Performance Max is either the best thing that happened to your Google Ads account or the fastest way to burn budget you’ve ever seen. Which one you get depends almost entirely on how you set it up and how well you feed the underlying AI. The campaign type itself is neutral — it’s the inputs that determine the outcome.
We’ve been running Performance Max across client accounts in the UK, US, Canada, and Australia for the past two years. The gap between campaigns that work and campaigns that don’t isn’t about budget or industry. It’s about how the algorithm is guided. This guide covers exactly how to do that.
What is Performance Max?
Performance Max (PMax) is a Google Ads campaign type that runs across every Google inventory surface — Search, Shopping, Display, YouTube, Gmail, Discover, and Maps — from a single campaign. It uses Google’s machine learning to decide which ad format to show, which audience to target, and when to show it, all in real time.
It replaced Smart Shopping and Local campaigns in 2022. For ecommerce advertisers especially, that transition was forced — you either moved to PMax or your campaigns stopped running. Understanding how to control it became non-optional.
The Core Problem With How Most People Run PMax
The number one mistake is treating Performance Max as a traditional campaign. Most advertisers set a budget, upload some assets, add a few audience signals, and wait. Then they check back in two weeks, see CPAs all over the place, and decide PMax doesn’t work.
What actually happened: the algorithm spent its first two weeks learning with incomplete information. If you didn’t give it enough audience signals, it cast too wide a net. If your conversion tracking was set up around the wrong events, it optimized toward the wrong behavior. If your asset quality was low, it defaulted to whatever format performed least badly.
Google’s own data shows that Performance Max campaigns drive an average of 18% more conversions at a similar or better cost per action compared to previous Smart Shopping and Local campaigns. But that’s an average across accounts that include some running it correctly and some not. The accounts dragging down that average are the ones treating PMax like a set-and-forget machine.
Performance Max Asset Groups: The Foundation You Can’t Skip
Asset groups are the building blocks of every PMax campaign. Each asset group contains your creative inputs — headlines, descriptions, images, videos, and the audience signals you want associated with them. The algorithm uses these to build ad combinations and test them across surfaces.
Most advertisers create one or two asset groups per campaign and call it done. The accounts that consistently outperform run five to eight asset groups, each aligned to a specific product category or audience intent. Here’s why that matters:
- A single asset group means the AI mixes messaging for different products and audiences, producing generic combinations that convert nobody well.
- Segmented asset groups let you write specific headlines for specific products, add audience signals that actually match those products, and measure which combinations work.
- When one asset group underperforms, you can identify it and fix it. With one combined group, you have no visibility into what’s failing.
Practically speaking: if you’re an ecommerce retailer selling clothing and accessories, run separate asset groups for each category at minimum. Better yet, separate by funnel stage — one group targeting new customers with brand awareness messaging, another retargeting past site visitors with specific product ads.
Audience Signals: How to Actually Guide the Algorithm
Audience signals are instructions to the algorithm, not hard targeting. You’re telling Google “start here” — it will expand beyond your signals if it finds better-performing users elsewhere. That’s both the power and the risk of PMax.
The signal hierarchy that consistently produces better results:
- Your own customer lists first. Upload your CRM — purchasers, high-LTV customers, email subscribers who convert. This is the highest-quality signal you can give. The algorithm uses it to find lookalikes that match your actual buyers, not a generic demographic profile.
- Website visitors with purchase intent signals. Add-to-cart events, checkout initiations, product page visitors (especially multiple visits). These tell the AI what pre-purchase behavior looks like for your specific product.
- In-market and custom intent audiences. These are Google’s pre-built segments based on search and browsing behavior. Layer them in as supplementary signals, not primary ones.
One thing that’s often overlooked: add audience signals to every asset group individually. If you add them only at the campaign level, the asset groups don’t inherit them in all configurations. Duplicate your signals across groups explicitly.
The Feed Quality Problem (For Ecommerce)
If you’re running PMax for a product-based business, your Google Merchant Center feed is doing more work than your ad copy. The AI matches products to search queries and audience signals partly based on what’s in your feed — product titles, descriptions, attributes, pricing.
