Google Performance Max is the most misunderstood campaign type in DTC advertising. Most brands running it are either watching their ROAS disappoint and blaming Google, or seeing numbers that look good while missing the fact that the budget is cycling through their own brand terms and existing customers. Neither version is working. The campaign is just hiding what it is doing badly enough that you keep funding it.
Google Performance Max replaced Smart Shopping in September 2022 and now accounts for the majority of ecommerce search spend across most DTC accounts. If you are spending on Google, you are spending on PMax. The question is whether you have set it up to acquire new customers at a profitable cost, or whether you are paying Google a premium to retarget people who were going to buy from you regardless.
What Performance Max Actually Does
Performance Max is a single campaign type that runs across all Google inventory simultaneously: Google Search, Google Shopping, Display Network, YouTube, Gmail, and Discover. You supply the inputs: creative assets (headlines, descriptions, images, videos), a product feed connected through Merchant Center, audience signals, and a conversion target. Google's algorithm handles placement, bidding, audience selection, and creative combination testing in real time.
The campaign is designed to maximise conversion value at your target return on ad spend. In practice, that means the algorithm finds the path of least resistance to conversions. Without guardrails, the easiest conversions on Google are people who already know your brand and are actively searching for it. They convert at high rates, cost per conversion looks low, and your ROAS dashboard looks healthy. None of it represents genuine demand generation.
The guardrails are not complicated. They require deliberate setup and a degree of patience during the learning phase that most operators do not give the campaign. That is the entire problem in most DTC accounts.
The Numbers: What Good Looks Like in 2026
Performance Max benchmarks vary significantly by category, average order value, and setup quality. These figures represent well-configured campaigns in DTC categories at the £500k to £5m revenue range.
ROAS
3.5x to 5.2x
Across drinks, beauty, wellness, and food DTC categories. Campaigns without audience signals or brand exclusions typically sit 30-40% below this range. The gap is entirely structural, not a category problem.
Learning phase
6 weeks
Google's official learning window for PMax, longer than any other campaign type. The algorithm targets 30 to 50 conversions per week to optimise reliably. Budget changes above 20% during this period reset the clock.
New customer rate
50-65%
What a well-configured PMax with audience signals and brand exclusions delivers in new vs returning customer split. Unconfigured campaigns routinely show 25-35% new customer rates, meaning the majority of spend is going to existing customers.
Feed quality impact
20-35% ROAS lift
The documented improvement from optimising product titles, descriptions, and category attributes in Merchant Center. Feed quality determines which search queries Google matches your Shopping ads to. Most DTC brands have Shopify default titles that leave significant search coverage on the table.
Real-world results confirm the benchmarks. A UK drinks brand spending £8,000 per month on Google rebuilt their PMax from scratch with audience signals from their Klaviyo buyer list and brand term exclusions. ROAS improved from 2.1x to 4.3x within six weeks, and new customer acquisition rate increased from 31% to 58%. A beauty brand with a poorly structured product feed updated their Merchant Center titles to include ingredient keywords and skin concern terms. Google Shopping impressions increased 140% within three weeks and cost per acquisition dropped from £34 to £22. A wellness supplement brand had been running PMax alongside a legacy brand campaign, creating internal auction competition. Consolidating and restructuring cut their average CPC by 18% and improved conversion rate by 24%.
The pattern across these examples is identical: the structural issues were fixable in a day. The results showed up within the 6-week learning window. The brands had been leaving this performance on the table for months.
How to Set Up Performance Max Correctly
The setup sequence matters as much as the individual steps. Configure everything before launch. Changes after the campaign goes live cost you learning-phase progress.
Fix your Merchant Center product feed first
Performance Max Shopping ads are only as good as the product data you feed them. Before creating the campaign, audit your Merchant Center: product titles should lead with the primary category keyword followed by brand name and key differentiators, not your internal product names. Descriptions should include the terms your customers use when searching in your category. High-resolution images with no text overlays consistently outperform lifestyle images in Shopping placements. If your feed is pulling directly from Shopify with default titles, you are leaving a significant share of search coverage unmapped.
Upload audience signals from your buyer list
Audience signals are the single highest-impact configuration input in Performance Max. They do not restrict who sees your ads. They give the algorithm a starting point for identifying similar profiles. Upload your customer email list from Klaviyo as a customer match audience. Upload your site visitor list and add-to-cart visitors as additional signals. Without signals, the algorithm spends the first 2 to 3 weeks of the learning phase testing broad, low-intent traffic. With signals, it has a clear pattern to model against from day one.
