The "80/20 Rule" in business suggests that 80% of your sales come from 20% of your customers. Even for the most popular brands, there will be customers who purchase once and never return. This means that if you improve your number of return purchasers by 1%, you'll see an improvement of 4% in your top line. It’s obvious why subscription brands should focus on reducing churn, but retention is incredibly important for non-subscription, consumer brands as well. The challenge is that all customers look very similar when you first acquire them, that they have purchased once so far. Once the cohort of new customers has "matured," the picture becomes clearer: some of these customers returned and made additional purchases, while the majority did not.
One may wonder if there is a better way to gain visibility into the future. Can you predict who your high-value repeat customers will be even before they discover your brand and make their first purchase? This is where predictive lifetime value (pLTV) comes in. We know that not all new customers are equal. Some will purchase once, while others will become your most loyal customers and advocates. Therefore, predicting the future value of each customer before they interact, and as they enter your funnel, is key to profitable growth.
Historically, pLTV signals have been leveraged for retention marketing. In simple terms, pLTV functions like an augmented reality of your current customers. It estimates the value a customer will bring to your business in the future, and helps you identify those at risk of churn as well as those who are likely to return and what products they may purchase next. pLTV helps you focus your communication efforts on customers to realize their full potential lifetime value. This includes paying attention to at-risk customers and personalizing every touchpoint with them by anticipating their needs. As customers interact with your brands more, each interaction becomes deep, durable and differentiated.
In addition, pLTV signals can be a highly valuable input for prospecting marketing, especially when available before and around the first transaction. Typically, advertisers send conversion signals to the ad platforms when someone transacts with the advertiser. That conversion signal is sent with in-the-moment information such as who bought it, when it was bought and how much they spent on the order. However, when you augment the current information with the future value of the customer just acquired, such as pLTV, then advertisers can start optimizing their campaigns to find higher future value customers, knowing that they are afford to pay a little extra to acquire such customers. They can afford to bid higher for similar future high value customers, or bid lower for those who are not likely to return. This approach allows advertisers to manage their acquisition funnel unit economics based on individual customer values, rather than just averages.
Before discovering your brand, prospects are essentially unknown. However, you can leverage third-party data, such as demographics, interests, psychographics, and purchase insights, to create a basic profile of potential customers. By comparing this with your high-value current customers (those with high lifetime value, or high LTV), you can make an estimate of potential lifetime value (pLTV) for these prospects. We'll call this the "Prospect pLTV", and it can be estimated for those prospects with the highest likelihood of becoming your customer.
Mathematically speaking, the Prospect pLTV estimation should be done concurrently with propensity estimation. This means that you should first identify the cohort that has a high enough propensity of becoming your future customers. Then, estimate pLTV only for this high propensity cohort.
Let's imagine that one of your high-value prospects has just engaged with your brand by clicking on an ad and visiting your online storefront. They browsed a few items, signed up for your newsletter, and then abandoned the store. Now you have some additional information about that prospect: You know how they interacted with your brand, which products they browsed, how long they visited each page, what they clicked on, how they reached your website, when they browsed and what type of devices they used. These are all first-party data signals that you can leverage in your pLTV model. Essentially, you can now mine the breadcrumbs and match them with your high-value customers' first interactions with your brand. This additional information raises the resolution of your pLTV model.
The prospect then returns to your site a few days later through Google search ads, and completes the first transaction. Congratulations! You have acquired a new customer. At this point you have even more information, namely the attributes of the first transaction: value, product purchased, total spent, how they paid for the product, what fulfillment or shipping option they utilized, which device they browsed on, what time of the day they made their purchase, what other products they browsed in this session, and so on. All these additional signals become meaningful in predicting if this customer is going to purchase again in the next 90-days and if so, how much they are going to spend. We call it the “Customer pLTV." Customer pLTV estimates the future value of all your current customers.
pLTV scores are further refined when customers return for their second and third transactions, however the sooner brands can derive a pLTV signal in their customers' lifecycle, the more useful that signal becomes for nurturing that customer and acquiring other customers who are similar to high-value customers.
Why do you need to predict LTV when you already have actuals? With actuals, you can look at the cohort of customers that have been with you for some time (say, 6 months or more) and tabulate how much each customer has spent with you during their first 6 months since acquisition. This is the actual LTV. You could use that information for many marketing analytics and optimization. However, there are certain limitations to this approach, namely:
You are using stale data, as you can't use the most recent acquisition cohort.
Your product may have evolved. You may have changed pricing and offers, and your early adopters may have had a different profile than the late adopters.
There are certain causations that this approach may be missing. Perhaps you acquired some of these customers by running an affiliate campaign that you since then discontinued, or you ran promotions, or your business has seasonal patterns.
With the importance of predictive LTV in mind, it's worth noting that businesses can create their own models to predict customer and prospect LTV. However, this requires data science and engineering expertise, investment in identity resolution and third-party data enrichment services and the necessary infrastructure to collect, process, and analyze large amounts of data. For many businesses, building and maintaining such system can be a significant barrier to implementing a predictive LTV strategy.
That's why we created Angler AI, a platform that enables businesses to generate predictive LTV scores for both their existing customers and prospects. Angler AI provides businesses with a no code configurable prediction engine, empowering businesses of all sizes and budgets to create and implement predictive LTV models. By leveraging Angler AI, businesses can gain valuable insights into customer behavior and optimize their marketing campaigns to maximize ROI.
In conclusion, predictive LTV is a critical metric that businesses can use to drive growth and profitability. By understanding the value that each customer brings to the business, businesses can make more informed decisions about how to allocate their marketing budgets and resources. With the right tools and expertise, businesses can create predictive LTV models that provide valuable insights into customer behavior and help them achieve their growth objectives.