LTV:CAC deep dive part 1: Misconceptions

In the first part we discuss misconceptions about the metric that lead to bad decision making and consequently bad investments

Introduction

LTV:CAC (lifetime value to customer acquisition cost ratio) is an extremely popular marketing efficiency metric. Despite its popularity though, it can get confusing when it comes to understanding its variations, calculations and interpretations. So in this multi-part post I will discuss from first principles the following topics:

1. Misconceptions

2. Measurement framework

3. Optimal LTV:CAC ratio

4. LTV:CAC & marketing spend optimization

Misconceptions

I have supported the proposition that ROI is the marketing efficiency KPI you should be focusing on. Why am I talking about LTV:CAC then? Well…

While ROI is a more generic metric (and a percentage instead of a ratio), within the context of marketing they should be the same thing, although most times they are not.

The reason? Wrong attribution.

While in the second part of the series I will discuss attribution as part of a measurement framework, let’s see why this metric can be used in a way that produces misleading results.

When you make an investment, you would like to know the benefits that came out of it. So the components in the ROI calculation (revenue, costs) should be causally connected, i.e. they were caused by the same activity.

Example 1

I bought a house which I rent and had a x% return within y years. The returns wouldn’t be there if I didn’t buy the house in the first place. You cannot say “I bought a house and I made some money from selling stocks so my ROI from the house is:”

This claim is misleading because it makes the investment look profitable when it might not be. I am bringing this up because this is how many businesses use LTV:CAC. They include in CAC costs related to marketing activities but then they spread these costs across a user base that is not causally related to this activity, therefore underestimating CAC.

Before we go on with another example, let’s look at the generic CAC formula that one can find with a quick search online:

Example 2

Let’s assume you have the following data on your performance marketing costs along with paid (performance marketing) and total (performance marketing + organic) acquisitions on monthly level:

month	cost	paid acquisitions	total acquisitions	cheap CAC	expensive CAC
Jan	$1,000	200	1,000	$5	$1,000
Feb	$2,000	350	1,001	$5.7	$1,000
Mar	$0	0	999	$0	$0

The cheap CAC is the version businesses usually go for. Here we divide total marketing costs with paid acquisitions as in the equation above.

But is this correct?

To answer this question let’s now focus on total acquisitions. If paid acquisitions went up from January to February and then down on March, we should be able to observe similar movements in the total acquisitions since they include paid acquisitions as well.

The March record shows us how many acquisitions we would get without any performance marketing spend. So you spent $1,000 on January and you got an additional acquisition compared to March. But it could be statistical error. Perhaps organic acquisitions happened to drop by the exact amount during the same time period. Then let’s compare January to February. The same pattern applies there. An additional $1k brought another acquisition. Can this be another coincidence? Most probably not.

The reality is that 2 customers were acquired for a total of $2k.

This cannibalisation effect is caused by last-click attribution. Here organic traffic (or unknown source traffic) is basically renamed to paid. This happens when a channel is so close to the final conversion event that doesn’t bring any new traffic but instead gets the credit for all traceable or untraceable channels before that. This phenomenon appears frequently in search engine advertising.

So what ends up happening is the same mistake we presented in Example 1. CAC is biased because the cost and the acquisitions are not causally connected by the same root cause.

So if the expensive and accurate CAC is what you want, then you need to adopt a more holistic view of the customers journey before reaching your door. This requires better attribution, use of causal inference methods and of course surveys. On the other hand, the cheap CAC can be calculated at any given point in time, with lower experimentation costs and lower measurement costs and this is why it ends up being the more popular, albeit wrong approach.

Next Up…

In part 2 I will discuss metric variations along with practical considerations and recipes for estimating LTV:CAC with a more holistic and unbiased approach.

Stay tuned!