A Google Ads management audit we ran across twelve ecommerce accounts found that feed quality issues were the primary cause of poor PMax performance in nine of them. The most common problems:
- Product titles missing key attributes (color, size, material) that match how people search.
- Generic descriptions that duplicate the title instead of adding context.
- Missing GTIN or MPN data, which reduces how often products surface in Shopping placements.
- Stale pricing or out-of-stock products left active in the feed.
The fix is a feed audit before you launch or optimize a PMax campaign. Prioritize your top-selling products first — clean those titles and descriptions to match actual search intent, not just internal product naming conventions.
Conversion Tracking: If This Is Wrong, Everything Else Is Irrelevant
Performance Max optimizes toward whatever conversion actions you’ve told it to value. If those actions are wrong, the optimization is wrong. This sounds obvious, but we see misconfigured conversion tracking in a significant portion of accounts we audit.
The common tracking mistakes that break PMax performance:
- Counting micro-conversions as primary goals. If “newsletter signup” or “page view” is set as a primary conversion action, PMax will optimize toward those — not purchases. Set macro-conversions (purchases, qualified leads) as primary. Set micro-conversions as secondary or informational.
- Duplicate conversion counting. If both Google Ads tracking and a GA4 import are counting the same purchase, your conversion numbers are inflated and your CPAs look better than they are. The algorithm optimizes toward these inflated numbers and overbids.
- Missing enhanced conversions. Enhanced conversions pass hashed first-party data back to Google, improving attribution accuracy — especially for conversions that happen after ad clicks are cleared from cookies. This is table stakes in 2026. If you haven’t set this up, you’re flying blind on a chunk of your actual conversion data.
Get conversion tracking right before touching anything else. Everything downstream depends on it.
Budget and Bidding Strategy: What to Expect in the Learning Phase
PMax needs data to learn. The minimum threshold Google recommends is 50 conversions in 30 days before automated bidding becomes reliable. If your account generates fewer than that, PMax will spend more time guessing and less time optimizing.
For accounts under that threshold, two approaches that work:
- Start with a Maximize Conversions bidding strategy (no target CPA or ROAS) for the first 4-6 weeks. This tells the algorithm to pursue volume, which builds the data it needs faster.
- Once you hit the 50-conversion mark, switch to Target CPA or Target ROAS, setting your initial target at or slightly above your current average. Setting an aggressive target too early starves the campaign of conversions because the AI becomes overly selective.
On budget: don’t set a daily budget so low that the campaign can’t exit the learning phase. A campaign running at $30/day with a $40 target CPA has almost no room to accumulate data. A rough heuristic — your daily budget should be at least 3-5x your target CPA to allow for normal conversion timing fluctuations.
For specific guidance on where Performance Max fits in a full Google Ads budget allocation, our Google Ads budgeting guide breaks down how to split spend across campaign types at different investment levels.
What PMax Won’t Do (And What You Need to Do Instead)
Performance Max gives you almost no keyword control at the campaign level. It runs on audience and intent signals, not explicit keywords. For most campaigns that’s fine — but there’s a specific problem: brand traffic.
PMax will happily bid on your own brand name and report those conversions as PMax-driven. That inflates performance metrics without representing any actual incremental gain. The fix is simple but often missed: create a separate brand Search campaign and use campaign-level negative keywords (available through a Google rep or the exclusion list tool) to prevent PMax from eating brand traffic.
Similarly, if you have high-performing, tightly controlled non-brand Search campaigns you want to protect, run them in parallel and monitor search term overlap. There’s no perfect solution to cannibalization — Google doesn’t give you full transparency — but comparing performance before and after PMax launch gives you a signal.
How to Read Performance Max Reports
PMax reporting is deliberately limited. You can’t see keywords, you can’t see placements by default, and you can’t see individual auction data. What you can see:
- Asset group performance ratings — “Best”, “Good”, “Low”. Replace “Low” rated assets. The algorithm won’t stop using them, but it will favor better-performing ones when you add them.
