The Ecomm Analyst

You divide ad spend by new customer orders over the same window. That is the whole formula. The reason it turns into a spreadsheet project is that the two numbers live in different systems, and neither one volunteers to meet the other.

Why this ends up in a spreadsheet

Shopify knows which orders came from first-time buyers. It will tell you that in the customer reports. Meta and Google know what you spent, broken out by campaign. What neither side knows is the other half. So the standard workflow is to export a customer report, export spend from two or three ad platforms, paste it all into a sheet, line up the date ranges by hand, and hope nobody renamed a campaign mid-month.

That works. I have done it plenty of times. The problem is that it decays. The moment you add a channel, rename a campaign, or want the same number for last quarter instead of last month, you rebuild the sheet. And because the join is manual, the errors are quiet. A mismatched date range does not throw an error. It hands you a number that is off by fifteen percent and looks entirely reasonable.

What computes it without the export step

Any tool that ingests both your order data and your ad spend can do this join for you. The thing to look for is whether new customer cost is a metric the tool actually carries, or whether it expects you to derive it from two other metrics yourself. Those are very different products in practice.

ThoughtMetric, which sponsors this blog, carries cost per new customer order as its own metric rather than making you build it, and it can be grouped by channel, campaign, ad set, or individual ad. It also carries new-customer ROAS separately from blended ROAS. Triple Whale and Polar Analytics both do the same join, and Northbeam reports new and returning revenue per visitor as distinct columns. The point is not which vendor. The point is that the join happens upstream of the number you look at, which is what stops it from rotting.

Decide what new customer means before you trust the output

This is where I see the most confusion, and it has nothing to do with tooling. A first-time buyer is usually defined as someone with no prior order under that email address. That definition breaks in ordinary ways. A customer who bought as a guest, then created an account, may count twice. A customer who bought through Amazon or a retail partner and then bought direct will look brand new to your store. Someone who ordered four years ago under an old email is new by any practical measure but old by the data.

None of that makes the number useless. It means the number is a decent proxy that runs a few percent optimistic on the new customer side. Know the direction of the error and you can work with it. Assume it is exact and you will make a bad call eventually.

What I check before I act on it

First, does total spend in the tool match what the ad platforms billed you? Not attributed spend, total spend. If those disagree, nothing downstream is worth reading. Second, does new customer orders plus returning customer orders equal total orders? If it does not, something is being dropped or double counted, and you want to know which before you divide by it.

Third, look at the number weekly rather than daily. New customer CAC on a single day is mostly noise, because order timing and spend timing do not line up cleanly. A customer who saw the ad Tuesday and bought Friday makes Tuesday look expensive and Friday look free. Weekly smooths most of that out without hiding a real trend.

How this relates to blended CAC

Blended CAC divides spend by all orders. New customer CAC divides spend by first-time orders only. The second is almost always the more useful number for acquisition decisions, because it does not let repeat purchases from customers you already paid for make your ads look better than they are. I walked through the blended version and where it goes wrong in this post on calculating blended CAC.

Run both. Blended tells you what the business costs to operate. New customer tells you what growth costs. When the gap between them widens, you are getting more of your revenue from existing customers, which is either very good news or a sign that acquisition is stalling. Which one it is depends on whether new customer order volume is holding up, so keep that in view alongside the cost.