Google Performance Max in 2026: What We Actually Turn Off

Google Performance Max has been the most-discussed — and most-misunderstood — campaign type in paid media for the past two years. At Choco Media, we run it across e-commerce, lead generation, and local service accounts. Our honest take: it works, but not on its own terms. The version that works is the one you’ve actively shaped by knowing exactly which settings to override, which signals to feed it, and which defaults to treat as mistakes waiting to happen.

This post is for media buyers, in-house marketers, and brand owners who have launched Performance Max campaigns and watched them do something other than what they expected. We’ll walk through the specific controls — asset group structure, audience signals, placement exclusions, and brand safety settings — that consistently make the difference between a campaign that burns budget and one that actually drives efficient conversions. If you’re evaluating Google as a channel for the first time, you may want to read our overview of Meta vs Google Ads first to frame where PMax fits.

Everything here is based on what we observe in real accounts. We won’t claim to know what’s in Google’s black box — but we can tell you what we turn off, what we turn on, and why it matters.

What Google Performance Max Actually Is (And Isn’t)

Performance Max is Google’s fully automated, all-inventory campaign type. A single campaign can show ads across Search, Shopping, Display, YouTube, Gmail, Discover, and Maps simultaneously. Google’s algorithms allocate budget across channels in real time based on conversion signals you provide.

The promise is elegant: one campaign, all of Google, optimised automatically. The reality is more complicated.

• PMax requires a large volume of conversion data to learn effectively — typically 50+ conversions per month at the account level before the algorithm stabilises.
• Without strong audience signals and clean asset groups, it defaults to the cheapest inventory, which is usually Display — and Display rarely drives the same quality of conversions as Search or Shopping.
• The campaign has very limited transparency. You can see asset performance labels, but you can’t see which specific placements, queries, or audiences are driving results within the campaign.

Understanding this is the starting point. PMax isn’t a set-it-and-forget-it solution. It’s a semi-automated system that produces better results the more intentional structure and signal you provide.

The Settings We Turn Off First

Automatically Created Assets

By default, Google will generate headlines, descriptions, and even images by pulling content from your website and combining it algorithmically. This feature, called “Automatically Created Assets” (ACA), sounds helpful and is often catastrophic for brands with any degree of voice or positioning specificity.

We turn it off in every account. Here’s why: Google’s generated copy tends to be generic, often pulls outdated text from pages you haven’t touched in two years, and has no understanding of your current messaging priorities. We’ve seen campaigns where Google generated headlines that directly contradicted a client’s current positioning after a rebrand.

• To disable: Campaign settings → Automatically created assets → turn off both Text and Image ACA.
• Replace with: Upload at least 5 headlines, 5 descriptions, 5 landscape images, 5 square images, and at least one 16:9 video per asset group.

URL Expansion

URL Expansion lets Google send traffic to any page on your website it determines to be relevant, rather than the landing pages you specify in your asset groups. This bypasses your conversion-optimised pages and can send paid traffic to blog posts, contact pages, or even the homepage when you’ve built a product-specific funnel.

We disable URL Expansion unless the account is specifically structured as a site-wide traffic driver. For e-commerce, keep it off and use well-structured asset groups that map to specific product categories. For lead generation, keep it off entirely — every conversion path needs to go through your tested landing page.

• To disable: Campaign settings → URL expansion → “Off” (or restrict to specific URLs in the URL contains filter).

Display Network for Awareness-Only

You can’t exclude Display entirely from PMax — that’s a known limitation of the campaign type. What you can do is weight against it by ensuring your conversion goals are tightly defined. If your primary conversion is a form fill or purchase, Google will gradually deprioritise Display inventory because it converts less efficiently against that goal. Feed it good conversion data and it self-corrects over time.

The single most effective thing you can do for a PMax campaign isn’t found in campaign settings. It’s making sure your conversion tracking is accurate, deduplicated, and not inflated by micro-conversions that don’t correlate with actual business value. Garbage in, garbage out — the algorithm will optimise for whatever you tell it matters.

Asset Groups: Structure Before Volume

Most PMax campaigns are set up with a single asset group covering every product or service the account sells. This is the most common structural mistake we see, and it has a direct impact on relevance and performance.

