- LTV is especially useful when compared with Customer Acquisition Cost (CAC);
- You can model LTV (even with no data);
- There are a few key formulae to get to grips with.
What is customer lifetime value?
Customer Lifetime Value, commonly referred to as LTV, is a very important business metric that sits outside standard financial reporting. LTV, in essence, tries to show how much every customer will be worth to you over the course of their lifetime with your business. The use of the verb ‘try’ here is intentional as almost always LTV is a arithmetically modelled calculation and hence will usually not be 100% accurate.
Why is customer lifetime value so important?
LTV is important when coupled with another metric, customer acquisition cost (CAC). These two metrics sit on opposite sides of a theoretical see-saw and jostle against each other to determine the success of your business. In very simple terms you can have three different scenarios:
- CAC > LTV aka Business Failure
In this scenario you are paying more money to acquire your customers than they are delivering you over their lifetime with your business. In this scenario the more customers you acquire the more money you will lose. Also, the faster you acquire customers, the faster you will run out of money.
- CAC = LTV aka Business Stagnation
In this scenario you are paying the same amount to acquire a customer as they are paying you back over their lifetime with your business. In this scenario your business is essentially flatlining. Bear in mind that from a cashflow perspective you’ll be negative for the same time as the customer’s lifetime with your business as it will take them their whole lifetime to repay the initial spend.
- CAC < LTV aka Business Growth
In this scenario you are repaying your CAC over the course of your customer’s lifetime with your business but crucially you are also generating additional revenue. In this scenario your business will grow and you should continue to drive further customer acquisition. If the equation holds true you should accelerate your customer acquisition as that will grow your business faster. Many entrepreneurs either overestimate LTV or underestimate CAC, which leads David Skok from Matrix Partners to assert that:
the second biggest cause of startup failure: the cost of acquiring customers turns out to be higher than expected, and exceeds the ability to monetize those customers
In order for your business to not face a LTV problem going forward, you need to try to model LTV.
How do you go about modelling LTV?
There are differing approaches, dependent on how long you’ve been trading and how much data you have.
I don’t have any data.
If you have no data, then the only choice you have is to base your LTV calculations on a lot of assumptions. Try not to be overzealous with these assumptions as you will most likely need to justify them when speaking to potential investors. First we need to define some LTV constants.
- s = average spend per booking
- c = average number of purchases per year
- a = average (gross) customer value per year
- t = average customer lifespan (years)
- r = customer retention rate (what % of customers this year will be customers next year?)
- p = average margin per customer %
- i = rate of discount*
- m = average gross margin per customer lifespan (a*t*p)
*rate of discount is a complex and abstract subject in its own right. A good, concise explanation can be found on the data.gov.uk website.
Unknown Discount rate, very basically, tries to model the fact that money is worth more to you now than it is in the future. For example, if you used a discount rate of 3.5% then £150 you receive in 5 years time is actually only worth the equivalent of £126 now. This £126 is the net present value of £150 in the future. This becomes important when modelling LTV as predicting a future value doesn’t tally correctly against a present CAC. In order to be able to accurately compare CAC against LTV you need to use the net present value of LTV. Using the example above, you could mislead yourself by saying that you could afford to spend £150 on CAC to break even as that is the LTV figure. However the net present value of that £150 is £126, so actually you’re £24 in the red. It’s a nuance, but important.
Using the constants in the table above we can then build up some LTV formulas:
|LTV 1||a * t * p|
|LTV 2||t * s * c * p|
|LTV 3||m * [ r / ( 1 - r + i ) ]|
|LTV 4||m * [ 1 / ( 1 - r ) ]|
You can use this Excel spreadsheet template to do all the legwork for you.
How do I use these results?
LTV is a projected figure and is not going to be accurate. With any type of modelling the further into the future you go, the less robust your results become so be wary modelling over five years (especially with no or very limited data). As you may have noticed the equations will, most likely, have given you a number of different results. One method to deal with these differing results it to average them all out. This is a method that has been endorsed by analytics guru Avinash Kaushik, however it’s one I disagree with. Modelling produces error and by simply averaging a lot of erroneous values together you’re not really solving anything, in fact you’re just losing the clarity of where the error might have come from. If you stick with one of the equations above you should be able to explain how you got to your end result, what the assumptions are, and where the technique falls over. That’s a stronger and more justifiable position than just averaging all of them out.
I do have some data.
If you have some data you’re already in a much stronger modelling position than if you didn’t have data, and now the key is to get the data to work as hard as it can for you. At the crux of any LTV modelling in a position when you have data is cohort analysis. Cohort analysis studies the behaviour of groups of customers over time. Your customers would normally be grouped into acquisition month and then studied over their time with your business. The first step in any cohort analysis is to structure your data in a way that facilitates easy pivoting. This will save you a lot of time going forward when you do this on a monthly (or more frequent) basis. My recommendation is to format your data like this:
|1st Purchase Month||Transaction Month||Customers||Revenue|
The example above is for a business with 3 months of data. Month 1 in January 2015 (2015-01) and month 3 in March 2015 (2015-03). With the data in this four-column format you can create a pivot table which will create a revenue array. You can see this demonstrated in this spreadsheet. 1st purchase month is what we call a cohort. Once you have your array, you can begin to accumulate revenue for every cohort you have. Your data could look something like this:
You should also know how many people transacted in their first month with you:
With these two data sets you can now start to tabulate customer lifetime revenue evolution. Divide the cohort revenue by the customer number for each cohort: You should also know the amount of money you spent acquiring these customers, this data could look like this:
Combine this cost data with the number of customers in that initial cohort to get the Customer Acquisition Cost (CAC):
Now use your previously calculated cohort LTV alongside the CAC to understand your payback amount and Return On Investment (ROI):
An evolving LTV positive business will be able to chart their data so it looks something like this:
This chart shows more recent cohorts having negative ROI, whereas cohorts from ‘m-10’ onwards have positive ROI. If this trend continues this business, in a LTV sense, is doing ok. The key for this business is to try and work out how to pull the ROI breakeven time (or Payback Time) closer to the cohort start time. Doing this will expedite the business’ success.