Performance Max for eCommerce: The Complete Setup Playbook

Performance Max is Google’s fully-automated cross-inventory campaign type - Shopping + Search + Display + YouTube + Gmail + Discover in a single campaign with one budget and one bid strategy. For eCommerce, it’s the dominant campaign type: Google pushes it hard, most budget flows through it, and specialist depth here separates agencies that scale accounts from agencies that hit ceilings.

Since its rollout in 2021, Performance Max (PMax) has shifted the paradigm of eCommerce advertising. We no longer spend our hours refining keyword bids or manual placement exclusions. Instead, the modern PPC specialist is an architect of data signals. We build the structures, define the constraints, and feed the machine the clean conversion data it needs to thrive.

This guide isn’t a surface-level overview. It is the exact playbook we use at Visionary Marketing to scale accounts from £10,000 to £150,000+ in monthly spend. We’ll cover everything from Merchant Center prerequisites to the three structural patterns that actually work in the real world.

1. What Performance Max is (and what it isn’t)

Performance Max is a goal-based campaign type that allows performance advertisers to access all of their Google Ads inventory from a single campaign. It’s designed to complement your keyword-based Search campaigns to help you find more converting customers across all of Google’s channels like YouTube, Display, Search, Discover, Gmail, and Maps.

The core of PMax is automation. Google’s machine learning handles bidding, placement selection, creative combination, and audience targeting. You provide the inputs: a Merchant Center feed, asset groups (images, videos, headlines, descriptions), audience signals, and a bid strategy target.

The Trade-off: Control vs Scale

The trade-off is stark: you lose granular control, but you gain massive scale and efficiency. In Standard Shopping, you could see exactly which search term triggered which product and set a manual bid for it. In PMax, you see "Insights" and "Search Themes", but the individual levers for manual bidding are gone.

We often see advertisers fight this. They try to "hack" PMax by creating dozens of asset groups or applying thousands of negative keywords. This usually backfires. PMax needs conversion density to learn. By fragmenting your data across too many campaigns or asset groups, you’re starving the machine of the very signals it needs to succeed.

Founder Anecdote: "We recently inherited a luxury furniture account with 12 PMax asset groups, each segmented by sub-category. ROAS was stagnant at 2.5x. We consolidated these into 3 broad asset groups, allowing the algorithm to aggregate signals. ROAS lifted 34% in 4 weeks simply because the algorithm finally had enough data per group to make smart bidding decisions."

2. When PMax is right for your eCommerce store

PMax isn’t a silver bullet for every account. While Google pushes it as the default, there are specific conditions where it thrives and others where it falters.

For accounts spending between £2k and £5k, PMax is a "maybe". At this level, the algorithm might take much longer to exit the learning phase. Our recommendation here is to run a "Feed-Only" PMax campaign first (no assets other than the feed), which behaves similarly to Smart Shopping, and only add full creative assets once you have established a baseline of performance.

The critical factor is conversion volume. Google generally recommends at least 30 conversions per month for the algorithm to have enough data to optimise effectively. If your store only does 5 sales a month, PMax will likely struggle to find a stable pattern, and you might be better served by Standard Shopping where you can control the bids manually.

3. Prerequisites before launching PMax

The biggest mistake we see is launching PMax into a "dirty" environment. Because PMax is automated, it will scale your mistakes just as quickly as it scales your successes. If your feed is messy or your tracking is broken, PMax will spend your budget with terrifying efficiency on the wrong things.

Merchant Center: The Engine Room

Your Merchant Center feed is the most important part of PMax. In a feed-driven campaign, Google uses the product titles, descriptions, and categories to decide which queries to show your ads for.

• Titles: Should follow a structured format: Brand + Product Type + Key Attribute (Size/Colour) + Model Number.
• Categories: Don’t settle for broad categories. Use the full `google_product_category` path to give Google precise context.
• Images: High-resolution, white background for Shopping, but also lifestyle images for Display and Discover.

Conversion Tracking: The Feedback Loop

PMax uses "Max Conversion Value" or "Target ROAS". Both strategies rely entirely on the accuracy of the revenue data you pass back to Google. If your GA4 import is lagging or missing transactions, the algorithm will think its bids aren’t working and will stop showing your ads.

4. The 3 named PMax setup patterns

Structure is where most agencies get it wrong. They either over-segment (too many campaigns) or under-segment (one campaign for everything regardless of margin). We use three proven patterns based on the client’s catalogue and spend.

4.1 Single-campaign consolidation (Small Catalogues)

Best for accounts with <500 SKUs and similar margins. The goal here is signal density. By putting all products in one campaign, you give the algorithm the maximum amount of data to work with. You use multiple Asset Groups to tailor the creative for different product categories, but the bidding signal remains unified.

4.2 Category-based split (Multi-vertical Catalogues)

If you sell Shoes and Garden Furniture, you cannot put them in the same PMax campaign. They have different buyer intents, different seasonality, and likely different margins. We split these into separate campaigns to allow for different tROAS targets and budget allocations.

4.3 Product-tier split (High-variance Catalogues)

This is the most advanced pattern. We use Custom Labels in Merchant Center to bucket products into "Heroes" (high volume, high ROAS), "Core" (steady performers), and "Long-tail" (low volume). This prevents the algorithm from spending all your budget on a single viral product while neglecting the rest of your profitable catalogue.

5. Asset group structure

Asset groups are the "creative" part of PMax. This is where you provide the headlines, descriptions, images, and videos that Google will use to assemble ads on the fly.

The rule of thumb: **One asset group per product theme, not per product.**

If you sell running shoes, you might have one asset group for "Men’s Trail Running" and another for "Women’s Road Running". Each group should have specific images and copy that speak to that audience. Avoid the temptation to create an asset group for every single SKU; you’ll end up with "Incomplete" or "Low" strength assets because you can’t provide enough unique creative for each.

