People have strong feelings about Performance Max. I've had clients tell me it's the only campaign they need, and others who won't let me go near it because someone burned their budget with it once.
I don't think either camp is right. PMax is a tool. It's very good at some jobs and fairly bad at others, and most of the problems I see come from using it before the account is ready for it.
What it actually does
Performance Max is a single campaign that can show your ads across nearly all of Google: Search, Shopping, YouTube, Display, Gmail, Discover and Maps. You give it a budget, a goal and a set of images, videos and text, and Google decides where each impression goes.
The important part is the goal. PMax will chase whatever you've told it counts as a conversion. It doesn't know your margins. It doesn't know your sales team thinks half the leads are rubbish. It only knows what the conversion tag tells it.
To be fair to Google, you can see a lot more inside PMax now than you could when it launched. There are search term reports, channel reporting, stats for individual assets, negative keyword lists at campaign level, and device and age settings. It's much less of a black box. But it still optimises toward the target you set, not the one you had in mind.
What I check before switching it on
First, I look at the tracking. If there are duplicate tags, or small actions like newsletter sign-ups set as primary conversions, PMax will find the cheapest way to get more of them. I fix measurement before anything else, every time.
Then I look at volume. Automated bidding needs enough conversions to learn from. If an account only gets a handful a month, results tend to bounce around and it's hard to tell what's happening. In that situation I'd usually stay with tightly run Search campaigns for a while.
Next, creative. PMax runs on YouTube and Display as well as Search, so it needs decent images and ideally some video. If the only assets are a logo and a couple of stock photos, it's going to struggle on those channels.
Last, brand searches. Left alone, PMax will happily show ads to people searching for your company name. Those people were probably going to find you anyway, and it makes the campaign look better than it really is. I add brand exclusions and keep an eye on the search terms.
If any of those aren't in place, PMax isn't really the question yet.
Where it tends to work
Online shops with a good product feed are where I've seen it do best. If the Merchant Center feed is clean, the titles make sense and you know your margins, PMax can find buyers that a Search-only setup would miss.
It's also useful once an account has more or less run out of room on Search. If you're already showing for most of the searches that matter and want more, PMax can reach people a bit earlier.
Local businesses can do well with it too, as long as calls, directions and store visits are being measured properly.
And it gets much better when you can feed it real results from your CRM, like which leads turned into sales. That gives it something worth chasing.
Where I'm more careful
Lead generation is the big one. If every form fill counts the same, PMax will find you lots of cheap leads, and a fair few of them won't be any good. Unless the sales outcome is coming back into Google Ads, I keep most of the budget in Search.
Small budgets are another. Spreading a small daily budget across seven channels usually means none of them gets enough to learn properly.
I'm also cautious with niche B2B. When there are only a few hundred companies that could ever buy from you, Search and LinkedIn give you much more control.
How I set it up
When it does fit, I keep it simple. One clear goal per campaign. Asset groups split by things that really are different, like product categories or price ranges, rather than one big group with everything thrown in. Audience signals from our own data where we have it. Brand exclusions on, and Search running alongside it so the brand and the most important keywords stay under direct control.
Then I leave it alone for a few weeks. Judging PMax after three days is like judging a new hire after one morning.
My view
I don't see it as a magic button, and I don't think it's a scam either. It's very good at amplifying whatever you give it. Give it clean tracking and decent creative, and it can do great work. Give it messy tracking and you'll get messy results, faster.
So if you're wondering whether you should be running it, start by asking whether your tracking is telling the truth.