Google Performance Max: Are You Feeding It the Right Data?

Performance Max is Google’s fully automated campaign type. You provide assets and conversion goals, and Google’s systems decide where, when, and to whom to show your ads across Search, Display, YouTube, Gmail, Discover, and Maps.

The theory is sound. The practice depends almost entirely on one thing: the quality of the conversion data you’re feeding it.

The Garbage In, Garbage Out Problem

PMax optimises toward your reported conversions. If your conversion tracking is wrong, PMax learns to optimise toward the wrong thing.

Common tracking failures that corrupt PMax learning:

Duplicate conversions — if your purchase event fires twice per transaction (common with Shopify and some WooCommerce setups), Google thinks you’re generating twice the revenue. PMax targets users who generate “conversions” — which in this case includes everyone who accidentally double-fires a tag.

Soft conversions counted as primary — if “newsletter signup” and “purchase” are both set as primary conversions with equal value, PMax will chase signups as happily as it chases purchases. It’s much cheaper to generate a signup. Left unchecked, you’ll see your signup volume spike and your ROAS collapse.

Missing value data — if you’re passing a static £1 value for all purchases instead of actual transaction values, smart bidding can’t optimise for high-value customers. All customers look equal. They’re not.

Consent Mode issues — if Consent Mode v2 isn’t properly configured, you’re losing modelled conversions. PMax gets less signal. Bidding becomes less efficient.

What Good Conversion Data Looks Like

For e-commerce:

• purchase event as the only primary conversion
• Dynamic value and currency parameters populated from actual transaction data
• transaction_id included (prevents duplicate counting)
• Enhanced Conversions set up (hashed email at checkout improves match rates by 15–30%)
• Consent Mode v2 in Advanced mode

For lead generation:

• Primary conversion = the qualified outcome (booked call, completed application, qualified form)
• Downstream funnel stages (MQL, SQL) imported from CRM via Google Ads conversion import
• Lead value assigned based on average close rate × deal value

Audience Signals Matter Too

PMax lets you provide “audience signals” — these aren’t targeting restrictions, they’re hints to accelerate learning. Good signals include:

• Your customer list (existing buyers)
• Website visitors who completed key actions
• Custom intent audiences built from search terms that indicate purchase intent

Without signals, PMax starts with zero data and figures out your audience from scratch. This “cold start” period burns budget. Good signals shorten it significantly.

How to Audit Your PMax Conversion Setup

1. Open Google Ads → Tools → Conversions
2. Check which conversions are set to “Primary” — there should be one per campaign objective
3. Check that value-based conversion actions have actual values (not static £1)
4. Go to Campaign → Insights → Auction Insights — unexpectedly high impression share from competitors at low cost often signals your targeting has drifted because of poor data

If you’re running PMax and your ROAS has been declining despite stable creative, conversion data quality is the first place to look. Book an hour and I’ll audit it with you.