Predicting Customer Lifetime Value (CLV) Before a Prospect Becomes a Customer

How Car Salespeople Figure Out Your Value to Them

Have you ever noticed how car sales professionals try to estimate your financial worth with small talk?

These salespeople tend to ask some quite simple but seemingly irrelevant questions before they even check your credit card or previous purchase history. At my last car purchase, I was judged by my looks, asked about the car I was driving, and questioned who the new car was going to be for. This conversation transpired long before the salesperson reviewed my application.

You might think these this is a strategy to make prospective customers comfortable. Of course not! It is rather a subtle way to determine your possible value, which allows salespeople to use their time wisely and get a higher conversion rate.

Idle chit-chat can be quite revealing—It can suggest where you fall on important decisions like whether you’ll lease or buy direct or even the type of cars that will interest you in the first place. All of this helps car salespeople determine your long-term value to their brand.

Customer Long-Term or Lifetime Value (CLV) simply refers to a customer’s long-term value to a brand in dollars. This metric gives businesses insight into the advertising and retention costs that will convert a prospect into a long-term loyal customer.

Unlike car salespeople in a dealership, online businesses don’t have the opportunity to make strategic small talk with their prospective customers before they make a purchase. They substitute the conversation with data such as income range, demography, gender etc. These attributes help digital marketers plan better because patterns in these variables often suggest the CLV each customer may have.

The Current State of Digital Marketing

The revolution of digital advertising in the last decade has helped bridged the significant gap created by traditional methods of marketing. Startups can now compete with giant rivals who have deep coffers to advertise in old media i.e. (print, radio and television).

Most online business marketers focus on short-termed successes with the Cost Per Acquisition (CPA) technique. CPA is the dollar amount required to complete a one-time business transaction, whereas Cost Per Click (CPC)refers to the amount required to get a user to click on your internet ad. By contrast, Cost Per Mile (CPM) is the cost per thousand impressions an ad receives.

A Better Way to Measure Success

The CPA technique is usually used for business who want instant conversions on their products and services. Hence, it is more expensive than all the other types of online ad. Focusing on a one-time transaction in this way is a common strategy for digital marketing campaigns, but this is a short-term success. The CLV metric allows marketers to understand how these customers can be converted to long-term loyal customers.

CPM, CPC and CPA are incredible techniques to get new leads and convert a prospect into new customer. However, marketers can’t use them to convert long-term loyal customers. CLV measures the value of a customer across their entire relationship with the brand–past, present, and future.

Peter Fader, a renowned customer marketing professor, in his book Focus on the Right Customers for Strategic Advantage,” explained the “Value Per Acquisition” metrics in depth. Simply put, not all customers are created equal. So, CLV is important to capture the present value of all future net cash flows associated with each individual customer.

Knowing CLV allows marketers to use their budgets to attract and retain only their best possible customers. Ad costs depend on the competitiveness of the industry, not the return on investment. That means going after the best possible ROI is the marketer’s responsibility, so CPA and CPC techniques waste advertising dollars by converting one-time customers instead of loyal ones.

Digital Advertising Costs Are On the Rise

Wordstream—a global advertising solution provider—recently released data capping advert cost on Google network. They pegged average CPA costs across all industries at $48.96 for search and $75.51 for display. B2B niches had the most expensive CPA costs at $116 for search and $130 for display, followed by tech industries at $134 for search and $104 for display in 2018.

The increase in Google advertising costs over the years has had little effect on CPC rate when compare to CPA. The average CPC rate across industries stands at $2.69 for search and $0.63 for display. Legal and Consumer Services rocked the CPC boat with expensive rates at $6.75 and $6.40, respectively.

The data also revealed that Adwords search has a higher click-through rate than the display network. Average CTR across all industries stands 3.17% for search and 0.46% for display in 2018.

According to Mary Meeker’s 2018 internet trends report, social media ad engagement is on the rise for e-commerce businesses. “With social media, ad engagement is rising, represented by Facebook’s e-commerce click-through rates which rose from one to three percent.”

Social Media advertisement costs are said to be rising faster than their reach, which means advertisers have to spend more to reach the same number of people as before. One might argue the effectiveness of social media and google network digital tools like remarketing, demography etc. These figures are still pointing to one fact – online ad costs are increasing.

CLV Can Fix Your Return on Investment in Digital Ads

One of Bluevine’s Technologies partners, Matt Estes, once insisted that CPA, CLV, and ROI are the only important marketing acronyms. He mentioned a subtle relationship between CPA and CLV which many marketers seem to have neglected since—CLV gives you enormous insight into what your CPA should be. The ratio of CLV to CPA should be 3:1 for best business practice, but some all-stars reach 4:1.

The importance of CLV is rising as customer acquisition costs seem to increase daily. With it, organizations can make informed decisions on product pricing and development, marketing budget, sales, and more. The fact that tech companies can pay $134 in advertisement cost for a one-time transaction has attested to the usefulness of CLV as the next most important metric for businesses to survive.

The Challenges of Calculating CLV

But, CLV computation is difficult. That’s because of the complex information needed to calculate the perfect CLV. Lifetime Value needs data like amount spent by customers and the frequency and duration of their purchases. Hence, it cannot be calculated for prospective customers nor a one-time transaction.