How We Structure Asset Groups

Each asset group should represent a coherent theme — one product category, one service line, one audience segment. The assets (images, headlines, descriptions, videos) within that group should all be contextually relevant to each other.

• E-commerce: One asset group per major product category (e.g., women’s outerwear, men’s footwear, accessories). Each group uses category-specific copy, product imagery, and landing pages.
• Lead generation: One asset group per service or offer type. If you offer three distinct services, build three asset groups with distinct USPs, proof points, and landing pages for each.
• Local: One asset group per geographic cluster if you serve multiple distinct areas, especially if the offer or pricing differs by location.

The practical benefit is twofold: Google’s algorithm has cleaner signal about what each group should achieve, and you can read asset-level performance labels (Low / Good / Best) to understand which creative elements are working.

Asset Quality Thresholds

PMax labels your asset group strength as “Poor,” “Good,” or “Excellent” based on asset variety and count. Aim for “Excellent” on launch, which typically requires:

• 15 headlines (at least 3 that include the target keyword)
• 5 descriptions
• 20 images (mix of square, landscape, and portrait formats)
• At least 1 video (we upload a 15-30 second cut even if it’s a simple brand slideshow — this prevents Google from auto-generating a video from your images, which it will do if you don’t provide one)

Audience Signals: Feed the Algorithm What It Needs

Audience signals in PMax are not targeting in the traditional sense. You’re not restricting who sees the ads — you’re giving Google a starting point for who to look for. It will expand beyond your signals if it finds better-performing audiences. The quality of your signals determines how quickly the campaign reaches a profitable state.

The Signals We Always Add

• Customer Match lists: Upload your existing customer email list (with consent). This is the highest-quality signal you can provide — you’re telling the algorithm “find more people like these.”
• Website visitors: Add your remarketing audiences, segmented by engagement level if possible (all visitors vs. cart abandoners vs. past purchasers).
• Custom intent segments: Build custom segments based on the specific search queries your ideal customer would use. These are more precise than broad interest categories.
• In-market audiences: Add the most relevant in-market segments as a secondary layer. They’re less precise but give Google additional directional context during the learning phase.

What We Don’t Add

Broad demographic targets, generic interest categories, and life events audiences tend to dilute the signal quality rather than improve it. If your customer file and website visitor data are strong, they’re the foundation. Everything else is supplementary.

For accounts without a substantial customer list (fewer than 1,000 emails), we lean harder on custom intent search term segments and combine PMax with a standard Search campaign running in parallel to build conversion history before PMax takes the majority of budget. This pairs well with the principles in our paid media account audit framework.

Brand Exclusions: A Non-Negotiable

Performance Max will bid on your brand terms unless you explicitly tell it not to. This means budget intended for prospecting new customers ends up going to users who were already searching for you by name — conversions you would have captured anyway through a low-cost branded search campaign or organic results.

How to Set Brand Exclusions

As of 2025, brand exclusions in PMax are managed at the account level through the “Brand lists” feature in Google Ads settings (not within the campaign itself). This is counterintuitive — you have to build the list in account settings and then apply it to the campaign.

• Navigate to: Tools → Shared library → Brand lists → Create a new brand list with your brand name and common misspellings.
• Apply the exclusion list: Within the PMax campaign settings, find “Brand exclusions” and add the list you created.
• Run a parallel branded Search campaign: Keep a separate, tightly controlled branded campaign running to capture brand demand efficiently and maintain visibility data for that traffic segment.

We treat brand exclusions as a day-one setup step, not something to add after the campaign has been running. Every week without the exclusion in place is budget that went to brand queries rather than prospecting.

Placement Exclusions: What You Can and Can’t Control

Placement exclusions in PMax are more limited than in standard Display or Video campaigns. You can’t build a full placement exclusion list at the campaign level the same way you would in a traditional Display campaign. What you can do:

• Content exclusions: At the campaign level, you can exclude sensitive content categories (tragedy and conflict, sexually suggestive content, etc.). Apply these as a baseline.
• Account-level placement exclusions: Managed Placements exclusions set at the account level apply to PMax as well. If you identify specific domains or apps that are consistently delivering low-quality traffic, add them here.
• Video inventory exclusions: Within PMax video asset settings, you can set “Inventory type” to “Standard” or “Limited” to reduce exposure on less brand-safe YouTube content.

