We used to hear about LTV/CLTV constantly in the days before “Big Data” became the latest buzzword. Now that customers are equipped with smartphones and shopping online, LTV has been pushed aside for Big Data.
Assigning CLTV to a single customer has always been difficult. With data flowing from mobile, online and direct sources, marketers are overwhelmed and no longer even pretend to understand or build lifetime value calculations. Instead, the customer experience leading up to a sale is at the forefront.
Did this change in focus happen only because calculating lifetime value is difficult?
Perhaps we need to answer a few other questions before we answer this one. Questions like: What is lifetime value anyway? Is this even a useful metric? Is any customer dedicated and loyal to any product or service over their lifetime?
What is Customer Lifetime Value?
CLTV predicts the future value of a customer relationship. At its most basic, CLTV assigns a dollar value to a customer. There are many algorithms used to calculate this value. The approach outlined below assigns a relative value to existing customers and describes a method for acquiring customers who fit the description of “best” or “most profitable.”
Why is this useful?
Assigning a value to a customer relationship can help us better segment customers and determine where to spend the next marketing dollar. Below are some steps marketers can take to begin the LTV journey.
Rather than putting the notion of customer lifetime value on the shelf, there are 5 steps marketers can use to assign value without building tedious mathematical equations:
Step 1: Identify the business challenge
First, identify your business challenge. Is it retention? Growing the customer relationship? Finding new customers?
All of these questions are likely a need of every business at any given point in time. Several challenges can be answered either with sentiment – that is, what your customers think of your product or service – or your own data collected from each customer contact. Can you identify how frequently they purchase? What they bought? How long ago? The dollars spent? Although your data may contain hundreds of other pieces of information, these four are the most significant. They represent how a customer behaves when interacting with your company.
Step 2: Organize the Big Data
Begin by collecting each customer’s purchase history. This likely will present a challenge if your business stores data in several systems constructed for different purposes. This data collection exercise may involve data discovery, collection, cleansing and organizing details around a customer. Another potential hurdle may be convincing other business groups in your company to share information.
Step 3: Solve for “value” – both customer and product
Once data is collected, a value can be assigned to each individual customer. Let’s say you own an upscale jewelry store. You have organized and collected your customer data and are ready to assign a value to each individual. You know how much money was spent, the number of visits (online and offline), the number of SKUs (or departments) visited and months since last visit. Also, you have collected data on the actual items purchased. For each of these criteria, a standard rank will be calculated.
First, step: calculate averages for each metric across the entire base of customers.
Next, calculate the standard deviation (this is a simple function in Excel or in a statistical software package).
In the jewelry store example, we find that customer history over the last 2 years yields the following:
- Average dollars spent: $475; standard deviation: $716
- Average departments (SKUs) shopped: 1.7; standard deviation: 1.4
- Average number of visits: 1.4; standard deviation 1.1
- Average months since last visit: 8.9; standard deviation: 3.6
Next, for each of the four measures above, subtract each customer’s individual value from the total customer base average and divide by the standard deviation. So, let’s say the first customer in our database has spent $600 at your online store. The average spend by all customers in your database is $475 (calculated above). So, $600 - $475 = $125. Now, divide by the standard deviation of $716. This calculation yields a standard score of .17.
We also know that this customer visited 4 times and purchased items from 6 different departments and visited 3 months ago. Performing the same calculation we did for spend, we calculate the following standardized scores: 2.3 for visits, 3.1 for departments shopped and 1.6 for months since last visit.
Next, we sum all of the standard scores for this customer producing a total standardized score value of 7.2. Once these scores have been calculated for each customer, the entire database can be ranked from high to low based on the assigned scores.
We could stop here, but not all products yield the same profit margins. Weighted values can be included based on product profitability. Before adding a fifth score to the list above, each product category can be assigned a ranking based on profit margin. Higher point values indicate the greatest margin as follows:
7 Points = Gold Jewelry/Pearls
6 Points = Chains
5 Points = Semi-Precious
4 Points = Precious
3 Points = Non-Karat/Bead/Silver
2 Points = Diamond Fashion
1 Point = Watches
If a customer purchased from both the Gold Jewelry/Pearls and the Diamond Fashion categories, a total of 9 points are assigned.
Next, the average and standard deviation is calculated for the department products above; each customer’s individual point value is tallied and then standardized as described for items 1-4 above.
Step 4: Segment customers
Now that each customer has a point value assigned based on their past behavior, the entire base of customers can be ranked by from highest total point values to lowest. This classification provides a starting point to measure the effect of your next campaign across five different dimensions as shown in the chart below. Customers with the highest aggregate point values are in decile 1 with lowest assigned point values ranking in decile 10:
Step 5: Build the Value Bridge
A Value Bridge is a method for applying the segmentation above to individuals who are not yet customers. To apply this in a new customer scenario, data attributes must be identified that are common to both customers and prospects. Typically, this data can be purchased and appended to individuals so a comparison can be made to discover which attributes correspond to highly ranked customers then find prospects who fit the description. For example, if we know that decile 1 above contains single women between the ages of 25 and 40 with income ranging from $100 - $150K, we would want to target similar individuals in the population at large.
So, a step-by-step approach would look like the following:
- Identify your best customers based on the ranked point values.
- Overlay external demographic/psychographic/behavioral data to a statistically reliable sample of customers and prospects
- Identify the attributes that describe your best customers via a scoring equation (model)
- Market to individuals who fit the profile of your top-performing customers
After creating the measures of customer value, the next goal is to manage customer value by fitting marketing campaigns to various customer segments. This will be an iterative process to best determine which marketing contacts generate ROI for each customer segment.
Customer LTV can be an extraordinarily effective way to get more value from all the data you manage. This calculation separates smart marketers from the ones who are just carried along by the tsunami of big data. Yes, it requires some work – but the payoff can be tremendous when you find you are focusing on the most profitable segments of your customers and prospects.