Skip to content
E-commerce performance dashboard

Performance Max for Shopping: Pros, Cons, and How to Win

· by Digitelia · 4 min read

Performance Max for retail goals (the campaign type that replaced Smart Shopping in 2022) has become the dominant Google Ads format for e-commerce. By 2026, it accounts for roughly 60-70% of Google Ads e-commerce spend across the accounts we manage. Done right, it delivers ROAS 20-40% higher than the Standard Shopping it replaced. Done wrong, it cannibalizes Search, eats your branded traffic, and obscures what’s actually driving sales.

This guide walks through the campaign architecture that wins in 2026, the failure modes that crater ROAS, and how to combine PMax with other formats.

E-commerce campaign performance

What Performance Max for retail actually is

PMax for retail is a single campaign that draws from your Merchant Center product feed and serves ads across:

  • Google Search shopping results
  • Google Shopping tab
  • Display Network
  • YouTube
  • Discover
  • Gmail
  • Maps (Local Inventory Ads if enabled)

The algorithm picks placement, audience, creative combination, and bid for every auction. You provide:

  • Connected Merchant Center catalog
  • Asset groups (creative themed by concept)
  • Audience signals (hints to the algorithm)
  • Conversion goal with Smart Bidding strategy
  • Budget

1. Catalog breadth. PMax discovers placement-by-placement which products win in which contexts. With 200+ SKUs, manual segmentation is impossible; PMax’s automated discovery is genuinely superior.

2. Multi-surface buyer journeys. Customers who research on YouTube, see a Display ad, then search and convert — PMax attributes and bids across that full journey.

3. Audience signal leverage. If you have rich first-party data (Customer Match lists), PMax leverages it more powerfully than Standard Shopping.

4. Performance scaling. Above $30K/month spend, PMax typically outperforms equivalent Standard Shopping. The algorithm benefits from data density.

5. Creative testing at scale. The asset group structure means you can supply 30+ creative combinations and let the algorithm pick winners. Manual A/B testing at that scale is impractical.

Where Standard Shopping (or Search) still wins

1. Small catalogs (<50 SKUs). PMax’s algorithm benefits from variety. With few SKUs, Standard Shopping’s manual control often outperforms.

2. Very low budgets (<$2K/month). Not enough conversion data for PMax to optimize. Standard Shopping’s simpler logic works better.

3. Highly specific product targeting needs. Want to bid 3× on holiday products, 0.3× on clearance? Manual Shopping campaigns give you that control; PMax doesn’t.

4. Audit and reporting needs. Standard Shopping reports cleaner query-level and product-level data. PMax’s Insights tab has improved but still less granular.

5. Margin-sensitive accounts. PMax tends to push toward high-volume, lower-margin products if not constrained. Manual Shopping with margin-based segmentation can squeeze more profit.

The asset group structure that wins

Performance Max’s asset groups are the equivalent of Shopping campaigns’ product groups, but themed differently. Within one PMax campaign, you can have 1-100 asset groups (recommendation: 3-7).

Naming/structure that works:

Pattern 1: Themed by product line.

  • “Best Sellers” asset group → top SKUs, high-volume creative
  • “Premium Line” → high-AOV SKUs
  • “Clearance” → looser tROAS for inventory move
  • “Seasonal” → time-bound promotional creative

Pattern 2: Themed by audience.

  • “Cold prospecting” → broader creative, awareness messaging
  • “Retargeting” → urgency messaging, repeat-customer signals

Pattern 3: Themed by margin (most effective for ROAS optimization).

  • “High-margin” → aggressive tROAS (e.g., 6.0)
  • “Mid-margin” → moderate tROAS (e.g., 4.0)
  • “Low-margin or sale” → loose tROAS (e.g., 2.5)

Each asset group should have at minimum:

  • 5+ headlines
  • 5+ long headlines
  • 5+ descriptions
  • 10+ product images (from your feed)
  • 1+ lifestyle/brand images
  • 1+ short video (15-30 seconds) — without one, YouTube placements get suboptimal creative

Product feed and creative library

Audience signals: how to actually use them

Audience signals are hints, not strict targeting. PMax will serve outside your signals — they bias, don’t restrict.

The hierarchy of signal quality:

  1. Customer Match (your email lists of buyers, leads). Best signal.
  2. Lookalikes of Customer Match (built automatically inside PMax). Strong.
  3. Site visitors (last 30-90 days), segmented by intent. Moderate.
  4. In-market audiences (Google’s algorithmic intent). Weaker but better than none.
  5. Affinity audiences (long-term interest categories). Weakest signal. Avoid relying on these.

For e-commerce PMax, always include at minimum:

  • Customer Match list of past buyers (Google builds lookalikes automatically)
  • Custom audience of cart abandoners (last 30 days)
  • Custom audience of product page viewers (last 60 days)

If you have <1,000 customers in your CRM, upload anyway. Even small Customer Match lists improve signal quality.

Bidding strategy by stage

For new PMax campaigns:

  • Weeks 1-2: Maximize Conversion Value (no tROAS). Let it explore.
  • Weeks 3-6: After 50+ conversions, layer in tROAS at a generous target (1.5-2× your minimum acceptable).
  • Weeks 7+: Tighten tROAS in 10-15% increments toward your real target.

For mature accounts:

  • tROAS adjustments in 10-15% increments every 21 days max.
  • Avoid changing strategy or major settings during learning periods.