In practice, we set account-level exclusions for mobile app traffic (app categories: Games) unless the client specifically sells apps or games. Mobile app Display placements are a consistent source of accidental clicks and inflated impression volume without meaningful conversion contribution.

Conversion Goals: The Highest-Leverage Setting

Performance Max is only as good as the conversion actions you’re asking it to optimise toward. This is where most underperforming campaigns have their root cause — not in creative, not in audience signals, but in what they’ve told Google to count as success.

Common Mistakes

• Optimising toward micro-conversions: If your account is optimising toward “Page views over 60 seconds” or “PDF downloads” because that’s the only conversion with enough volume, you’re optimising for engagement, not revenue. The algorithm will send you a lot of that behaviour regardless of whether it correlates with actual customers.
• Including all goals equally: Google allows you to weight conversion actions by value. A purchase should be weighted much higher than a newsletter signup. If you’re treating them equally, budget shifts toward whichever has lower friction, which is usually the lower-value action.
• Conversion windows too short: If your sales cycle is 30–60 days, a 7-day conversion window means the algorithm is operating on incomplete data. Match the window to your actual decision timeline.

What We Set

For e-commerce: purchase as the primary conversion, with cart additions as a secondary lower-weighted goal. Revenue-based bidding (target ROAS) once the account has at least 30–50 purchases per month from the campaign. Below that threshold, we use “Maximise conversion value” without a target until there’s enough data for a stable ROAS target.

For lead generation: form fill or phone call as the only primary conversion. We disable micro-conversion goals from PMax specifically, even if they’re useful elsewhere in the account for audience building. You can find more detail on our attribution approach in our post on GA4 and server-side tracking.

Monitoring Without Full Transparency

The lack of search term transparency in PMax is a genuine frustration. You can see search themes (added in 2024) but not individual queries. You can see asset performance labels but not which placements or audiences drove which results. Here’s how we work around it:

• Insights tab: Check it weekly. It surfaces audience segments, search themes, and product/asset performance signals that aren’t visible elsewhere.
• Asset group reports: Review “Best” and “Low” labels monthly. Rotate out “Low” assets — don’t just add more on top of underperforming ones.
• Cross-campaign search term data: Run a Search campaign in parallel with broad match keywords and look at the search terms report there. It gives a proxy view of the query types PMax is likely capturing.
• Segmentation by channel: In the campaign performance report, segment by network (Search, Display, YouTube, etc.) to understand where budget is going. If Display is consuming 60% of budget with 10% of conversions, that’s a signal to adjust conversion goal weighting or investigate creative quality.

We review PMax campaigns more frequently than standard campaigns during the first 30–45 days — three times per week rather than weekly. The learning phase is where most bad habits form, and early corrections have an outsized impact on long-term performance.

When to Use PMax and When to Reach for Standard Campaigns

Performance Max is not the right first campaign type for every account. Our rough framework:

• Good fit: E-commerce accounts with product feeds, at least 30 conversions/month, a customer email list, and the creative assets to populate multiple asset groups properly.
• Good fit with support: Lead generation accounts with strong conversion tracking, a defined audience, and a willingness to run a parallel Search campaign during the learning phase.
• Not a good fit yet: New accounts with limited conversion history, businesses with very specific targeting requirements (narrow geographic radius, specific job titles, niche B2B intent), or accounts where brand safety is a hard constraint that can’t be managed through the limited exclusion options.

For accounts in the “not a good fit yet” category, we typically start with Standard Shopping (for e-commerce) or a tightly structured Search campaign (for services), build conversion history over 60–90 days, and then layer in a PMax campaign once the algorithm has a signal foundation to build on.

If you’re working with a paid media partner and wondering whether your PMax setup reflects these principles, our contact page is the place to start — we audit accounts regularly and the findings are often immediately actionable.