Performance Max: what it is and how to tell if it works

Your Performance Max campaign shows a $44 CPA, the best in the account. Three months later, total revenue is roughly flat and spend is up $20,000 a month. Where did the result on the screen go?
This happens all the time in e-commerce and lead generation accounts. Not because Performance Max is bad, but because it is very good at one specific thing: finding conversions that were going to happen anyway and putting its name on them. Understanding what Performance Max is and how it makes decisions is the first step to knowing whether it is adding results or just moving them around.
Performance Max (PMax) is a Google Ads campaign type that uses a single setup to advertise across every Google channel at once: Search, Shopping, YouTube, Display, Discover, Gmail and Maps. You provide conversion goals, budget, creative and audience signals; the algorithm decides where, when and to whom each ad is shown, adjusting bids in real time.
What is Performance Max in Google Ads, really?
Think of PMax as an automated media buyer that gets a target (CPA or ROAS) and a budget, and is free to spend it on any Google inventory. You don’t pick keywords or placements. You supply raw material and the system builds the distribution.
That raw material has four parts:
- Conversion goals: what counts as a result (purchase, lead, call). The most important input, and the most neglected.
- Asset groups: headlines, descriptions, images, videos and, for retailers, the Merchant Center feed.
- Audience signals and search themes: hints about who your customer is. They are suggestions, not restrictions.
- Budget and bid strategy: maximize conversions or value, with or without a target CPA/ROAS.
The scale is big. According to Google, Performance Max was used by over one million advertisers as of April 2025, and the 90-plus improvements shipped in 2024 lifted conversions and conversion value by more than 10% for advertisers using it (Google Ads, April 2025). Keep in mind this is Google’s internal data, measured with Google’s own yardstick.
Why is Performance Max called a black box?
Because for a long time it reported the consolidated result and hid the path. You saw conversions and cost, but not how much went to YouTube, how much to branded Search, or which queries triggered the ad. No media buyer enjoys defending a number they can’t break down.
That has improved a lot. In April 2025, Google announced channel performance reporting, a search terms report with the same granularity as Search campaigns, and campaign-level negative keywords (Google Ads, 2025). August 2025 brought negative keyword lists shared across campaigns, a search theme limit raised from 25 to 50 per asset group, plus age and device exclusions (Search Engine Roundtable, August 2025).
So the box now has windows. But visibility is not causality. Knowing that 30% of spend went to Search doesn’t tell you whether those sales would have happened without the campaign.
How to tell if Performance Max is actually working
The right question is not “what is the PMax CPA?” It is “how much did the whole account gain because of PMax?” Those are different questions, and the gap is usually large.
1. Look at the whole account, not the campaign
Compare total conversions and revenue for the account (or for the business, in your CRM or order system) before and after launching PMax, against total spend. If PMax reports 450 conversions and the whole account only grew by 210, the other 240 came from somewhere: other campaigns that lost ground.
2. Calculate the incremental CPA
A worked example (a simulation, not research). A store was spending $33,000 a month on Search: $30,000 on generic terms (300 conversions) and $3,000 on brand terms (400 conversions). Total: 700 conversions.
It launches PMax with $20,000. The following month PMax reports 450 conversions at a $44.44 CPA. But branded Search drops to 200 conversions (PMax started bidding on the brand name) and generic Search drops to 260.
| Metric | Before | After |
|---|---|---|
| Total spend | $33,000 | $53,000 |
| Account conversions | 700 | 910 |
| Conversions reported by PMax | 0 | 450 |
| PMax CPA (on screen) | n/a | $44.44 |
| Incremental CPA ($20,000 / 210) | n/a | $95.24 |
The real PMax CPA is more than double what the dashboard shows. It may still be worth it, depending on your margin. But the decision changes completely once you look at the right number. To set that ceiling, revisit your break-even ROAS.
3. Open the channel report and the search terms report
In channel performance, check how much spend went to Search and Shopping versus YouTube and Display. Then, in the search terms report, filter for queries containing your brand name. If a meaningful share of conversions comes from branded searches, PMax is harvesting demand that already existed.
