A Decision Making Framework for Meta attributed nCPA (New Customer CPA)

Written by
Michael @UpStackified

I've been getting some questions from brands and marketers regarding tracking nCPA (New Customer CPA) and nROAS in Meta once it has been enabled by UpStackified. I will explain below in depth my thinking, on what, why, and how to use Meta nCPA and how it compares to true nCPA, and which to use when.

A large DTC brand asked me today if they should trust the Meta attributed nCPA or if they should use blended nCPA which they calculate manually? If blended nCPA is the answer, then what is the value of Meta attributed nCPA and how should they provide direction to their media buyer on metrics for success?

Quick Backstory: A few months ago we introduced NC-CPA (aka nCPA, aka new customer CPA) and nROAS metrics reported directly in Meta. (though we weren't the first, we did popularize the adoption cc: @johnhickey1970). Since popularizing this data in Meta, most of our main CAPI competition has fast-followed and introduced similar features into their platform with varying degrees of success.

So, Is Meta attributed nCPA accurate and worthwhile or is true(blended) nCPA most accurate??? Can we use FB attributed nCPA for decision making, or should we be using blended nCPA?

This is a great question. Both FB attributed nCPA and blended nCPA are correct and I'll do my best to explain my thinking on the topic.

Our in platform (FB) nCPA that UpStackified provides is very accurate for measuring the NC which facebook has attributed the acquisition. UpStackified does improve the attribution on Facebook significantly and reduces Facebook reliance on data modeling for attribution. (among many other benefits).

Facebook attributed nCPA is a great tool for decision making on campaigns, comparing performance of ads, and establishing your facebook attributed nCPA baseline metric to use it directionally to understand performance. (more on this below)

However, marketing broadly, like Facebook, usually has a large "halo effect", where platform spend and attributed performance increase, so does un-attributed performance and conversions. Generally these are highly correlated.

Your true(blended) nCPA is the business KPI, not a facebook metric, and is calculated by the following:

true nCPA = total spend / # total NC (using Shopify NC is fine)

Because the nCPA of facebook is a calculation of attributable new customers, there will always be a pool of NC who went unattributed, sometimes for technical reasons, like they came back on another device, or fell outside the attribution window, and sometimes just because humans are unpredictable and 100% attribution for any marketing campaign, on any platform, has never been, and is never going to be achievable. (Do you think the guys on Mad Men were tracking attribution?)

It is important to think about Facebook as a marketing channel, but make business decisions using all the available information, including real numbers based on actual business accounting (true nCPA).

Now to get to the core question of how to determine success on facebook and give good direction to the media buyer?

Typically, true nCPA and facebook attributed nCPA are HIGHLY correlated, but rarely the same. Most brands allocate between 70%-90% of their spend on Facebook, and it is reasonable to assume that it is also mostly responsible for your NC acquisition.

Let's use a real world example with some round numbers.

Assume that you have established that for your brand you need to meet a bare minimum nCPA of $50

  1. If Facebook nCPA is < $50 , assuming you are 80%+ spend on Facebook, your true nCPA should be LOWER than facebook attributed nCPA (due to "halo effect") and the decision is easy. Scale scale scale
  2. If Facebook nCPA is < $50 and your true nCPA is HIGHER than facebook attributed nCPA -- it means your other channels are performing really poorly (worse than facebook) and dragging up your nCPA and you need to be looking at optimizing those channels, but you should be scaling facebook until nCPA breaches $50
  3. If Facebook nCPA is > $50 (let's assume FB attributed nCPA is $60) - you need to look at your true nCPA and weigh the decision more carefully at the following:
    • If true nCPA is also > $50 -- you need to work on creative / ads / reducing true nCPA further, unless you are willing to spend more to acquire a NC
    • if true nCPA is < $50 - you will have a decision to make, based on how much lower true nCPA is that your $50 target - and this will be different for every brand
    • here is where it will help to establish the correlation between true nCPA and fb attributed nCPA -- and see if that correlation is relatively consistent by monitoring over time
      • if true nCPA is $40, but fb attributed nCPA is $60
      • you have a differential of $20
      • true nCPA is telling you that you are $10 under target (things are going well)
      • Now if you are running the vast majority of your budget on Facebook, the question I would be asking myself is
        • if I know my true nCPA is $40, but I spend all or most of my marketing dollar on FB -- then where are these NC really coming from? SEO? Brand Recognition? or most likely, FB as unattributed NC ("halo effect")
        • if that is the case, it should make sense to scale up your campaigns as the BUSINESS is making money. (these are business decisions, not marketing channel decisions)
        • Watch your true nCPA and how it correlates to fb attributed nCPA as you scale.
          • does the correlation remain intact while you scale?
          • are you still hitting your targets of $50 true nCPA as you scale?
          • can I continue to scale and is facebook attributed nCPA maintaining or increasing?
  • the lower true nCPA is under your $50 target, the more confidence you should have that everything is working, and you should press the gas
  • the closer true nCPA is to your $50 target is when you ease off and work the marketing / targeting / creative - killing or reworking the campaigns with the highest facebook attributed nCPA

For the media buyer - facebook attributed nCPA is highly relevant (and typically highly correlated to true nCPA)

They should be using it for the following:

  1. How is this campaign / ad performing relative to our established facebook attributed nCPA?
  2. Should I kill or keep this campaign? Does it need revision / retargeting?
  3. is the campaign truly accretive to the brand (acquiring NC) or is it just retargeting RC (purchases)?
  4. the farther away a campaign facebook attributed nCPA is from your baseline facebook attributed nCPA -- the worse it is performing to acquire NC
  5. campaigns / set out-performing baseline facebook attributed nCPA should be broadened and locked in as winners for scaling

This combination approach is a good scaffolding for developing your own decision making process as it relates to your business-specifics.

Facebook is just a marketing channel, but your brand is a business and business decisions on where to allocate spend require more contextual information than simply looking at what the marketing channel is attributing.

Interested to hear what the community thinks about this. Please Share if valuable Hopefully this sparks a good discussion.