How to actually calculate
your NCAC

The Version Everyone Uses (And Why It's Lazy)

The standard formula goes like this: total ad spend divided by the number of new customers acquired. Clean, simple, and wrong in almost every case it gets applied.

The problem starts immediately with what goes into the numerator. Most brands only count media spend - the actual pounds handed to Google and Meta. They leave out agency fees, creative production costs, attribution tool subscriptions, and any internal headcount that touches paid acquisition. The result is an NCAC that flatters the channel and understates the real cost of growth.

If you're paying a £3,000/month retainer, spending £15,000/month in media, and running a £500/month attribution tool - your true acquisition cost numerator is £18,500, not £15,000. That difference compounds materially once you divide it across new customer volumes.

But the numerator problem is manageable. The denominator is where things genuinely fall apart.

The Denominator Is The Real Problem

Ask most marketers where they get their new customer count and they'll tell you: Google Ads conversions, filtered by 'new customer' in the segment section.

This is a fundamental misunderstanding of what Google Ads is actually measuring.

Google's new customer classification is based on cookie and device signals - it doesn't have access to your CRM, your order history, or your actual customer database. A customer who purchased six months ago, cleared their cookies, and converted again will likely appear as a new customer in platform. A customer who bought yesterday on a different device might also register as new.

Google fundamentally (even when you upload customer match lists), just isn't that great at reliably determining who is a new customer.

Platform-reported new customer data is directional at best. Using it as a hard denominator in your NCAC calculation is building your unit economics on sand.

Where To Actually Get Your New Customer Count

Your source of truth for new customer volume is your back-end data - Shopify, your CRM, your order management system. Not the ad platform.

The definition of a new customer should be simple and consistently applied: a customer placing their first ever order with your business. Not first order from a specific channel. Not first order in the last 12 months. First order, ever.

If your data isn't clean enough to distinguish first-time buyers reliably, that's a data infrastructure problem that needs fixing before you can trust any acquisition metric you report.

Once you have a reliable new customer count from the back end, you have an actual denominator. Now you can build a formula that means something.

The Full Formula

With the inputs sorted, the calculation looks like this:

Most brands will use a simplified version that only includes media spend and agency fees - which is fine as a working number, provided it's applied consistently and everyone understands what's been excluded. The important thing is that you define it once and don't change the definition mid-reporting period.

Calculate it monthly. Track it as a trend line, not a point in time. A single month of NCAC is almost meaningless - seasonality, promo cycles, and channel mix shifts all move it. Rolling 4, 8 & 12 week averages tells you far more about where acquisition efficiency is actually trending.

Allowable NCAC: What Your Ceiling Actually Is

Knowing your NCAC is only half the job. The number only has meaning in the context of what a new customer is actually worth to your business - and that depends entirely on your contribution margin and your repeat purchase rate.

Your allowable NCAC is subjective to the business's cash flow, profitability, retention rate & growth ambitions. This could be the maximum you can spend to acquire a new customer before you lose money per customer.

For a one-purchase-only business - low repeat rate, customers rarely come back - your allowable NCAC is constrained to a fraction of CM2 from that first order. You need to acquire profitably on the first transaction or you never recover it.

For a business with strong repeat purchase behaviour - supplements, consumables, subscriptions, any category where customers come back reliably - you can afford to loss-make on the first transaction. The first order is a customer acquisition event, not a profit event. The economics only make sense when you look at contribution across the lifetime.

This is why blanket ROAS targets are such a blunt instrument. A 3x ROAS target on a first purchase in a high-repeat category might be unnecessarily profitable - you're constraining acquisition volume when your LTV justifies spending more to bring customers in the door. A 3x ROAS target in a low-repeat, high-return category might be structurally loss-making once you account for all the variable costs.

If you have strong cohort data showing customers reliably return 2.4 times in 12 months at 40% CM, you can quantify what that future value is worth in present terms and use it to justify a higher first-order NCAC.

If you don't have cohort data, don't assume a favourable repeat rate. Default to a conservative allowable NCAC based on first-order CM2 only, and build the LTV case from actual data over time.

Connecting It Back To The Account

Once you have a reliable NCAC figure and an allowable ceiling, you have the foundation for running paid acquisition intelligently - rather than chasing a ROAS target your agency set in an onboarding call two years ago.

The practical application: if your NCAC is trending above your allowable ceiling over a rolling period, something structural is wrong. Either your spend is rising without a proportional increase in verified new customer volume - which usually means your tROAS is set too high and Google is recycling existing customers - or your channel mix is off, or you've hit saturation point in your current audience pool.

If your NCAC is comfortably below your allowable ceiling, that's a signal to push harder - you're acquiring profitably and there's likely headroom to scale. This is where most brands leave money on the table. They hit a ROAS number that looks good in the platform and stop scaling, without realising the true acquisition cost is well within a range that justifies more aggression.

The in-platform ROAS number and your back-end NCAC will rarely tell the same story. That's expected. The platform is showing you attributed efficiency on whatever cookie trail it can follow. Your back-end is showing you what actually happened. Run both, triangulate between them, and make decisions based on the back-end trend - not the dashboard.