How to Calculate Marginal CAC (Part 1)

Ok. It’s time. Time to do a deep dive into Marginal CAC (the marginal cost to acquire a new customer). Or thought of differently, a measure of how effectively the least effective dollars of our marketing budget are operating so we can decide when and where to modify our marketing spend to maximize our growth within the confines of our long-term objectives - most notably, long-term profit growth.

As we started digging into this topic, we realized that there’s a ton of depth here so we’re breaking our post into (at least) 2 parts. This will be part 1 of our exploration.

So buckle up, it’s about to get mathy.

Why do we care about Marginal CAC in the first place?

We’ve discussed this in a previous post, but briefly, things like Marginal CAC matter because we don’t want to waste our precious resources making suboptimal capital allocation decisions. Rather, we want to maximize the number of new customers we acquire in such a way that every single customer acquired is a profitable allocation of capital. When we’re using blended CACs for instance, we might miss opportunities to increase investment in a particular campaign or channel because we’re above our blended CAC, but our marginal CAC is well below our threshold for incremental long-term profit generation. And, since we care deeply about growing our business in the absolute best way possible, it behooves us to become intimately familiar with what’s happening at the margin.

Marginal CAC isn’t the Whole Story

There are many ways to think about Marginal CAC - but what I want to address quickly up front (that will inform the rest of the context here) is that Marginal CAC is only ½ of the equation for truly assessing how much incremental marketing dollars you should be allocating to any given campaign, channel, tactic, etc.

The other, extremely vital, piece of the equation is the REST of the cost to actually sell that item. Your shipping fees, your product costs, etc. And this number should be, just like our CAC, calculated on a marginal basis.

Some examples of reasons for this are:

1. Your campaign could be driving great acquisition but on a product that has an extremely low gross margin (big sales and promotions are often a culprit here, as are those products that are the best “deal” on your site). Thus, if you’ve built your understanding of Marginal CAC off of a blended unit economic model (average Gross Margin across your whole business), all of a sudden you’re underwater on your contribution margins because of this cost variability.

2. Your product costs changed in the past month because of a change in freight rates, manufacturing costs, etc. - so your unit economic model for the year is no longer a true accounting of your marginal profitability on each new order.

3. Your shipping costs are higher than your blended average - this can happen on larger-than-average items, or for discounted or higher priced items (usually folks model shipping rate as a % of the order revenue, but if the revenue from an order changes while the shipping rate stays the same, you’ll have variability in cost). Alternately, if you start pushing packs, for instance, and a materially larger % of orders hit a new shipping rate tier due to the increased weight or box size, you might also see this issue.

So, when thinking about Marginal CAC, to me it’s most useful to think - instead - about Marginal Contribution, or Marginal Variable Profit, from your incremental marketing spend.

Zooming in on Marginal CAC

That said, Marginal CAC is a big piece of this equation, so that’s what we’ll focus on today - but note that we’ll also want to discuss Marginal COGS, Marginal Fulfillment costs, etc. in a future post.

So, let’s dive in - how the heck should we think about calculating marginal CAC?

There are many ways and it depends on your tools and how much you demand perfection versus direction.

Note: at the end of the day, as with all things in capital M Marketing, we can’t get to the exact number for today, this week, this month, etc. However, we can be accurate enough to make confident, probabilistic, decisions - and any attempts to get closer to a true marginal CAC are well worth it in order to build your business for profitable long-term growth.

And now, without further ado - let’s get into the…

Methods to Calculate Marginal CAC

1) The Extremely-Basic, Anyone-Can-Calculate-and-Decide-What-to-Do-With-the-Data, But-it-Can-Sometimes-Cause-Head-Scratching, Method

The core question we’re asking when we ask about Marginal CAC is - with the extra money I allocated into my marketing stack, how many new customers did those extra $$ acquire?

So, a very simple calculus is to simply compare the amount of spending increase against the amount of new customer increase.

Thus, the equation, where S = media spend on acquisition, and N = new customers acquired, for any time period where you’re attempting to calculate Marginal CAC:

Equation 1
∆S / ∆N = Marginal CAC

where ∆ represents the change in the given variable period-over-period

Pretty basic, yeah?

And, for the most part, is much better than not paying attention to this number. It will give your marketing team much more direction and help you hone in on your true Marginal CAC into which you can spend and operate comfortably and confidently.

However, it has glaring issues and should not simply be taken at face value. It has three critical issues:

1) First, it assumes that all of your new customer acquisition is driven by marketing spend. This assumption is dangerous and very, very wrong (as anyone who started a brand without any money or marketing budget can tell you).

