Is Performance Max worth it? A straight answer, with conditions

The short answer: worth it under four conditions

Is Performance Max worth it? It is, if four things are true: your conversion tracking is accurate, you get enough conversions each month for the model to learn from, you have genuine creative assets, and, if you sell products, a product feed worth showing. Miss any of those and Performance Max becomes an expensive way to find out.

That is the verdict. The rest of this post is the working.

PMax is not a good or a bad campaign type. It is an amplifier. It takes whatever your account currently believes about a good customer and buys more of that across every Google surface at once. If the account believes something wrong, you get more of the wrong thing, faster and at scale.

So the useful question is not whether Performance Max works. It is whether your account is in a state where automation has anything worth learning from. Most of the disappointing PMax stories we hear come from accounts that were not ready, running a campaign type that assumed they were.

When Performance Max earns its budget

Four conditions, in the order they matter.

Your conversion tracking is trustworthy

This is the whole ball game. PMax bids toward whatever conversions you report back to it. So conversion tracking that fires reliably, does not double count, and carries a value where you have one is the difference between a machine that learns your business and a machine that learns your measurement errors. If your thank-you page fires twice, PMax will patiently find you more of whatever caused that.

You have enough conversion volume to feed the model

Automated bidding needs a steady run of conversions to work out what a buyer looks like. Dozens a month, not a handful. This bites harder in Australia than in larger markets: a service business in one capital city can be doing everything right and still generate too few conversions a week for the model to separate a real pattern from a coincidence.

Your creative assets are actually good

PMax builds ads by mixing your headlines, descriptions, images, logos and video across surfaces you cannot see individually. Give it three headlines and one stock photo and it will assemble ads from three headlines and one stock photo, everywhere Google sells space. Give it several real angles, proper photography and at least one decent video and you have handed it something worth amplifying. Missing assets get auto-generated, which is rarely how you would describe your own business.

You have a real product feed (for stores)

For e-commerce, the feed does the job keywords used to do. Titles written the way people search, complete attributes, accurate pricing and stock, all mapped properly to Google’s product data specification. A store with a clean feed and honest conversion values is the case where PMax is genuinely hard to beat, and it is the centerpiece of most Performance Max campaign builds we run for retailers.

There is one more case worth naming: accounts that have run out of keyword list. If your Search campaigns have been mature for years and impression share is already high on the terms you know about, PMax reaches demand a keyword list could never describe. How that matching works is covered in does Performance Max use keywords.

When Performance Max is not worth it

This is the part most guides skip. Some accounts should not be running Performance Max this quarter, and spending less is the correct answer for them.

• Your tracking is shaky. Not "mostly working". If you cannot say with a straight face that every conversion in the account is real, deduplicated and worth roughly what it claims, fix that before handing the budget to a model that optimizes toward it.
• You get a handful of conversions a month. With thin volume the model never leaves the guessing stage, and the weekly numbers swing enough to justify whatever you already believed. A tight Search campaign gives you control and readable data while volume builds.
• You cannot separate brand from generic. More on this in the next section. It is the single most common reason a PMax campaign looks brilliant and changes nothing.
• You need control for a real strategic reason. Wildly different margins across products, stock you cannot afford to sell out of, a line you are deliberately not promoting, a regulated category where placement matters. PMax gives you influence, not control, and sometimes control is the actual requirement.
• Your budget cannot fund a learning period. PMax spends across a lot of inventory. Split a small budget over every Google surface and you buy a thin layer of nothing in a lot of places. Work out what one result is worth to your margin before you decide the campaign type.
• Lead quality varies and you cannot pass that back. Optimized toward raw form fills, PMax will find you an impressive number of raw form fills. Lead generation accounts need qualified-lead data flowing back before automation is an upgrade rather than a volume machine.

If tracking is broken, pause the victory lap and fix measurement first. Everything Performance Max does well, it does by copying what your conversion data tells it.

The brand traffic problem that flatters every report

Here is the thing that makes a mediocre Performance Max campaign look like a triumph. Unless you tell it otherwise, PMax can serve on searches for your own business name, and it counts those conversions as its own. Those are people who already knew you, typed your name into Google, and were arriving anyway. Very sporting of it.

The result is a campaign return on ad spend that looks excellent while total revenue has not moved. You have not bought growth. You have bought a receipt for traffic you already had, and you are now paying a click fee on your own brand.

Two fixes, both worth doing before launch:

• Turn on brand exclusions so PMax cannot serve on your business name and its close variants.
• Run a separate branded Search campaign so brand traffic is visible, cheap and reported on its own line, where you can see exactly what it costs and what it returns.

Then judge PMax on what it added, not on what it collected. Compare account totals and business revenue before and after, and be suspicious of any campaign whose numbers improve while the business total stays flat. Where your impressions land deserves the same skepticism: low-quality placements are the other quiet leak, and Performance Max placement exclusions are how you stop paying for them.

A return on ad spend figure that includes your own brand searches is not a performance number. It is a count of the customers you already had.

Performance Max vs Search campaigns

This gets framed as a fight, which is the wrong framing. They do different jobs.

• Search campaigns capture demand that already exists. You choose the keywords, you read the search terms, you match the message to a specific intent. When someone types "emergency electrician Adelaide" at 9pm, that is not a job for a discovery machine.
• Performance Max finds demand across surfaces you would not have thought to buy, using signals no keyword list can express. Broader reach, coarser reporting, less control over the details.

Most healthy accounts run both. Search takes the high-intent terms and the brand, PMax covers the ground beyond the keyword list and gets fed clean conversion data from the rest of the account. The order matters for smaller advertisers: start with Search, build conversion history, then add PMax once there is enough signal to make automation a decision rather than a hope.

If what you want is broader matching inside Search rather than one campaign buying everything, that is a different feature entirely. Google Ads AI Max sits inside your existing Search campaigns and keeps most of the reporting you are used to.

How to decide in your own account

1. 1Audit tracking first. Every conversion action, its value, its deduplication, checked against your CRM or store rather than trusted because Google Ads displays a number.
2. 2Count your monthly conversions. Dozens means you have signal. A handful means build volume in Search first and revisit in a quarter.
3. 3Set the guardrails before launch. Brand exclusions on, placement exclusions applied, a separate branded Search campaign live, and full asset coverage including video.
4. 4Fund a real test in AUD. Enough budget to buy a readable number of conversions each week for at least six weeks, because the first fortnight is learning, not performance.
5. 5Judge on business outcome. Cost per qualified enquiry, or profit after cost of goods, measured at account and business level. Campaign-level return on ad spend is where PMax is at its most flattering.
6. 6Decide once, calmly. If it works, scale slowly. If it does not, the cause is often the offer, the feed or the landing page rather than the campaign type.

That sequence is what our Google Ads management works through before recommending Performance Max to anybody, and reasonably often the recommendation is "not yet". PMax is a multiplier, and a multiplier is only worth having once there is something in the account worth multiplying.

Frequently asked questions