Customer Lifetime Value

CLV background
For those unfamiliar with customer lifetime value (CLV), it’s a prediction of how profitable a customer will be to your company. If we know how much our customers are going to spend, we have a good view of how our business is performing.

A formula for CLV would be:
CLV = gross profit * expected spend - marketing costs

It sounds basic
Just predict what our customers are going to buy, how much they will spend and how much promotions (marketing costs) we send to them. 

Companies have started applying AI/ML to predict CLV, creating an arms race to see who can produce the most accurate predictions. The explosion of data has allowed practitioners to throw in any sort of data - app interactions, demographics, website behavior - anything to help their predictions. In addition, companies are using the most sophisticated AI/ML models. Airbnb tried 27 different algorithms to see which one gave the best prediction for their own CLV model. Google shows how you can apply deep learning to CLV on their cloud website.

The arms race spreads to personalization
Predicting how much a customer will spend will only get us so far. Now companies are turning their focus on using AI/ML to proactively increase CLV. Once again, these companies are deploying state of the art machine learning to either recommend you new offers or send you tailored discounts.

Starbucks is a good case study of personalization evolution. After Starbucks deployed their original ML personalization software, Starbucks saw increased engagement and revenue through the program. Now, the company is using reinforcement learning, one of the most complicated machine learning techniques available, to produce even better personalization.

Does all of this sophistication actually work?
Airbnb found a 5% improvement in prediction results when deploying their CLV solution, which they claim led to a material impact on sales. McKinsey estimates that “personalization, fully implemented, can unlock significant near-term value for businesses - such as 10 to 20 percent more efficient marketing and greater cost savings and a 10 to 30 percent uplift in revenue and retention.“

This makes sense. If you can more accurately predict what people want, even if it is just by 5%, you can expect some sort of CLV increase. Instead of just sending $10 off coupons to people, determining which offer is the best to send can lead to cost savings.

Robotics/Automation drive future CLV
Companies realize convenience is a driver for high CLV.  Morgan Stanley estimates that Amazon prime members spend 4.6 times than regular members with over 90% retention rate. While Prime isn’t a AI/ML innovation, it’s a glimpse into the future of automation.

Multiple companies are introducing new AI/ML innovations, including curbside delivery and delivery by robots, to make it easier and quicker for people to order.  Walmart is experimenting with robots to fetch goods for curbside delivery and autonomous/robotic delivery of groceries. Walmart recently partnered with Google to enable grocery ordering through Google Home.

Amazon is going to extreme lengths. Last year, they were awarded a patent on shipping a product to someone who doesn't even realize they need it. There is also this terrifying video of a blimp filled with packages, which are then delivered by drones.

In closing
If you think you are impervious to these algorithms, look at your spend from year to year at different retailers. 

Just don't kill the messenger