibalizing and you need brand exclusions and possibly tighter keyword exclusions on the PMax side.

How many asset groups should a PMax campaign have?

As many as you have meaningfully different product or service themes — and not one more. For a small ecommerce brand with three product categories, three to five asset groups is right. For a large retailer, you might have 15–20. The wrong answer is one asset group for everything, and the other wrong answer is 50 asset groups that are so granular the algorithm never accumulates enough data to optimize any of them. A useful rule of thumb: each asset group should be able to generate at least 20–30 conversions per month on its own.

Why does PMax report great ROAS but my overall revenue isn’t growing?

This is the cannibalization problem in numbers. PMax is likely claiming credit for conversions that were already happening through other channels — branded Search, direct, organic. The campaign’s reported ROAS looks strong because it’s measuring last-click or data-driven attribution that often favors the most recent Google touchpoint. Pull your total account conversion volume and revenue before and after PMax launch. If overall numbers are flat while PMax reports stellar results, it’s almost certainly taking credit rather than generating incremental value.

What’s the minimum budget for Performance Max to work?

There’s no official minimum, but in practice, PMax needs enough budget to generate 50+ conversions per month within the campaign to optimize meaningfully. In most markets, that means at least $3,000–$5,000/month — and in competitive B2B verticals with CPCs above $20, often significantly more. Running PMax on $500/month means the algorithm will spend most of the month in a learning phase and most of your budget testing placements rather than exploiting what works.

Does Google’s Search term report work for PMax?

Partially. Google shows Search category insights under the “Insights” tab for PMax, which gives you a rough sense of query themes, but not individual search terms the way the standard Search terms report does. This is a real limitation and one of the legitimate criticisms of PMax transparency. The workaround is to cross-reference your PMax Search insights with your Search campaign search term reports — if you see query themes appearing in PMax insights that you’d normally exclude, add them as negative keywords at the account level or campaign level where available.