Why this matters
By default, ad platforms optimise toward “any purchase.” A returning customer who would have bought anyway counts the same as a brand-new customer the brand has never reached before. Algorithmically, the platform learns to find people likely to convert, which often means people who were going to convert without seeing the ad at all.
The result: high reported ROAS, low incremental ROAS, slow new-customer growth. The campaign looks efficient on the surface but isn’t actually growing the business.
New-customer optimization fixes this by telling the platform “this conversion was a new customer. This one was a returning customer.” The algorithm can then weight them differently, or optimise specifically for new ones.
How it works
The architecture:
- Your server determines whether each conversion is from a new or returning customer (lookup in your customer database)
- The conversion event sent to the ad platform (via CAPI, Enhanced Conversions, etc.) includes a custom parameter or event type indicating new vs. returning
- The ad platform’s algorithm reads the parameter and weights accordingly
Concrete examples per platform:
- Meta, “Estimated New Customer” signal in CAPI. Can configure campaigns to optimise for new-customer purchases specifically
- Google, “New Customer Acquisition” goal type in Performance Max. Uses Customer Match lists to determine new vs returning
- TikTok, similar via Events API custom parameters
Why server-side is required
You can only tell the platform “this is a new customer” if your server knows. Client-side pixel events don’t have access to your customer database in real-time.
The full chain:
- User completes purchase
- Your server records the order
- Server queries customer DB: is this email/customer ID known? When did they first buy?
- Server fires the conversion event to Meta CAPI / Google Enhanced / TikTok Events API with the new-vs-returning flag attached
Without server-side tracking, this signal cannot be reliably sent. Client-side pixels just don’t have the context.
What lift to expect
DTC brands that implement new-customer optimization typically see:
- 15-30% lift in new-customer count at the same CAC
- 10-20% reduction in blended CAC over 30-60 days
- A drop in “reported in-platform ROAS” (because over-attributed returning purchases are now weighted down)
- A rise in blended ROAS (because growth becomes more real)
The drop in reported ROAS is usually the most surprising part, and the one that proves the system is working. The previous high number was inflated by repeat customers. The new lower number is the actual incremental performance.
Common mistakes
- Defining “new customer” inconsistently. Is it the first lifetime purchase? First purchase in the last 365 days? Each platform expects a definition. Pick one and use it consistently.
- Sending the signal without optimising for it. The signal needs to be paired with platform configuration (Meta’s New Customer Acquisition objective, Google’s NCA goal). Just sending the parameter doesn’t change bidding.
- Skipping CAPI. Client-side pixels can’t access the customer database to make the new-vs-returning determination.
FAQ about New-Customer Optimization
What is new-customer optimisation?
New-customer optimisation is a paid-media setup that signals “this conversion came from a brand-new customer” to ad platforms, so their algorithms can prioritise prospecting (new customers) over retention (existing customers buying again).
How do I tell Meta or Google a conversion is from a new customer?
Your server determines new-vs-returning by looking up the customer in your database, then includes that signal in the CAPI / Enhanced Conversions event payload. Client-side pixels cannot do this, server-side tracking is required.
What lift can I expect from new-customer optimisation?
DTC brands typically see 15-30% lift in new-customer count at the same CAC, plus a 10-20% drop in blended CAC over 30-60 days. The previously-reported in-platform ROAS also drops as over-attributed repeat purchases get weighted down.