A GUIDE TO CLTV!
Applying Customer Lifetime Value
Understanding, Calculating & Predicting CLTV!
There are several Metrics to evaluate how a Business is performing, Marketers talk about must have measurable Metrics for optimized or enhanced performance in terms of Business, Advertising, Social Media effectiveness & working towards to achieve that goal and similar other things.
However, one of the important and most underrated Metric is Customer Lifetime Value (CLTV/CLV/LTV), which is often neglected by the Business.
- What is Customer Lifetime Value?
Customer Lifetime Value (CLTV/CLV/LTV) is a metric which allows a Business to see how much a paying Customer during their Lifetime might bring Revenue (Value) by spending on their Products & Services.
For an example: If a Customer purchased your Product & subscribes for some paid services provided on a Monthly basis, & renews regularly for a year or continues, & the Money paid during that period will determine his/her Lifetime Value.
2. Why is CLTV becomes such an Important Metric?
To be precise, CLTV tells you how much a Customer is worth to your Business, it helps you understand their overall value.
Marketing Teams, understands better than anyone that, acquiring New Customers is costlier than retaining existing Customers, so, by ascertaining CLTV, it helps you to make decision on how much to spend on Customer Acquisition & also on Customer Retention.
That being said, let’s get an understanding of Customer Acquisition Cost (CAC); it’s the cost of acquiring a Customer to Purchase a Product/Service.
3. How to Calculate CAC?
The simple method it divides the Total Marketing Costs to acquire New Customers by the Total number of Customers acquired for a defined period.
For an example: No Company would like to be unprofitable, by spending $1,000 on acquisition of a 100 New Customers, when expected CLTV is $900, since it would drain $100 of Value per Customer acquired.
Especially, CAC combined with CLTV becomes really an important Metric, particularly for SaaS Companies, even then, with the Advanced techniques like Machine Learning & it’s subset Deep Learning which is somewhat evolution of ML has made Predictions more accurate & improving every day, it becomes equally important to other Industries as well, to manage their expenses, see their growth, expand if Business allows.
4. Historical CLV Using Cohort Analysis to understand Customer Behavior:
Cohort analysis takes the ARPA approach one step further. Instead of calculating an overall average monthly revenue per user, Cohort Analysis calculates an ARPA per month per Cohort (a Cohort is a group of customers who share an attribute or set of attributes; in this case, a Cohort is defined as those who joined or made their first purchase in a particular month).
Overall, Cohort Analysis is part of Exploratory Data Analysis, which helps in understanding the Data Patterns & Customer Behavior.
5. How to Calculate CLTV? There are several ways to get it, let’s make it with two simple Methods:
A. First Method:
Also, before we get into on how to Calculate CLTV, we need to understand Avg. Revenue Per Account (ARPA), Churn Rate, Retention Rate for a Defined Period, which we can get during Cohort Analysis.
For example: If in a Cohort some 30 Customers generated an Avg. $15,000/Customer in Revenue for every recurring Month over a Time Period of 1 Year.
However, for 100 Customers if ARPA is $2,000 every recurring Month for a Time Period of 12 Months, out of which 30 Customers are New & rest are Existing, in the same last Period there were total 90 Customers with an ARPA of $2,200, so:
B. Second Method:
You are a Customer who does shopping in a Camera & Accessories Online Store which it specializes, provides reasonable discounts without compromising on quality of the products sold & services provided. Your Avg. Purchase Frequency is once every Quarter, Avg. Order Value is about $200 & your Avg. Customer Lifetime is three years. Using the formula, your CLTV would be $267. Overall, that’s a pretty good number for a niche store that likely count one-time purchases as a large part of its revenue.
Note: You should keep your CLTV higher than your CAC, again it depends on your Business strategy adopted, maybe you want to capture Market Share first & once achieved, then think about Profit! Otherwise, it’s important.
6. Predicting Future Customer Lifetime Value with Probabilistic Models & Machine Learning.
So, far we’ve tried to understand the importance CLTV Metric in a Business & several other Metrics associated with it, which can help come up with a better Marketing Strategy, by focusing on Customer Retention along with successfully adding New Customers, we’ve also discussed on how to calculate all the Metrics discussed manually, that will help, However, there are more Advanced Algorithms which learns based on Historical Data & it can make some fairly accurate future CLTV Predictions.
A. Probabilistic Models.
Probabilistic Models incorporate random variables and probability distributions. The random variables represent the potential outcomes of an uncertain event, whereas the probability distributions assign probabilities to the various potential outcomes. Examples of Probabilistic models include BG-NBD, Regression Models, Monte Carlo simulation, and Markov Models. It formalizes relationships between variables in the form of fixed Mathematical Equations.
Below Modeling Techniques are considered proven & powerful in Predicting Future CLTV Accurately:
a) BG-NBD
· The model allows to assign a value to each of those future.
