LTV, aka Lifetime Value, aka Customer Lifetime Value (CLV) is an indicator of the customer value for the whole lifetime period. It shows how much money one user will bring in an average for all the time of using the product. LTV indicator is universal, it is calculated both in web analytics and mobile analytics. He is calculated for most kinds of products, whether it's Starbucks coffee houses, mobile operators, banks, SaaS-products or games.
In this article, we are going to talk about what you need to know LTV for and how to use it. It is necessary to make a reservation that Lifetime Value indicator, calculated for all user base is somewhat of a Spherical Cow. It can be used, but in order to have a more accurate result for LTV indicator, it should be calculated in separate cuts. What are those cuts, you may read below.
LTV > CPI
This is the main formula of all traffic analysis and the main condition for the effectiveness of acquisition. The user should bring more money than was spent on his acquisition.
By CPI (Cost Per Install) in this case, we mean the average cost of acquiring a single user across all channels. If you are more familiar with the abbreviation CPA (Cost Per Acquisition), traditional to web-based products, use it for the further formulas in this article.
In fact, the average CPI or average CPA - are pretty much conditional indicators, as we usually pay one partner the A amount of money, the other - the B amount of money, the third - the C amount of money, while the overall average CPI - is likely to value that is not equal to A, B, or C. LTV is best to be calculated separately for acquisition channels, by campaign, and so we come to the following use of this indicator.
Evaluation of the quality of the traffic source
Each source has its own acquisition price (CPI and CPA) and its own traffic quality, and, therefore, its own LTV. This is why it is more efficient to calculate LTV for each channel separately.
By doing this you may get an overall average LTV greater than the total average CPI, but in the cut of acquisition channels, you would be able to see the ineffective channels, where this condition is not met. What should be done in such a situation? You may, certainly, immediately turn off the traffic channel that fell out of favor. However, it would be more efficient to do a detailed study of it, in the cuts of the campaigns, countries and platforms, and turn off those where LTV is less than CPI. And even better - to introduce a similar analysis in regular practice and turn off the underperforming SubIDs attributed to the traffic channel.
Metrics are used by analysts, and the money comes from owners and investors. And for these serious people, it is important to know whether their investment will pay off. For this purpose, the ROI (Return On Investment) metric was invented, which takes into account both LTV and the cost of acquisition.
ROI might be calculated in different ways though now we are talking about the formula ROI = LTV / CPI * 100%. According to the results of calculations, ROI must be greater than 100%.
We recommend also count ROI for certain fixed time intervals from the time of registration (first entry) of the user: ROI N days = Cumulative ARPU N days / CPI * 100%.
Here we introduce a new metric Cumulative ARPU N days, which shows how much money on the average was brought by one user in the first N days of using the product. By selecting different N values, you will better understand the dynamics of ROI and will be able to calculate another important indicator. Namely…
When will the money invested in the project pay off? Look at the chart:
The blue line is an indicator of Cumulative ARPU, it shows how much money on the average is brought by one user in N days of using the product. LTV indicator - is the limit of Cumulative ARPU when N tends to infinity (although, in practice, usually fixed values of N such as 120, 180, 360 days are taken).
If the business works well, and the traffic pays off, so there is a point T, where the blue line (the money brought by the user) is higher than the green (money spent per user). And the day when such an important event fires is called the payback period. Now you can tell the owner exactly when the money invested will pay off and when ROI will exceed 100%.
Planning the costs
Let’s go back to the main formula: LTV > CPI.
In the calculations, it is important to know about the concept of net LTV, ie LTV net other costs: the commission of the store, the commission of the publisher and royalties, and also taxes.
With CPI, everything is not as easy also. To start buying traffic, you must first negotiate and agree on terms and conditions (management salary), then sign a contract (lawyers salary), then integrate (programmers salary), and that we do not yet take into account the fixed fee when signing for some channels. So from CPI, we move on to the effective price of acquisition, eCPI (by analogy with an effective bank rate).
Typically, in a project, there is also the cost of running the user's activity - technical support, community management, server, and others.
The final formula takes the following form: Net LTV> CPI + costs for 1 user (variables, constants).
It follows that the costs should be planned in such a way that the condition is performed, after the deduction of all commissions from LTV and addition of all the costs to CPI.
The dynamics of the project
LTV is based on the value of the many metrics. It is influenced by retaining users (retention), by the proportion of paying users (Paying Share), and by the revenue of paying user (ARPPU). Instead of tracking the dynamics of multiple metrics, you may follow the dynamics of LTV, - it will show you how effective are the changes that you are making in the project.
If LTV is growing from month to month - well, keep up the good work. If it falls (as the majority of projects has a downward trend of LTV on the time axis) - it's time to take action and improve the project.
Forecasting future revenue
If you know how to predict LTV, and even calculate it in the cut of channels, countries, platforms, etc., then you may well be able to predict how much money you will earn in N months. For example, you would be able to answer the following questions:
- what will happen to the revenue in 3 months, if we reduce the paid traffic by 50%;
- if we release at the market of a new country in April, how much money will we get from it by the end of the year;
- if we make a change in the project, that increases the retention of users by 3%, how will it affect our revenue;
- when will the traffic that we purchased from partner X pay off;
- and so on.
As you may see, LTV is the most important indicator in the project analytics. But there is one issue: it takes a time to calculate it, and as a rule, there is always lack of time. If you calculate LTV for the short term, the forecast will not be the most accurate. If you make a calculation over an extended period, then the question of prediction ceases to be irrelevant as the future catches you up.
So how should LTV be calculated?
This is a question that is not answered in one sentence. For example, the screenshot about the calculation of LTV from the book "Database Marketing: analyzing and managing customers" (good and powerful book on analytics and marketing, if you like hardcore):
There are many ways of calculating LTV, and among them, there are those that allow you to get an accurate forecast for a few days. So that you could better understand in practice how to use this indicator, we invite you to a free webinar. It will take place February 2, 2016, 9:00 AM - 10:00 AM PST. I will conduct a review of methods for calculating LTV and select the most effective, as well as consider a few questions on forecasting and usage of this metric in practice. Sign up!
P.S. And, yes, at the webinar there would be no such hardcore as seen in the last screenshot :)