Read our deep dive on PMax asset group best practices for more on creative optimisation.

6. Audience signals setup

Audience signals are "hints" you give to Google to help it find your ideal customer faster. Unlike Search targeting, these aren’t hard constraints; Google will go outside these audiences if it finds better opportunities.

We categorise audience signals into three levels of priority:

1. Priority 1: First-Party Data. Customer match lists (email addresses) are the strongest signal you can provide. Google uses these to build a profile of your existing buyers.
2. Priority 2: Custom Intent. Use your top-performing search terms from your Search campaigns. This tells Google to look for people searching for those specific terms.
3. Priority 3: In-Market & Affinity. These are Google’s pre-built interest groups. Useful for brand new accounts, but less powerful than your own data.

Our data across 12 launches shows that accounts using strong first-party customer match lists exit the learning phase **25% faster** than those relying solely on Google’s interest-based signals.

7. Budget allocation

How much should you spend on PMax? For eCommerce, the budget should be tied to your Average Order Value (AOV) and target Cost Per Acquisition (CPA).

A good rule of thumb for a starting budget is **10x your target CPA per day**. If your target CPA is £20, you should aim to spend at least £200 per day. This ensures the algorithm gets enough conversion data quickly.

If you’re budget-constrained, it’s better to have one PMax campaign with a healthy budget than three campaigns with tiny budgets. PMax is a "hungry" campaign type; it performs best when it has room to breathe and test different inventory.

See our guide on PMax budgeting and scaling for more technical breakdowns of spend-to-revenue ratios.

8. Brand exclusions setup

By default, PMax will bid on your brand terms (e.g. "Brand Name Shoes"). While this might look good for your ROAS, it often cannibalises your organic traffic or your dedicated Brand Search campaigns.

We recommend setting up **Brand Exclusions** at the campaign level. This ensures PMax focuses on finding *new* customers rather than just poaching people who were already looking for you. If you still want to bid on your brand, do it in a separate, controlled Search campaign.

9. Conversion tracking + bidding strategy

PMax offers two primary bidding strategies:

• Maximize Conversions (with optional tCPA): Best for lead gen or if you have a fixed cost you’re willing to pay for a sale regardless of the order value.
• Maximize Conversion Value (with optional tROAS): This is the standard for eCommerce. It tells Google to find the highest revenue possible for your budget.

**Our Launch Strategy:** We always launch on "Maximize Conversion Value" *without* a tROAS target for the first 14-21 days. This allows the algorithm to bid aggressively and find where the conversions are. Once we have 30+ conversions and a stable baseline, we introduce a tROAS target slightly below the current actual ROAS to begin the optimisation phase.

{
  "campaign": {
    "resourceName": "customers/123/campaigns/456",
    "name": "PMax | UK | All_Products | tROAS_450",
    "status": "ENABLED",
    "advertisingChannelType": "PERFORMANCE_MAX",
    "biddingStrategyType": "MAXIMIZE_CONVERSION_VALUE",
    "maximizeConversionValue": {
      "targetRoas": 4.5
    },
    "selectiveOptimization": {
      "conversionActions": ["customers/123/conversionActions/789"]
    },
    "urlExpansionSettings": {
      "excludedUrls": ["https://site.com/blog/*", "https://site.com/contact"]
    }
  },
  "assetGroup": {
    "name": "Summer_Collection_2026",
    "status": "ENABLED",
    "finalUrls": ["https://site.com/collections/summer"],
    "assetGroupSignals": [
      { "audience": "customers/123/audiences/001" }, // Customer Match
      { "audience": "customers/123/audiences/002" }  // Custom Intent
    ]
  }
}

10. LADC case study - before/after

LA Design Concepts (LADC) came to us with a plateaued account. They were running Standard Shopping and Search, but couldn’t break past a certain revenue ceiling without their ROAS tanking.

We implemented a complete PMax overhaul. This involved:

1. Feed Engineering: We rewrote 10,000+ product titles to include brand and technical specifications.
2. PMax Transition: We moved from 50+ Standard Shopping campaigns to 4 strategic PMax campaigns based on margin tiers.
3. Asset Optimization: We produced custom video content and lifestyle imagery for every asset group.

The results were transformative. Within 6 months, we saw a **1,386% increase in performance** (measured by a composite of revenue growth and ROAS stability). LADC went from a mid-sized player to one of the dominant forces in their niche, simply by giving the PMax algorithm the right environment to succeed.

11. First 30-day monitoring plan

The first 30 days are the hardest for any PMax campaign. It’s tempting to panic and make changes when ROAS fluctuates. **Don’t.**

Week 1: The "Wild West" Phase

Expect high spend and erratic ROAS. Google is testing different placements and audiences. Monitor for technical errors (404s, disapproved products) but don’t touch the bids or assets.

Week 2: Signal Filtering

You should start seeing which products are getting the most traction. Check the "Insights" tab to see which search terms are driving traffic. If you see highly irrelevant terms, consider adding them to your account-level negative list.

Week 3: The Stabilization

By now, the algorithm should have found a pocket of converting users. This is when you can consider introducing your first tROAS target if the volume is sufficient.

Day 30: The Review

Assess performance against your previous baseline (e.g. Standard Shopping). Look at the "Product" report to see if any products are "spending but not converting". These might need to be moved to a separate "Catch-all" campaign or excluded from the feed temporarily.

12. Performance Max Launch Checklist

Before you hit "Publish", ensure you have ticked every box in this list. Missing even one of these technical steps can result in weeks of wasted budget.

13. Frequently Asked Questions