- Search category insights — found in the Insights tab, this shows categories of search queries your campaign matched. Not the exact queries, but directionally useful for spotting misalignment.
- Audience insights — shows which in-market and affinity segments are over-represented in your conversions. Use this to add or adjust audience signals.
- Channel performance breakdown — how spend and conversions distribute across Search, Shopping, Display, YouTube, etc. If YouTube is eating 40% of budget with 5% of conversions, that’s a signal to adjust asset weighting or add YouTube exclusions.
Check the Insights tab weekly, not just campaign-level metrics. The signals there inform your asset group adjustments, audience additions, and budget decisions in ways the top-line numbers don’t.
Performance Max vs Standard Shopping: When to Use Each
Standard Shopping campaigns still exist and still make sense in some situations. PMax isn’t the right answer for every account. Here’s the honest breakdown:
- Use PMax when you have enough conversion data (50+/month), a solid feed, good audience lists, and you want scale across channels beyond Shopping.
- Use Standard Shopping when you need granular control over bids and placements, your account is new with little conversion history, or you’re selling products where Shopping placements drive almost all your volume.
- Run both when you want PMax for scale and a Standard Shopping campaign for your highest-margin or highest-conversion products where you don’t want to lose keyword-level control.
Our Google Ads optimization guide covers the specific campaign structure decisions in more depth, including how to sequence PMax alongside existing Search campaigns without cannibalizing performance.
Frequently Asked Questions About Performance Max Strategy
How long does the Performance Max learning phase take?
Performance Max typically needs 4 to 6 weeks to exit the learning phase and stabilize performance. During this period, avoid making major budget or bidding changes — frequent adjustments reset the learning clock. The campaign needs at least 50 conversions in 30 days for automated bidding to work reliably.
Can you add negative keywords to Performance Max campaigns?
Yes, but the process is limited. You can add account-level negative keyword lists that apply to PMax, and you can request campaign-level negatives through your Google rep or the campaign exclusion tool. You cannot add negative keywords directly at the campaign level through the standard interface the way you can in Search campaigns.
How many asset groups should a Performance Max campaign have?
Most accounts perform better with 5 to 8 asset groups per campaign, each aligned to a specific product category or audience intent. One or two asset groups is the most common setup but tends to produce generic ad combinations that convert poorly. Segment by product type, audience stage, or offer — whichever distinction is most meaningful for your business.
Does Performance Max replace Search campaigns?
No. PMax and Search campaigns serve different purposes. Search campaigns give you keyword-level control and transparency. PMax gives you scale across all Google inventory using AI targeting. Most accounts benefit from running both — PMax for broad reach and full-funnel coverage, Search for high-intent branded and non-branded keywords where you want tighter control.
Why is my Performance Max campaign spending on Display and YouTube instead of Search?
PMax allocates budget across channels based on where it predicts conversions will come from. If it’s spending heavily on Display or YouTube, it either lacks enough conversion data to confidently target Search, or the Search inventory for your targeting is too narrow to absorb your budget. Check that your conversion tracking is correct, ensure your audience signals include purchase-intent visitors, and consider running a parallel Search campaign to capture the high-intent traffic directly.
The Bottom Line on Performance Max in 2026
Performance Max works when you work it. The accounts we see failing with it are the ones that uploaded three headlines, set a budget, and expected the algorithm to figure out the rest. The ones winning have clean conversion tracking, segmented asset groups built around specific audiences, customer lists uploaded as primary signals, and optimized product feeds updated on a regular cadence.
None of that is technically complicated. It’s just operational discipline — treating the AI as a powerful tool that needs good inputs rather than a magic solution that replaces strategy.
If you want a specific read on where your current Performance Max setup is leaving performance on the table, book a Google Ads audit with Advertizingly. We’ll go through your campaign structure, asset quality, conversion setup, and bidding configuration and give you a prioritized list of what to fix first.
For more on building a full paid search strategy around Performance Max, see our Google Ads management services page and our guide on performance marketing strategy.