(Preferably used for continuous purchases made at any time)
b) Markov Chain
The model computes transition probabilities for each state
(Preferably used for discrete purchases made at fixed time)
Model Evaluation: SMAPE
Output: Creating Quantiles based on Predicted Revenue by utilizing Historical Data, classifying them as High(H), Medium(M) & Low(L), further creating Deciles for High value Cohorts, which in turn help Business to prioritize based on elbow curve plot for their Marketing Strategy & Customer Engagement.
B. Machine Learning Models:
While Artificial Intelligence (AI) is the broad science of mimicking human abilities, Machine Learning is a specific subset of AI that trains a machine how to learn.
A supervised Machine Learning Algorithm that can learn from data without relying on rules-based programming.
Machine Learning has evolved, it has been categorized into many types now, one of them which is very important has been highlighted below.
I. Learning Problems:
a. Supervised Learning: Supervised learning describes a class of problem that involves using a model to learn a mapping between input examples and the target variable.
b. Unsupervised Learning: Unsupervised learning describes a class of problems that involves using a model to describe or extract relationships in data.
c. Reinforcement Learning: Reinforcement learning describes a class of problems where an agent operates in an environment and must learn to operate using feedback.
Below Machine Learning Modeling Techniques are considered proven & powerful in Predicting Future CLTV Accurately.
a) Random Forest Regressor
- The model fits several classifying decision trees on various
sub-samples. - Uses averaging to improve the predictive accuracy and control
over-fitting.
b) Deep Neural Network
- The model has capacity to execute feature engineering on its own.
- DNN handles both linear relationship and non-linear relationship.
c) XG-Boost
- This robust algorithm uses level wise splitting.
- Uses parallel learning and regularization to avoid over-fitting.
d) Light GBM
- Light GBM uses leaf wise splitting enables it to converge much faster.
- Uses parallel learning and has low memory usage.
Model Evaluation: RMSE, Adjusted R Squared & SMAPE
Output: Creating Quantiles based on Predicted Revenue by learning from Historical Data, classifying them as High(H), Medium(M) & Low(L), further creating Deciles for High value Cohorts, which in turn help Business to prioritize based on elbow curve plot for their Marketing Strategy & Customer Engagement.
We can also classify, new Cohorts based on some already existing attributes in the Data, even though we do not have enough Transactional Data for these fresh Cohorts. This could be possible due the fact that, our Models have fair enough Learnings to Classify these new Cohorts into High, Medium, Low, this could really help Marketers act fast on their Marketing Strategy, also equally beneficial to Sales & CRM Teams.
7. How to Apply Customer Lifetime Value?
Once, we have calculated Customer Lifetime Value, along with Predictions, & by providing a Marketing Window, Retailers can utilize this Data to Further refine, optimize their Strategies, New Customer Acquisition, Customer Retention & other aspects which may boost Revenue & Profitability.
I. Here are some actionable ways you can extend the life of your Customer and get LTV Higher:
a. Customer Retention:
Utilizing CLTV Data, Marketers can target the right Customers, Retailers can provide better Customer Experience, which could improve Customer Satisfaction further, & many more.
High value Customers can be better engaged, Passive Customers can be re-engaged.
b. Cross sell/upsell:
By engaging, with H, M & L Value Customers, Marketing & Sales Team can come up with a better Strategy for Cross Sell/Upsell by coming with a customized offering for all the categories a Customer belongs to.
With such a Strategy in place, there is a huge possibility for growing Customers LTV, resulting in growth in Revenue & Profitability.
c. Product line expansion, Awareness & Pricing:
Not just giving Customers only add-ons will help, try to create an entirely new product portfolio suitable to all categories of Customers as per their CLTV Classification.
They are here, so make sure you capitalize on that through adding pieces of product to their workflow. Overall, these solutions should be at least related to your core product to promote synergy amongst your product and customer success teams, but this isn’t an absolute requirement. Price Structuring by knowing Customers requirement through Value Metrics would help a lot to grow business & improve Customer Satisfaction.
Conclusion: By Calculating & Predicting Lifetime Value of a Customer, it can bring a positive impact on Business growth, it’ll help in better engagement & re-engagement with Customers, with quality feedbacks from Customers, Products/Services can improve, Retailers can bring Products/Services awareness to the Customers by improving their relationship, which could lead to better Customer Satisfaction, reduction in Churn, resulting in improved Retention Rate. Also, Marketing Budget can be optimized by targeting the right Prospect, when an effective Strategy is put into work, backed by CLTV Data.
Disclaimer: The views, thoughts, and opinions expressed in the text belong solely to the author, and not necessarily to the author’s employer (past or present), organization, committee, institutions, or another group or individual.
Author can be reached at: nandish.mr@live.com
