Helping a skincare brand see the sales its Meta ads were already making

The numbers didn't add up

The brand sells cleansers, serums and moisturisers through its own Shopify store, and most of its new customers come from Facebook and Instagram ads. The founder had noticed something odd. In a good week Shopify would show a couple of hundred orders, while Meta's Ads Manager claimed far fewer.

Going by Meta's numbers, several campaigns looked like they were losing money, and the plan was to cut them.

Finding the gap

Before changing any ads, we compared the two sets of numbers order by order for a month. Meta was seeing only about 62% of purchases. The rest were going missing because of browser privacy settings, ad blockers and Apple's tracking changes on iPhones, which stop a lot of what normally happens in the browser.

This is very common, and it causes a second problem that's easy to miss. Meta's algorithm learns from the purchases it can see, so with four in ten missing it was getting steadily worse at finding buyers.

Fixing the tracking

We set up Meta's Conversions API, which sends purchases to Meta straight from the server instead of relying on the customer's browser. We made sure each order was only counted once, even when both the browser and the server reported it, and we sent the order value along so Meta could tell a £15 travel-size order from a £90 bundle.

Then we checked the figures against Shopify every week until they lined up. A few weeks in, Meta was seeing about 96% of orders.

What we found once we could see properly

The biggest surprise was a campaign built around customer reviews. It had looked like the weakest of the lot and was first in line to be switched off. With accurate data, it turned out to bring in the customers most likely to buy a second time.

We moved budget towards it and away from a broad campaign that mostly reached people who clicked and didn't buy. We also gave returning customers their own campaign, so the brand wasn't paying to win back people as if they were new.

Three months later

Measured return on ad spend went up by 41%, and the cost of winning a new customer came down by 23%. Some of that improvement is simply Meta seeing sales that were already happening. The rest came from budget decisions that were only possible once the numbers could be trusted.