Consider the figure below where both the first and second customer has the same number of purchases but their lifetime value is different because one customer has stopped spending and the other is still likely to spend in the future.

In order to accurately predict CLV, businesses must have the following at hand:

  • Lots of customers
  • Customers with more than a year of purchasing activities
  • Lots of computation power.

So far, Professor Fader and his team have done the best job computing CLV based on lots of customer history.

However, their method (also known as the Pareto-NBD method) still has few limitations and pitfalls. Pareto-NBD and similar analysis heavily rely on lots of historical data about people’s purchasing behaviour to predict CLV. This means that by the time you compute CLV at the individual customer level, you have already spent your acquisition dollars.

Imagine if a car dealer had to wait for several car payments worth of history before deciding what deal to give me? The best companies today using these methods compute CLV way before the first few transactions.

How to Estimate CLV Before Leads Become Customers

What if you could predict CLV in advance of a customer’s second purchase—or even before first purchase? In today’s business ecosystem, enterprises can easily predetermine certain characteristics of their customers based on some key actions. One example of an action to look at is the time of day and week the user lands on the website. Another is to pay attention to the products and pages they like to interact with the most. Moreover, you can add more data by connecting with data from third-party providers.

The approach we have used at Retina takes advantage of our very rich dataset of customer profiles and marketing and product interactions prior to any transaction. We use modern classification algorithms so we can take full advantage of this data.

CLV is counter-intuitive in the sense that it’s not only telling who your good customers are. It also shows you who your “one-time” customers are. If you can predict this information in advance, you don’t need to worry about it when they eventually do not come back.

How to Calculate Customer Lifetime Value

The most comprehensive approach is to combine the best elements of Pareto-NBD models with modern machine learning techniques. This happens in four steps:

  1. Identify four types of datasets in a database: transaction history, customer demographics, profile data, marketing actions and product/website/app actions.
  2.  Forecast every existing customer’s behavior and predict the number of future transactions, churn date, and expected spend amount
  3. Split the data into two sets; training and test.
  4. Add hundreds of features and attributes about the customer to this dataset.

Now the dataset should look very familiar and ready for machine learning models explained above.

Note: There are several variations which were not covered that normalize for cohort behavior and multicollinearity. After doing that, you can model out e-CLV (Estimate of CLV using pre-data) or predicted CLV based on attributes that are not related to customers’ transaction behaviour.

What can you use CLV for?

CLV can drive a lot of value to a business because of the following use cases:

Call Center:

  • Determine how much of a refund/promotion an unsatisfied customer is worth. Offering a $10 rebate to satisfy and ultimately retain a customer worth $500 is certainly worth it.
  • Call centers can also be used to push the right kind of offers.
  • Always remember that keeping customers happy increases CLV.

Advertising:

  • Invest in ads designed to target high CLV demographics.
  • Adjust your creative to appeal to your highest CLV customers.
  • Personalize email campaigns to engage customers based on their CLV segment. Personalization is key to success, so use CLV and transaction history to customize emails.

Business Insight:

  • Dig in to find the reason why certain customer segments have low CLV.
  • Answer vital questions like, “Is there a way we can enhance their experience?”
  • Use surveys to explore low-CLV demographics.
  • Compare low-CLV product interaction with high-CLV product interaction.

Budget Allocation:

  • Justify expanding or contracting your budget per acquisition channel and/or promotional content.

Business Strategy:

  • Explore markets with high CLV but low penetration.
  • Look at your CLV/CAC ratio to understand how healthy your business model is. Remember, optimizing this ratio is key to growth, so strive for a ratio of 3:1 CLV to CPA.
  • Loyalty programs encourage repeat purchases and retention—It costs less to retain a customer than acquire a new customer.
  • Use referral programs to reward your loyal customers and generate new ones. A prospective customer is 4x likely to buy something recommended by a friend.
  • Leverage convenience/vertical integration to build loyalty, trust and a stronger customer base.

Other Uses:

  • Early value driver’s analysis can be used to see which behaviors are good indicators of high CLV.  You can also model your acquisition on reaching prospects who look like the highest CLV customers. 

About Retina

At Retina we obsess over increasing customer CLV and reducing CAC (Customer Acquisition Cost) through data science.

As the cost of customer acquisition skyrockets, it is important to get focused on CLV at the customer level. Companies that computed their CLV to CAC ratio incorrectly are failing in an increasingly competitive environment i.e. (Blue Apron, Wayfair, Chef’d). Companies that get it right are achieving sustainable profit and growth i.e. (FAANG companies, Dollar Shave Club, Ring, etc.).

The Retina Platform computes predictive-CLV and early customer behavioral drivers of CLV at the customer level. Furthermore, it uses next-gen machine learning algorithms (Built on 30 years of academic research). Within a matter of days, Retina automatically builds audiences (Facebook, Google, LinkedIn, Snap) for your marketers to acquire new high-CLV customers and retain your existing high-value customers.

Interested in what we do? Log on to http://staging-retinaai.kinsta.com/story for a more detailed presentation about what we do, how we work and how we can help your business grow to the next level.