Campaign structure: how many PMax campaigns?

The default question: one big PMax campaign or several?

One big campaign: maximum data pooling for the algorithm. Pros: faster learning, better optimization. Cons: less granular margin control.

Several campaigns by margin tier: separate Best Sellers, Mid-margin, Clearance into different campaigns with different tROAS. Pros: margin control. Cons: data splits — each campaign optimizes independently.

Several campaigns by audience tier: one for cold prospecting, one for retargeting. Pros: budget control by funnel stage. Cons: usually overcomplicates.

For most e-commerce accounts: 2-4 PMax campaigns. One per major margin tier or one per major product line.

The branded keyword problem

PMax automatically serves on search queries — including your branded queries. This means it can eat your existing branded Search campaign or even branded organic clicks (you can’t pay for organic, but PMax replacing them with paid is wasteful).

The solution: brand exclusion lists. Tools → Account Settings → Brand exclusions → add your brand keywords. PMax then won’t serve on those queries.

Always set this up before launching PMax. Critical for any account running branded Search campaigns separately.

Combining PMax with Search campaigns

The hybrid structure that works for most e-commerce:

  • PMax campaigns for catalog and multi-surface acquisition (70-80% of budget)
  • Branded Search campaign for brand defense (10-15% of budget, exact match)
  • Non-branded head terms Search campaign for terms you want full visibility and control on (10-15%)

This combo lets PMax do what it does well (catalog + multi-surface + algorithmic discovery) while preserving control on branded and competitive head terms.

Common PMax failure modes

1. No video assets. PMax with no video defaults to bad auto-generated thumbnails for YouTube placements. Always include 1-3 short videos per asset group.

2. Single asset group with everything in it. No creative thematic discipline. Algorithm has no way to bucket creative for different intent.

3. tROAS set too aggressively at launch. Campaign learns slowly, conversions stall, you panic. Start loose, tighten gradually.

4. No Customer Match audience signals. Cold start with weak signals. PMax leans on in-market and affinity, which are weakest.

5. Mixing high-margin and clearance in one campaign. Algorithm optimizes for blended ROAS, starving high-margin products. Split by margin.

6. Cannibalizing branded Search. No brand exclusion list applied. PMax wins your branded auctions at higher CPCs than Search would have.

7. Insufficient catalog data quality. Disapproved products, missing GTINs, poor titles. PMax can only work with what’s in the feed.

8. Daily budget too small. Below $50/day per PMax campaign, the algorithm can’t gather enough signal. Either consolidate or increase budget.

Reading PMax reports

PMax reporting is opaque by design — Google wants you to trust the algorithm, not micromanage. The reports that matter:

1. Insights tab. Top queries (limited but growing in 2026), audience segments performing well, asset performance ratings.

2. Asset details report. Per-asset performance ratings: “Best”, “Good”, “Low”. Replace “Low” assets with new ones.

3. Search Term Insights. Aggregate query data showing themes PMax is winning on. Less granular than Search but useful.

4. Audience Insights. Which audience segments PMax is serving to. Useful for refining signals.

5. Product Insights (e-commerce). Which products are getting the most impressions, which are converting. Critical for spotting underperformers to feed-optimize or exclude.

A 60-day PMax launch plan

Days 1-10: Setup.

  • Confirm Merchant Center health (no disapprovals).
  • Build 3-5 asset groups themed by product line or margin.
  • Upload Customer Match lists.
  • Set brand exclusion list.
  • Launch with Maximize Conversion Value bidding, no tROAS.

Days 11-30: Learning.

  • Don’t touch campaign settings.
  • Monitor for 50+ conversions to exit learning phase.
  • Refresh underperforming creative assets weekly.

Days 31-45: Optimization.

  • Add tROAS target (generous, 1.5-2× minimum).
  • Identify and pause underperforming asset groups.
  • Add new audience signals based on Insights.

Days 46-60: Scale.

  • Tighten tROAS in 10-15% increments.
  • Increase budget on winning campaigns 20% per week.
  • Consider splitting into margin-tier campaigns if scale warrants.

Frequently asked questions

Can I see what queries PMax served on? Partial. Insights tab → Search Term Insights shows aggregate themes. Not individual queries like Search campaigns. For full visibility, request the search terms insights report via API.

Should I run PMax alongside Standard Shopping? Generally no. They compete, and PMax wins most auctions because of broader signal. Pick one for catalog-level acquisition.

What’s the minimum budget for PMax to work? Roughly $50/day or 10× your tCPA per week, whichever is higher. Below that, the algorithm doesn’t have enough volume to converge.

Does PMax replace Performance Max for lead gen? Performance Max for retail (with Merchant Center catalog) is different from Performance Max for lead gen (Search/Display goals, no catalog). Different campaign types, similar names. Pick the one that matches your goal.

Can I exclude specific products from PMax? Yes — via supplemental feed with a “excluded_destinations” attribute, or by deactivating products in Merchant Center. Useful for stopping PMax on clearance items you don’t want pushed.

Performance Max for retail in 2026 is the highest-leverage e-commerce campaign type Google has — when you set it up right. The accounts that complain it doesn’t work usually haven’t fixed feed quality, haven’t added Customer Match signals, or haven’t structured asset groups deliberately. Done well, PMax + Search hybrid + clean feed is the modern e-commerce paid acquisition stack.

Tagged

#performance-max#shopping#google-ads#ecommerce#asset-groups#roas