4. Split new customers from returning ones
Cheap remarketing inflates any campaign. If PMax mostly converts people who already bought or already had items in the cart, it is doing the job of an email. Turn on the new customer acquisition goal, or match orders against your customer base to measure the share.
5. Run a holdout test when you can
If you have the volume, the most reliable approach is to pause PMax in some regions (or for a few weeks, watching for seasonality) and compare with regions where it keeps running. It isn’t perfect, but it is the most honest way to measure incrementality outside the dashboard. Our post on conversion attribution explains why every platform tends to claim credit for itself.
When to use Performance Max (and when not to)
In practice, PMax works best when the algorithm has enough signal to learn and you have an external yardstick to check what it tells you.
| Scenario | Worth testing PMax? | Why |
|---|---|---|
| E-commerce with a clean feed and dozens of conversions per week | Yes | Shopping and dynamic remarketing are the campaign’s strong suit |
| Local business with call or directions goals | Yes, carefully | Maps is part of the inventory, but check call quality |
| B2B lead gen where the conversion is “form submitted” | Risky | The algorithm optimizes for cheap forms, not qualified leads |
| New account with few conversions per month | Not yet | Thin signal slows learning and scatters spend |
| Strong brand with lots of name searches | Only with brand exclusions | Without them, PMax cannibalizes branded Search |
The B2B case deserves a warning. If the configured conversion is “form submitted,” PMax will find the people most likely to fill out forms, who are not always the ones most likely to buy. The fix is importing offline conversions from your CRM (qualified lead, closed deal) and optimizing for those. Without that, CPL drops and sales don’t follow.
The most common PMax setup mistakes
- Polluted conversion goals: a WhatsApp click, a page view and a purchase all count the same. The algorithm optimizes for the easiest event.
- No brand exclusions: the campaign takes over searches for your name and looks brilliant.
- Thin creative: without a video, Google may auto-generate one from your images, and the result rarely represents the brand well.
- Tweaking every week: each big change partially resets learning. Set an evaluation window (four to six weeks) before judging.
- Judging only by Google’s dashboard: Google is both referee and interested party.
If you already have an account review routine, the right order for analyzing Google Ads campaigns helps you fit PMax into it instead of treating it as an exception.
Frequently asked questions
Is Performance Max worth it?
It is when you have well-defined conversions, enough volume for the algorithm to learn, and a way to measure results outside Google’s dashboard. For e-commerce with a well-structured feed, it usually pays off. For small accounts or B2B leads without offline conversions, the risk of paying for results you would have gotten anyway is high.
What is the difference between Performance Max and a Search campaign?
In Search you choose keywords and only show on search results. In PMax you don’t choose keywords: Google uses your assets and signals to advertise across all channels, including search. That is why the two can end up competing for the same queries.
How long does Performance Max take to show results?
Google talks about a learning period of a few weeks. In practice, evaluate with at least four to six weeks of stable data and no major changes midway, compared with the same period before launch.
Can you add negative keywords to Performance Max?
Yes. Since 2025, Google allows campaign-level negatives and negative keyword lists shared across campaigns, on top of brand exclusions. Use them to block your brand terms, “free” searches, or any intent that doesn’t convert for you.
What formats does Performance Max use?
Text (headlines and descriptions), square, landscape and portrait images, videos, a logo and, for retailers, the product feed. The more complete the asset group, the more inventory the campaign can compete for with good ads.
PMax isn’t the problem; the yardstick is
Performance Max delivers what it promises: scale with little setup. What it doesn’t deliver is an impartial verdict on its own performance. The people who make good decisions with PMax cross-check Google’s dashboard with real revenue, the CRM and the account’s other campaigns, and always ask how much was added, not how much was reported.
That cross-check is exactly the tedious work almost nobody does every week. With Sherlok, you connect Google Ads, GA4 and your CRM and ask in plain language, for example, “how many new sales has Performance Max brought in since we turned it on?”, and get the analysis back with numbers for the whole account.
Your AI agents already know what to do in your account.
Turn them on free, no credit card.
- No credit card
- Read-only access
- Revoke it any time