Twice the Meta budget, and the return held.

A Singapore womenswear brand gave Meta audiences built from each shopper's predicted value. The budget doubled over 79 days, and return on ad spend stayed within 8%.

Book 30 minutes All case studies

Womenswear · Shopify · Singapore, SEA, US, Australia
One Meta campaign, 79 days, mid-2026
Revenue from Shopify orders, not ad reports

Meta campaign · 79 daysads

Value-based lookalikenew customers

Visitors, weighted by predicted valuemid-funnel

Customersbottom

Daily budget

2.1×

ROAS within 8%

Buyer rate

+65%

profiles reached

2.1×daily budget over 79 days, return on ad spend held within 8%

+88%store revenue per click, same 60 days, 2025 against 2026

+65%buyer rate of profiles the campaign reached, matched 30-day cohort

+33%revenue per customer, same 60 days year on year

The problem

Meta could recognize less than a third of its own visitors.

Facebook brought half or more of the store's traffic. Only about 30% of those visitors carried an identifier the pixel could use, because most arrived in Facebook's in-app browser.

Of 3,000 abandoned browsers on a typical day, 101 could be retargeted. Reported purchases were counted 1.8 to 1.9 times. Audiences came from broad demographics and past behavior.

What the agent changed

Meta learns who is worth finding

Audiences from future value

For every shopper, the agent predicts spend over the next 12 months. That seeds a lookalike of new people most likely to become high-value customers.

Complete purchase signal

Every session is captured server-side, including in-app browser traffic, and Meta gets clean events, each purchase counted once.

One campaign, three audiences

A value-based lookalike for new customers, visitors weighted by predicted value, and existing customers.

Results

Same season, a year apart

The same 60 days, 23 May to 22 July, in 2025 and 2026. Revenue from Shopify orders. A year-on-year comparison, not a controlled test.

Measure	2025, before Releva	2026, with Releva	Change
Meta spend, indexed	100	89	−11%
Store revenue per Meta dollar	6.09	6.45	+6%
Store revenue per click, indexed	100	188	+88%
Revenue per customer, indexed	100	133	+33%

The spring-to-summer drop in efficiency was −24% in 2025 and −12% in 2026.

Inside the campaign

The future-value audience led on return

Over 72 days, with the same creative in the same countries, the value-based lookalike returned 5.67 on ad spend, against 5.28 for a broad demographic audience.

In a matched 30-day cohort, profiles the campaign reached became buyers at 0.56%, against 0.34% for the old campaign: 65% higher.

Ad sets · 72 days · return on ad spendads

Value-based lookalike5.67

Customers and visitors5.46

Broad demographic5.28

How we counted

Store data. Revenue figures are Shopify orders, not Meta's attribution.

Same season, not a controlled test. The yearly comparison uses the same 60 days, so the season matches. Other things changed too, so it is not a controlled result.

Get the method and setup notes

How each number on this page was counted, the definitions, and what was switched on in the store, in order. Enter your work email and the notes open; we also send you the link.

Thanks. Open the method and setup notes.

We use your email to know who reads our case studies, and to send the link. Privacy policy