Performance Max is the campaign type Google most wants you to run, and the one most likely to quietly waste your budget. Both things are true simultaneously, and the difference between them is almost entirely down to how well your conversion tracking works.
What PMax actually is
A single campaign that places ads across Search, Display, YouTube, Gmail, Discover and Maps. You supply assets — headlines, descriptions, images, video, a product feed — plus a goal and a budget. Google's machine learning decides who sees what, where and at what bid.
You give up granular control. In exchange you get access to inventory that separate campaigns cannot easily reach, and an algorithm that is genuinely good at finding buyers — provided it can tell what a buyer looks like.
The core problem
PMax optimises toward the conversions you report. It has no independent way of knowing whether those conversions are valuable. So it will relentlessly, efficiently pursue whatever you have told it to count.
If you count “visited the contact page” as a conversion, PMax will buy you a great many cheap contact page visits from people who will never enquire. If you count newsletter signups and sales equally, it will chase signups, because they are easier. The algorithm is not making a mistake — it is doing exactly what you asked, and your instructions were wrong.
The rule: do not launch PMax until conversion tracking is verified end to end and you are counting only actions that genuinely represent business value, with sensible values attached. In a poorly tracked account, PMax is a very efficient way to spend money on nothing.
Before you launch: a checklist
- Verify every conversion action fires once, on the right event. Test it yourself. Double-counting is common and skews everything downstream.
- Assign realistic values. A form submission worth ₹5,000 in expected profit and a brochure download worth ₹200 should not be equal.
- Mark only the real ones as primary. Everything else is secondary and observational.
- Accumulate at least 30 conversions in 30 days on the account before expecting PMax to perform. Below that the algorithm has nothing to learn from.
- Add brand as a negative at account level so PMax does not take credit for people who were already searching for you.
Setting it up properly
Asset groups are your only real targeting lever
Build one asset group per distinct product or service, each with its own headlines, images and audience signals. One asset group covering everything gives the algorithm no way to match message to intent, and is the most common structural fault we see.
Audience signals are hints, not limits
Feed in your customer list, website visitors, and custom segments built from the search terms you know convert. PMax treats these as a starting point rather than a boundary — good signals meaningfully shorten the learning period.
Supply real video
If you do not upload video, Google auto-generates one from your images and text. It is invariably poor and it will run on YouTube representing your brand. Even a simple 15-second clip shot on a phone is better than the machine's version.
Exclude what you cannot serve
Location exclusions, brand exclusions, and for e-commerce, out-of-stock and low-margin products removed from the feed. PMax will otherwise happily spend on all of them.
The reporting problem
PMax reporting is deliberately limited. You cannot see full search term data, you get restricted placement visibility, and channel-level breakdown is partial. This is Google protecting the black box, and it is a legitimate reason for scepticism.
Two workarounds worth knowing:
- The Insights tab exposes search category themes — coarser than a search terms report, but better than nothing
- Scripts exist that surface asset-group-level placement data the interface does not show
Run PMax alongside a tightly structured Search campaign rather than instead of one. The Search campaign gives you the search term visibility PMax withholds, and the two together tell you far more than PMax alone.
When PMax genuinely wins
- E-commerce with a healthy product feed — this is what it was built for, and it is very good at it
- Accounts with strong, clean conversion data and enough volume to learn from
- Businesses that can produce decent creative, since assets are most of what you control
- Remarketing at scale, where it reaches inventory Search cannot
When to avoid it
- Fewer than 30 conversions a month
- Conversion tracking you have not personally verified
- Very small budgets — PMax needs room to learn before it performs
- Highly regulated categories where placement control matters
- B2B with long sales cycles, where the conversion signal is thin and delayed
The honest summary
PMax is a good campaign type wrapped in a sales pitch. Google's reps will push it into every account regardless of readiness, because it is easy to sell and it increases spend. In a well-tracked account with real creative and enough volume, it earns its place. In a poorly tracked one, it is the most efficient budget-burning machine in the platform.
Fix tracking first. Everything else follows from that.
Our free Google Ads audit checks conversion tracking end to end before recommending any campaign type. See how we run ads or request the audit.