2) Second, it assumes that “today” (the current time period), is operating exactly the same as “yesterday” - from a cost of media perspective, a per session value perspective, creative performance perspective, seasonality perspective, etc. If any of those variables change period over period, the diminishing returns curve of your media will also change, which changes how we need to think about Marginal CAC.

A case study.

Let’s say I’m currently spending $100k/week in acquisition spend, and I’m driving 1k new customers for a blended CAC of $100.

Then, I increase my spend to $105k/week, and I see an explosion up to 2k new customers!

Whoa! My $5k spent was incredibly effective, for a marginal CAC of $5!! Wahoo! 10x budgets!

Of course, the reason for your expansion to 2k new customers wasn’t just the extra $5k you spent. Instead, it was likely a whole gamut of things - per session value increases, better ad creative, changes in organic traffic and its conversion rate, etc. - that improved the blended performance of the existing $100k that you were already spending such that the $5k was a cherry on top. This often happens when you’ve kicked off a sale, launched a new product, launched a new feature, launched new ad creative, etc.

3) Finally, we need a mental model to account for what happens when spend decreases and/or our Marginal CAC goes negative. For example, if we cut budgets by 10% and customer acquisition declines by 50% - how should we interpret that? Was that a good move and we should cut even more because our business is underperforming, or was that a bad move because that 10% budget reduction actually caused the decrease in performance? Furthermore, what happens if we cut budgets but customer acquisition actually increases?

Clearly, this method is quite basic, so let’s evolve our model to the next-most-basic way to calculate Marginal CAC to start to account for some of the above issues.

2) The Slightly Less Basic, but Still Incomplete, Method

The next easiest way to deal with the above problem is to take “organic” new customer acquisition out of the equation entirely and evolve our approach to marginal CAC to focus explicitly on our paid channels such that major growth in organic volume doesn’t affect our model more than it should (because naturally some paid media results in organic purchases).

Here, most businesses we find are in a few various states of evolution when it comes to assessing their paid media performance:

• Don’t have attribution software in place and rely on platform reported, last click web data, and post-purchase survey data to come to a general understanding of their marketing performance and paid media performance

• Do have attribution software in place and/or use a single tool as the source of truth (maybe with some multipliers attached) of actual attribution of performance to media spend - with plenty of understanding of the frailties of attribution in general

• Do have attribution software in place and use multiple tools as their source of truth of actual attribution of performance to media spend, with plenty of understanding of the frailties of attribution in general

• Have an oracle tell them exactly how their media is performing to the last dollar spent

In each scenario, you have a team of folks (or an oracle) trying to assess the actual performance of your media spend. The big difference is the tooling you’re using to come to your conclusions, but at the end of the day, every business has a team of people who are trying to figure this out.

So, let’s have them build a model of how much of this week’s revenue we want to attribute to paid sources vs. organic and use that as our determination of what customers and revenue to attribute to “paid” channels.

If you don’t have any attribution in place, this likely looks like reviewing your source/medium report in Google Analytics and simply applying different %’s of the traffic or revenue that you’d ascribe to paid media, and revising that on a weekly basis given changes in those traffic sources, knowledge of your marketing calendar, etc.

For example, for traffic that shows up as last click attributed to your Facebook ads, you’d maybe assess 100% of that as paid traffic. For your traffic coming in through organic search, you might assess something like 30% of that in any given baseline week to paid. For email, maybe it’s 25%, and so on.

If this is your baseline model, but in a given week you send out an absolute banger of an email (major sale/promotion), maybe you’d revise that week’s email paid % downward to account for the massive growth out of that channel. Or if you kicked off a national TV campaign and saw your organic search traffic/revenue grow in accordance, you could attribute a higher % to paid.

Yes, it’s hand-wavy. Yes, it’s not perfect. But it sure as heck is better than not doing this exercise. Intuition gets a bad rap these days, but I’m a strong proponent of using data + intuition to build a testable mental model that is incrementally better than not doing this exercise.

As you build this model, you then can neatly attribute PAID new customer acquisition and UNPAID new customer acquisition. Thus your Marginal CAC can focus explicitly on the PAID side of new customer acquisition as that is what your marketing dollars are explicitly impacting.

To revise our equation let S = total acquisition spend in a time period, and N_p = Paid New Customers Acquired in a time period.

