LTV Series
When do you need to use the LTV projection?
The 4 most common use cases of using LTV in marketing.
From time to time start-up communities talk about how important it is to forecast LTV (Life-Time Value). Some experts argue that knowing LTV will give you some advantages in performance marketing. Other experts warn you that LTV is just a projection and you should use it cautiously. In fact, both points of view are right.
In today’s post (it’s going to be series about LTV), I will touch on the 4 most common scenarios where using LTV could be beneficial:
- What’s the maximum cost we should pay to acquire one customer in a specific channel/campaign?
- Which channels/campaigns should we prioritize based on their LTV/CAC?
- Which user segments do we need to prioritize first in terms of win-back campaigns?
- Which offers does it make sense to have for which LTV user segments?
Let’s get started.
What’s the maximum cost we should pay to acquire one customer in a specific channel/campaign?
I’m not a big fan of projecting LTV to infinity periods as projection is unreliable then. That’s why let’s define first the time horizon we are interested in. In most cases, the time horizon from 12 months to 3 years is quite reasonable.
To show you the main ideas I have projected LTV for the 36 months. As the first few questions are related to Paid user acquisition, I subset users from the Paid channel.
The LTV projection for the Paid channel by period could look like this:
In 36 periods the projected LTV for the Paid channel = $130.4. By itself, the LTV projection is not very informative. It’s really difficult to say whether it’s a good or bad result.
To get some sense from LTV projection let’s add to the chart the Customer Acquisition Cost (CAC) metric, where the [CAC] = [Costs (of cohort)] / [# buyers (in a cohort)].
Now the chart will look like this:
From the chart above we can learn:
- From period 0 (signup period) till the 9th period CAC > LTV. This means that the CAC payback period = 9 months.
- Starting from 9th period LTV > CAC. It means that starting only from the 9th-period cohort will start earning money for the business.
It seems that having at hand only CAC or LTV is not as useful as contrasting them with each other.
One possible way to contrast costs and revenue is to calculate the metric [LTV / CAC]. This ratio can give us some sense of whether our marketing investment is successful.
There is common wisdom that the [LTV / CAC] ratio, on average, is in a range [2, 4], where the ideal scenario is 3 or higher. Why?
The logic behind this rule is the following:
- 1/3th of LTV will cover User Acquisition costs
- 1/3th of LTV will cover User Retention costs
- 1/3th of LTV will contribute to the Profit Margin
Even if we assume that we don’t have User Retention costs (but we always have them) then the breakdown will look like this:
- 1/2th of LTV should cover User Acquisition costs
- 1/2th of LTV should contribute to the Profit Margin
Based on the last breakdown we can easily come up with a CAC that we can afford: [CAC max] = INT( [LTV] * 0.5 ).
In our case [CAC max] for the Paid channel = $130.4 / 2 = $65.0.
In our case, if we manage to acquire new users with a cost of less than $65.0 then we have a great chance to make money. Otherwise, likely we can incur losses.
Can [CAC max] calculated for the Paid channel be used on the campaign level?
It’s always convenient to have one universal figure that can be used in different scenarios. Unfortunately, it happens very rarely.
Let’s calculate all the necessary metrics for each paid campaign within the Paid channel.
The summary table of paid campaigns could look like this:
From the paid campaigns summary table above we can learn that:
- The metric [LTV to Date] varies greatly from campaign to campaign (from $66.7 up to $204.4) while on the Paid channel level, it’s $130.4.
- Because of the [LTV to Date] the metric [max CAC] also varies greatly from campaign to campaign (from $33.0 up to $102.0) while on the Paid channel level, it’s $65.0.
So, if we simply use the overall Paid channel metrics:
- for some paid campaigns we will overpay for new users and incur losses
- for other paid campaigns we will underinvest in new users and will earn less than we could
This brings us to the following rule:
LTV (or any other metric) on the Paid channel level is a sort of average value. We should always calculate all corresponding metrics on the campaign level (or even more in-depth levels) and adjust campaigns based on their performance results.
