Cost per new customer rises while purchase CPA stays flat

What you see

Ad spend per verified new customer rises while reported cost per purchase stays flat or falls. Returning customers take more of the matched orders.

Illustration: cost per new customer rises from $50 to $75 while reported purchase CPA stays at $30. — Purchase CPA hides the rising cost to acquire a new customer.

Check this first

Split matched campaign orders into first and repeat purchases before comparing costs.

Data needed

Campaign spend and reported purchases by date
Order ID, customer key, purchase time, valid-order status, and full available order history
One campaign assignment per first purchase, plus unmatched status

Run the check

#1
Use complete customer order history to mark each customer’s first valid purchase.
#2
Apply one stated source-assignment rule. Keep unmatched customers visible.
#3
Compare spend, purchase CPA, new customers, and new-customer cost over equally mature periods.
#4
For prospecting campaigns, inspect customer exclusions and new-customer settings.

Calculate

Campaign acquisition cost = campaign spend / distinct verified new customers assigned once to that campaign. Compare with platform purchase CPA separately. Also report total paid spend / all store new customers as a blended planning ratio.

Compare groups

Campaign purpose
New, returning, and unknown customer status
Source and device
Promotion period

What a healthy result looks like

New-customer cost stays within the brand’s agreed acquisition limit, and better purchase CPA also accompanies better new-customer results.

When to investigate

Use the brand’s comparable-period range and stated acquisition limit. There is no universal acceptable gap between purchase CPA and new-customer cost.

Possible causes

Returning buyers receive more delivery because customer exclusions are incomplete.
The campaign now reaches more people who already know the brand.

Rule out these explanations

Changed customer matching or incomplete order history
An intentional repeat-customer campaign
Different attribution windows or delayed orders

What to do next

Correct customer classification or exclusions when they're wrong. Set separate reporting goals for acquisition and repeat purchases.

What this check can tell you

Source assignment describes where orders are credited. It doesn't prove that advertising caused those orders.

Use this check with AI

Run a read-only check for: Cost per new customer rises while purchase CPA stays flat.
First confirm the available sources, columns, row grain, date basis, currency, and customer definition. Use only authorized data.
Required inputs: Campaign spend and reported purchases by date; Order ID, customer key, purchase time, valid-order status, and full available order history; One campaign assignment per first purchase, plus unmatched status.
Check: Ad spend per verified new customer rises while reported cost per purchase stays flat or falls. Returning customers take more of the matched orders.
Calculate: Campaign acquisition cost = campaign spend / distinct verified new customers assigned once to that campaign. Compare with platform purchase CPA separately. Also report total paid spend / all store new customers as a blended planning ratio.
Slice by: Campaign purpose; New, returning, and unknown customer status; Source and device; Promotion period.
Use this comparison rule: Use the brand’s comparable-period range and stated acquisition limit. There is no universal acceptable gap between purchase CPA and new-customer cost.
Show the source totals, calculation, unknown groups, missing inputs, and result. Don't invent fields, thresholds, customer matches, or causal effects.
Rule out: Changed customer matching or incomplete order history; An intentional repeat-customer campaign; Different attribution windows or delayed orders.
Describe the observed signal separately from possible explanations. If the check is incomplete, state the exact data needed.
Make no account or budget changes.

Sources and definitions

Google: new-customer reportingPlatform definitions and detection can differ from first-ever purchase. Align definitions and retain unknown status.Read source
Shopify: customer reportsFirst-purchase history, new versus returning customers, and equal-age cohort comparisons.Read source
Google: attribution settingsTime zone, attribution settings, and conversion windows affect comparisons.Read source