Exclude brand terms at the account level
This is the most commonly skipped step and the most expensive omission. Add your brand name, all common misspellings, and your core product names as negative keywords at the account level. This forces PMax to compete for non-branded demand. Your brand search terms should be captured by a separate, tightly controlled brand campaign with lower CPCs and a distinct budget. If you do not separate these, PMax will route budget into branded searches because they are easy conversions. Your blended ROAS will look acceptable while your incremental new customer acquisition quietly deteriorates.
Structure asset groups by product category, not by SKU
Each asset group should represent a distinct customer intent or product category with its own creative assets and audience signals. A DTC skincare brand might run one asset group for moisturisers, one for serums, and one for SPF. Running an asset group per SKU fragments conversion data, prevents the algorithm from learning at any useful rate, and creates a management overhead with no performance benefit. Consolidate to 2 to 4 asset groups maximum at launch. Expand only once each group is generating 30 or more conversions per week.
Set a target ROAS, not maximise conversion value
The default bidding strategy in PMax is maximise conversion value, which gives the algorithm complete discretion over bidding with no efficiency floor. For most DTC brands, this produces early performance that looks strong before CPAs climb as the budget scales. Set a target ROAS from launch that represents your minimum acceptable efficiency. This creates a ceiling on how aggressively the algorithm bids into low-quality placements. Start with a target 20% below your historical average to give the algorithm room to learn, then tighten it once the campaign is out of the learning phase.
Do not touch the campaign for six weeks
Budget adjustments above 20%, pausing and reactivating, adding or removing asset groups, and changing the bidding strategy all reset the learning phase. Set your budget at launch based on what you can sustain for six weeks. Plan your creative refresh cycle to avoid needing to make changes during learning. Review performance data weekly but make no structural changes until week 6. The instinct to optimise during learning is the primary reason most DTC PMax campaigns underperform.
The Four Mistakes Killing Your PMax Performance
These patterns appear in almost every DTC Google Ads account that is underperforming on PMax. Each one is fixable within a day.
Running PMax without brand exclusions
The algorithm is optimising for conversions, not for new customer acquisition. Your brand-name searches have the highest conversion rates on Google. Without exclusions, PMax routes an increasing share of budget there as it learns that branded clicks convert well. By month two of a typical unconfigured campaign, 35 to 50% of spend is on branded queries. You are paying Google Shopping CPC rates to intercept people who were already on their way to your site. Separate your brand terms. Run them in a controlled brand campaign with manual CPC bidding. Keep PMax focused exclusively on non-brand demand.
No audience signals uploaded
Performance Max without audience signals is an expensive experiment in broad-match inefficiency. The algorithm needs a model of who converts for you. Without it, it tests wide. Existing customer lists from Klaviyo, past purchaser segments, and high-intent site visitors all serve as anchors that focus early learning. Many DTC brands have not connected their Klaviyo customer data to Google Ads as a customer match audience. That connection alone typically improves early-phase performance by 20 to 30% and cuts the effective learning period from 6 weeks to 4.
Using Shopify default product titles in the feed
Shopify exports product titles in the format your team uses internally, which almost never matches the search terms your customers use on Google Shopping. A drinks brand selling a "Mango Passionfruit Daily Greens" product needs a title that leads with the search category: "Greens Powder Mango Passionfruit, 30 Servings." A beauty brand selling "Radiance Serum No.4" needs "Vitamin C Face Serum for Brightening, 30ml." The feed title is the primary signal Google uses to match your Shopping ads to search queries. Invest two hours optimising your top 20 products in Merchant Center and the impression coverage increase is immediate.
Running legacy campaigns alongside PMax in the same categories
If you have Shopping campaigns, Dynamic Search Ad campaigns, or Standard Shopping campaigns covering the same product categories as your PMax, they are competing against each other in Google auctions. This inflates your CPCs across all campaigns and splits conversion data between campaigns, slowing learning in both. The most common version is a brand that ran Standard Shopping before PMax launched and never turned it off. Audit your campaign structure. PMax should be the primary campaign for your product categories. Legacy campaigns for the same inventory should be paused or consolidated.
How to Read PMax Data Without Being Misled
Performance Max reporting is designed to show you what Google wants you to see. The top-line conversion numbers look clean. The channel breakdown is buried. Understanding what is actually driving performance requires navigating a few levels below the campaign dashboard.
The most important question to ask about your PMax campaign is not "what is my ROAS?" It is "what percentage of my conversions are from new customers, and what search terms is Google Shopping matching my ads to?" If you cannot answer both questions, you do not know whether your Google spend is working.
The new versus returning customer split is available in the asset group reporting under "Audience insights." If your new customer rate is below 50%, your brand exclusions are either missing or not comprehensive enough. Check the Merchant Center for the search terms your Shopping ads matched to. If you see branded terms appearing, they are getting through the exclusion list and should be added as negatives.