Which channels/campaigns should we prioritize based on their LTV/CAC?
Since in our case we have only one Paid channel, let’s focus on answering the question related to campaign prioritization.
From the table above we have already learned that there are 12 paid campaigns. Frankly speaking, this figure is not big. In a real-case scenario, a company could have thousands of paid campaigns.
As paid campaign optimization is a non-stop process then to run it efficiently it makes sense to figure out which paid campaigns should be dealt with the first, which should be dealt with the second, and so on.
To prioritize our campaign list let’s sort it by the metric [LTV / CAC] and color groups using the rule { >2, [1–2], <1 }.
The colored campaign list could look like this:
There are 3 segments in the list:
- [LTV / CAC] < 1. This segment contains 2 paid campaigns that do not pay back at all. The reason for this is that these campaigns have [CAC] above the average and at the same time have [LTV to Date] below the average. Taking into account that there is a very tiny chance that these campaigns will have higher actual LTV, then it makes sense to shut them down.
- [LTV / CAC] > 2. This segment contains 3 paid campaigns that make sense to scale. Two of these campaigns have actual [CAC] that is lower than [max CPC]. We can scale them without worrying too much about [CAC] increase.
- 1 ≤ [LTV / CAC] ≤ 2. This segment contains 7 paid campaigns. As a rule, this is the biggest segment. Campaigns in this segment need to be heavily adjusted. As marketers have much more leverage to reduce CAC than to increase LTV, it makes sense to start from high to low LTV and then try to optimize costs from top to down.
There are some other approaches about how to prioritize paid campaigns. I will show you some of them in the following posts.
Which user segments do we need to prioritize first in terms of win-back campaigns?
As we have already seen, [LTV to Date] can vary a lot from campaign to campaign. Also, there is a big chance that different users within each campaign spend different amounts of money on our product.
Let’s calculate LTV percentiles { 20, 40, 60, 80, 100 }, and create LTV segments by aggregating payment stats for all users (from Paid, Organic, and Direct channels) within each segment.
When we would like to launch win-back campaigns it makes sense to segment buyers based on their payment stats (LTV and ATV):
- % of buyers from the segment pctl_1.00 is just 12%, but they bring us 73% of revenue. Each such buyer generates $715, which is 6x higher than the average buyer. If we lose even a few such buyers it will be noticeable for business.
- % of buyers from the segment pctl_0.20 is 41%, but they bring us just 2% of revenue. Each such buyer generates $5.3, which is 22x lower than the average buyer. Even if we lose a large % of such buyers the risk for business is very tiny.
As marketers are very busy with different stuff, I would recommend starting with the segments at the top of the list: if you don’t have time to contact users from the pctl_0.20 segment it’s OK, but you should do your best to keep regularly working with users from the pctl_1.00 segment.
Which offers does it make sense to have for which LTV user segments?
Another important moment is what to offer for our lost buyers. Depending on their ATV (Average Transaction Value) it makes sense to adjust win-back offers:
- for segment pctl_1.00 I would recommend an offer that after discounting will be higher than $50 (as ATV for this segment = $49.0). For example: order from $60 and get a 5% discount.
- for segment pctl_0.80 I would recommend an offer that after discounting will be higher than $30. For example: order from $40 and get a 10% discount.
- for segment pctl_0.60 I would recommend an offer that after discounting will be higher than $20. For example: order from $30 and get a 15% discount.
- for segment pctl_0.40 I would recommend an offer that after discounting will be higher than $10. For example: order from $15 and get a 15% discount.
Remember “1/3th of LTV will cover User Retention costs”? Discounts are paid from that source of money.
If we are talking about prioritization, I would recommend focusing at the top of the list as a few high-value segments are the top priority.
SUMMARY:
- Don’t use Paid channel metrics to assess the performance of paid campaigns as overall metrics can be very misleading.
- Try to prioritize campaigns based on a set of simple rules: shut down, scale, adjust.
- Don’t send one-offer-fits-all-size: segment buyers based on their LTV and create offers using segment payment stats (for example, by using ATV).