LTV Series

CAC payback period vs LTV / CAC

Which campaign optimization strategy will win?

In the last few posts, I explored the [CAC payback period] from different perspectives. I researched the relationship between the [CAC payback period] and the other metrics. Also, I reviewed Lenny’s [CAC payback period] benchmarks for the B2C business model.

Today it’s time to summarize all my previous findings.

Let’s start with the fact, that there are two camps, each of which believes in the superiority of one approach over another:

• [CAC payback period] fans are big believers in velocity. They argue that LTV projection for a long time horizon is a very questionable task and that’s why they focus on a much shorter period (period of payback) as a much more doable task that requires fewer assumptions.
• [LTV / CAC] fans are big believers in capital efficiency. They admit that LTV projection is a challenging task but at the same time, they know that there is no guarantee that campaigns with short payback periods will generate much more revenue in the future.

I must admit that depending on the situation, each of the arguments above could sound very persuasive.

To determine which approach is better and, more importantly, in which cases, let’s try to compare these strategies.

To compare these [CAC payback period] benchmarks I defined similar by meaning benchmarks for the [LTV / CAC] metric. Then, I calculated the same set of metrics for both optimization strategies and put them into tables with the same structure.

[CAC payback period] and [LTV / CAC] optimization strategies with corresponding metrics look like this:

Two optimization strategies: CAC payback period vs LTV / CAC.

Important caveat: all figures above are actual, not projected. It means that we put the question of LTV projection precision out of brackets.

My CMO experience tells me to start analyzing from the bottom of lists.

The logic is quite straightforward: if I manage to figure out quickly which campaigns have a long CAC payback period and, in fact, they don't work, I can stop them and reinvest their budget into campaigns from the top of the list. This move will not require approval from the CEO and will give me some room for improvement.

So, let’s start with LOSSES campaigns (the bottom line of each table):

• if I stop LOSSES campaigns in the [CAC payback period] strategy I will free a budget of $17,680 (39%). This move will lead to the following consequences: (a) I will not generate revenue of $10,002 (15%) and (b) will not acquire 162 buyers (32%).
• if I stop LOSSES campaigns in the [LTV / CAC] strategy I will free a budget of $7,381 (16%). This move will lead to other consequences: (a) I will not generate revenue of $1,518 (2%) and (b) will not acquire 47 buyers (9%).

Firstly, while the [CAC payback period] strategy will enable us to free a big portion of the budget, this will heavily impact our customer base growth and noticeably impact revenue. Are we ready for this scenario?

Secondly, an attentive reader might have noticed that 22 LOSSES campaigns from the [CAC payback period] strategy are the sum of 10 LOSSES and 12 BAD campaigns from the [LTV / CAC] strategy.

Turning off 10 LOSSES campaigns with an average CAC = $7,381 / 47 ~ $157 makes a lot of sense taking into account that the average CAC of the whole Paid channel is $44,912 / 500 ~ $90. Actually, these campaigns generate 76% of all losses.

At the same time, turning off 12 BAD campaigns with an average CAC = $10,299 / 115 ~ $90 doesn’t make that much sense as this CAC is equal to the average CAC of the whole Paid. It means, that while users from these campaigns are slow payers, however, we haven’t overpaid for them.

Moreover, if I focus on CRM marketing and improve # of transactions and/or ARPU of these buyers, I will be able to partially improve their LTV over time.

To summarize, the [LTV / CAC] optimization strategy is much better at distinguishing campaigns that generate huge losses from campaigns that generate manageable losses. [CAC payback period] strategy is not capable of handling this task.

Now let’s focus on Exceptional campaigns (the top line of each table).

There is a big difference in number of these campaigns per optimization strategy:

• [CAC payback period] strategy has only 2 such campaigns. It’s just 3% of all campaigns.
• [LTV / CAC] strategy has 8 such campaigns. It’s 13% of all campaigns.

Obviously, having only 2 campaigns is not enough to train Google/Facebook Ads optimization engines. The same issue will arise if you would like to train an ML model to predict such campaigns.

So, focusing on Exceptional campaigns in the [CAC payback period] strategy is not practical.

Finally, we come to the most important part of this research — high-value campaigns with a short CAC payback period.

Relationship between LTV / CAC and CAC payback period.

As you might have noticed on the scatter plot above, the CAC payback period could merge campaigns that contain users with very different spending power.

We have already discussed above the situation with 2 clusters of campaigns with very long CAC payback periods.

Let’s talk a bit about campaigns with short CAC payback periods.

There are 22 campaigns with very short [CAC payback period] ≤ 3. Supper! The issue is that they are distributed in 4 segments: OK, GOOD, GREAT, and Exceptional. The actual [LTV / CAC] spread is from 1 up to 7!

In fact, the [CAC payback period] has only part of the information about user spending power. There is no way to figure out how committed the users are to the product from these campaigns in the long run.

The chart above brings me to the following conclusions:

• Knowing that the [CAC payback period] is in a range from 6 up to 30 periods (2 clusters in the middle) can be used as a rule of thumb that campaigns that have such a payback period will not generate great Return on Investment. It’s useful.
• Knowing that the [CAC payback period] ≥ 30 periods (2 clusters on the right side) is a bit useful. Sure, we can turn them off. But a better approach is to calculate [LTV / CAC] and separate campaigns with long CAC payback periods into two clusters: huge losses and manageable losses.
• Knowing that the [CAC payback period] ≤ 3 periods (1 cluster on the left side) is not insightful. It’s extremely important to calculate [LTV / CAC] and separate campaigns with short CAC payback periods into a few sub-clusters: ‘scale’, ‘adjust’, and ‘leave as is’.

SUMMARY

1. [CAC payback period] is a good rule of thumb for assessing which campaigns don’t have potential.
2. [CAC payback period] barely can help you prioritize which campaign can be safely turned off and which campaign should be scaled in cases when CAC payback periods are very long or very short.
3. For very short or very long CAC payback period campaigns you need to somehow project LTV to figure out what specifically to do with each of them. Thus, the ROI-based approach cannot be completely replaced.

In the next post, we will talk about behavioral signals and their relationship to LTV.