For a complete picture of Google's actual contribution to revenue, pair PMax data with a post-purchase survey asking customers how they found you. Platform-reported attribution from Google consistently over-credits last-click conversions that would have happened through organic search regardless. For how to build a multi-source attribution model that does not rely on Google's self-reporting, see our guide to DTC marketing attribution.
Asset group performance reporting shows you which creative combinations Google is serving most frequently. Unlike Meta's ASC, which gives granular creative-level data, PMax shows you asset-level signals rather than precise conversion attribution. The "Learning" status in asset quality scoring tells you whether individual assets are contributing or holding the group back. Replace low-quality assets that have been running for more than 4 weeks without improvement signals.
PMax and Meta: Running Both Correctly
The DTC brands scaling most efficiently in 2026 are running Google Performance Max and Meta Advantage+ Shopping Campaigns in parallel, not choosing between them. They serve structurally different demand.
Google captures intent-led demand: people actively searching for your product category. Meta creates and converts demand among audiences who match your buyer profile but are not yet searching. The two channels are additive, not competitive. Brands running both correctly with proper exclusions and audience signals typically see 15 to 25% improvement in blended new customer acquisition cost compared to running either in isolation.
The creative brief for PMax is different from Meta. Google Shopping creative lives or dies on the product feed and the first image. For Display and YouTube placements within PMax, the same VOC-led angles that work on Meta apply: lead with the customer's problem, use real product-in-use imagery, and put the key benefit in the first three seconds of any video asset. For a detailed framework on creative testing velocity, see how the top DTC brands structure their testing cadence across both channels.
Growth Audit
Want to Know If Your Google Spend Is Actually Driving New Customers?
I will audit your current Google Ads structure, identify whether your PMax is consuming brand budget it should not be, and give you a clear action plan to recover that spend within 30 days. No pitch, no agency retainer required.
Book Your AuditFrequently asked questions
What is Google Performance Max for ecommerce?
Google Performance Max is an AI-driven campaign type that replaced Smart Shopping in 2022. It runs across all Google inventory: Search, Shopping, Display, YouTube, Gmail, and Discover. You supply creative assets, a product feed, and a budget. Google's algorithm handles where, when, and to whom your ads appear. For ecommerce brands, PMax is now the default campaign type and accounts for the majority of Google Shopping spend.
Why is my Performance Max ROAS so low?
Low PMax ROAS almost always comes from one of three causes: missing audience signals (the algorithm defaults to broad, low-intent traffic), missing brand exclusions (30-40% of budget routes to existing customers searching your brand name), or a weak product feed (poor titles and descriptions mean Google matches your ads to irrelevant queries). Fix the feed, add audience signals from your existing buyer list, and exclude brand terms from PMax. These three changes typically move ROAS by 30-60% within 30 days.
Should I exclude brand terms from Performance Max?
Yes, always. Without brand exclusions, PMax will route a significant share of your budget into branded searches because these convert at high rates, which looks good to the algorithm but drives zero incremental value. Existing customers who would have bought anyway get credited as conversions. Add your brand name, misspellings, and product names as negative keywords at the account level to keep PMax focused on net-new demand.
How many asset groups should a DTC brand run in Performance Max?
Start with one to three asset groups, each mapped to a distinct product category or customer intent. Running too many asset groups too early fragments the data and slows learning. A supplement brand might run one asset group for bestsellers, one for bundles, and one for new customer acquisition. Once each group is generating consistent conversion volume, you can expand. Do not create asset groups for every individual SKU.
Can Performance Max work alongside Meta Advantage+ Shopping Campaigns?
Yes, and the two campaigns serve different roles. Meta ASC operates within Meta's ecosystem and is strongest for awareness and mid-funnel conversion among interest-based audiences. Google PMax captures existing demand: people actively searching for your product category or brand. The two are not in competition. Brands running both correctly typically see a 15-25% improvement in blended ROAS compared to running either channel in isolation.
How long does Performance Max take to learn?
Google's learning phase for Performance Max is 6 weeks, longer than most other campaign types. The algorithm needs at least 30 to 50 conversions per week to optimise reliably. Do not make structural edits during this period. Budget adjustments of more than 20% reset the learning phase. Set a realistic budget from launch, add your audience signals and brand exclusions before going live, and review performance with at least 14 days of data before drawing conclusions.
About the author
Caner Veli built Liquiproof from zero to 3,000+ global retailers in under 6 years. He now helps DTC and CPG brands fix broken growth engines and scale 2x-15x in 90 days.