Equation 2
∆S / ∆N_p = Paid Marginal CAC

Awesome - now we’re much closer! And, again, this one is pretty easy to compute and you’re already doing all of the attribution thinking anyway - this is a great opportunity to put that work to use, and it’s much better than the slightly-more-basic version.

And, a little value-added bonus here for any folks who are doing a deeper dive into attribution and actually building their attribution models out to the channel or campaign level (which is likely most of you) - you can copy-paste the formula above, but just with that attributed view of customer acquisition. Thus, the equation becomes:

Equation 3
∆S_c / ∆N_c = Campaign Marginal CAC

Where S_c is spend in the campaign and N_c is the number of customers acquired by the campaign based on your attribution methodology

Now, at this point, you may be thinking - “wait a minute - isn’t our organic revenue today just the output of our media spend in the past???” Definitely, to a certain extent. Like I said, hand-wavy. But that’s our line of work. If you wanted specifics, you shoulda gone with accounting.

Note: it’s compelling to overestimate the work your historical paid media is doing today without any way to measure it. I would caution against that. Again, as anyone who’s been forced to build a brand where they literally couldn’t spend money on media can tell you - your organic work can go a long way. The conversations people have on the street. The serendipitous interactions with a stranger wearing a sweet new pair of shorts that you’ve never seen before. The organic social content that gets shared between friends. The emails that get forwarded. That “media” does a huge amount of work for you - much more than the lingering memory of a 2 second interaction with a Facebook ad ever would. However, it’s still an issue, and to account for it, you can decide how you want to set your Marginal CAC threshold to allow for some performance of media over the next 30, 60, 90 (pick your time period) days.

Ok great - so now we’ve adequately resolved critical issue number 1. High fives all around!!

But. I have some bad news my dear reader - critical issue number 1 was the easy issue to resolve.

We’ve still got critical issues #2 and #3 to deal with. Because even with the modification to remove unpaid customer acquisition from our analysis, clearly we’re still not taking into account the other variables that are nearly always different when comparing period-over-period like CPMs, CTRs, Conversion rates, etc. AND we’ve not yet built our mental model for spend decreases and negative Marginal CACs - so we can still get better.

And it’s critical issues #2 and #3 that are the real doozies.

However - FEAR NOT. For we have a couple of more methods for your testing and feedback that incorporate and understand these differences between time periods with varying degrees of complexity and sophistication, and also a few frameworks for how to think about spend decreases/negative marginal CACs.

And, since resolving those issues is - as noted - a real doozy, it’s going to require at least one more post to do it justice. We’re already working on the post, but it might take a couple of weeks so hang tight!

In the meantime - if you don’t already have a method for calculating and evaluating marginal CAC please test out the above methods - particularly method 2, and even more particularly using Equation 3 to math it out to the campaign or tactic level - and let us know what you think.

Some questions to direct your feedback:

• Where did you find more hang-ups? Where did it create unnecessary complexity?

• Did it help you find more profit? Did it not?

• Was it easy to message to your team? How easily did they adapt to this paradigm?

• How frequently are you using this calculation?

• How does this theoretically fit with the way you think growth teams should operate? How does this practically fit with the way you think growth teams should operate?

• How did you think about a negative Marginal CAC? What nuance did you uncover as you built your own mental model around that?

Please let us know any and all questions, thoughts, concerns, praise, etc. We’ve got a ton of extremely intelligent folks on this newsletter and would absolutely love it if we can all learn together, so, please, no touch too light in sharing any sort of feedback whatsoever.

On that note, I got some feedback from some of the Bodacious readers that it wasn’t super clear how to actually pass any sort of feedback along (they sent me a carrier pigeon), so I wanted to offer up a few methods:

1) respond directly to this email with any thoughts, notes, etc.

2) send an email directly to Tom ([email protected]) or Preston ([email protected])

3) respond to either of our Linkedin posts either in DMs or in the comments section

Looking forward to your feedback.

Now let’s get back to building.

Cheers,

P & T

—
Thanks for reading. If this post was helpful, here are 4 things you can do right now to 

• Get more strategic + tactical nuggets on brand building, 

• Tactics on connecting brand building to financial impact, 

• And specific things you can do to build strong emotional connections with your audience: 

1) Subscribe to the Brand Builders podcast on YouTube or Spotify, and Apple Podcasts. 
2) Apply to join our brand builders slack community (suuuuper small group exclusively for brand founders and operators), 
3) Follow Tom and Preston on Linkedin for regular posts on this stuff
4) And heck, share this with someone