Performance Max is the campaign type Shopify advertisers love to hate. The reporting is opaque, the controls are limited, and the default setup Google nudges you into is rarely the best one. But for ecommerce, PMax tends to outperform standalone Shopping campaigns when it's structured well — and most stores never get past the default structure.
This guide is the asset group framework we use for client accounts running PMax at $5K-$200K/month in spend. The principle is simple: give PMax enough variety to optimize, but not so much fragmentation that no single asset group has signal.
What asset groups actually do
An asset group is PMax's version of an ad group. Each one has its own creative, audience signals, and listing groups (the products eligible for that asset group). PMax distributes your campaign budget across asset groups based on which is converting cheapest.
The two big mistakes we see:
- One asset group with all products jumbled together. Google can't tailor creative or signal, and asset rotation suffers.
- One asset group per product. Signal too thin, none of them ever exit learning.
The right answer is in the middle.
The structure we use by default
For a typical Shopify catalog with 50-500 SKUs across a few product categories, our base structure is:
Campaign 1: Core PMax (catch-all commerce)
- Asset group A: Hero SKUs (top 10% of revenue, high margin)
- Asset group B: Best-sellers (next 30% of catalog by revenue)
- Asset group C: Long tail (everything else, simpler creative)
Campaign 2: Brand defense PMax (only if you have brand search volume worth defending)
- Single asset group with brand-focused creative
Campaign 3: New customer PMax (only at $50K+/month spend)
- Asset group focused on new-customer acquisition with brand exclusions and audience signals weighted toward unfamiliar prospects
The asset group split inside the core campaign is what does most of the work.
Why hero / best-seller / long tail beats category split
The instinct most operators have is to split asset groups by product category — hats, shirts, accessories, etc. That instinct is usually wrong for PMax.
Here's why: PMax optimization works best when each asset group has a distinct conversion profile. Category splits don't always create distinct profiles. Hero/best-seller/long-tail does.
- Hero SKUs need premium creative, broader audience signals, and budget guarantees
- Best-sellers need volume creative and broader catalog presentation
- Long tail needs lighter creative investment and product-feed-driven distribution
When you split by category, your "hats" asset group ends up with hero hats and long-tail hats fighting each other in the same group.
If you have very different category economics (apparel + supplements, say), then a category split makes sense. For most single-vertical stores, the revenue-tier split wins.
Asset group setup details
For each asset group:
Listing groups. Use product type, custom labels, or item ID lists to gate which products show in each asset group. We typically use Shopify custom labels (custom_label_0, custom_label_1) that we sync from a Google Sheet via a feed management tool.
Audience signals. Add your customer list, your top 1% website visitors, and any in-market segments relevant to the asset group. Audience signals don't restrict targeting — they nudge optimization. More relevant signals = faster optimization.
Final URLs. Send to category pages or collection pages, not the homepage. PMax respects final URL expansion by default — turn it off if you have legal or brand reasons to control which pages traffic lands on.
Creative. 5+ headlines, 5+ descriptions, 4+ images, 1+ logo, 1+ video minimum. Yes, even if you don't have video — Google will auto-generate one if you skip it, and it'll be terrible. Spend $200 on Fiverr for a 15-second product video before launching.
What goes in each asset group's creative
We tailor creative to the asset group's role.
Hero SKU asset group: Premium product photography, brand-led headlines, longer-form descriptions emphasizing positioning. Think "the chair Architectural Digest covered" not "save 20% today."
Best-seller asset group: Conversion-focused creative — social proof, ratings, specific benefits. "10,000+ five-star reviews" energy.
Long-tail asset group: Lighter creative investment, more reliance on the product feed itself. Generic but well-written headlines that work across many SKUs.
Audience signals: what to actually use
Audience signals are PMax's nudge for who to optimize toward. They're not targeting; they're seed data.
What works:
- Your customer list (uploaded as a Customer Match audience)
- Top 1-5% of website visitors (high-intent custom segment)
- In-market segments relevant to your category
- Detailed demographics if you have meaningful skews
- Custom segments built from search terms ("running shoes" if you sell running shoes)
What doesn't work:
- Affinity audiences (too broad)
- Stacking 20 audience signals (dilutes)
- Random in-market segments because they sounded relevant
We typically include 3-5 audience signals per asset group, weighted toward customer match and top-visitor segments.
Excluding brand search from PMax
By default, PMax happily eats your branded search traffic at lower CPA than your brand campaign — but the customers it serves were going to buy anyway. This inflates PMax ROAS and starves your brand campaign of impressions.
Use the brand exclusion option in PMax campaign settings to keep brand terms out. Then run a dedicated brand search campaign with manual CPC bidding for brand defense. Your blended CAC drops because you stop paying PMax to retarget your existing customers.
This is now table-stakes setup. If your PMax campaign isn't excluding brand, you're paying for the privilege of inflating your dashboards.
Common asset group mistakes
Five asset groups with $30/day campaign budget. PMax can't optimize across five groups at that spend. Consolidate to two.
Different asset groups serving the same products. Listing groups should be mutually exclusive within a campaign. Overlap means your asset groups bid against each other.
No customer list as audience signal. This is a free 10-15% performance lift that operators routinely skip.
Final URL expansion disabled with no replacement. If you turn off URL expansion, you're sending all traffic to the URL you specified. Make sure that URL is actually high-converting.
Treating the asset group score as gospel. Google's "Excellent" rating means your assets are loaded, not that they're performing. Check actual asset performance reports, not the badge.
How to read PMax reports (since they hide things)
PMax reporting is famously opaque, but you can extract more than the default dashboards show.
- Asset group performance: click into each asset group, look at conversion volume and CPA per group, not just campaign blended.
- Asset reports: under each asset group, you'll see "Best," "Good," and "Low" performance ratings on individual assets. Replace "Low" assets monthly.
- Search categories: the Insights tab shows search themes converting through PMax. Use this to inform your search campaign keyword expansion.
- Audiences insights: which audience segments are converting. Use this to refine audience signals.
The reports won't give you the granular keyword/placement data old Shopping gave you. Accept that and focus on what PMax does show.
What to do this week
Audit your current PMax. Are you on a single asset group? Split into hero / best-seller / long tail this week. Are brand exclusions enabled? If not, enable them. Are your asset groups loaded with the minimum 5/5/4/1/1 creative count? If not, fill them in — even if some assets are placeholder quality at first.
For more, see our Google Shopping feed optimization checklist, the Google Ads search vs PMax cannibalization guide, and our broader Shopify Google Ads guide.