Today’s launch of our new Predictive LTV model is something we at Lifetimely have been looking forward to for years.
In fact, the original vision for the model came before we even launched the first version of Lifetimely. Ever since, each iteration of the app was designed to bring us closer to where we could project the future value of any customer.
Before we get into details of how the model works, let’s look at why we made it our mission to calculate future LTV, what it took to build the model, what makes it different, and how you can use it to your advantage.
For many of you, this story will sound familiar.
At my last job at a major Finnish retailer, we faced the same questions as any other retailer or D2C brand:
The challenge was that our customer base was constantly changing. We knew that to be a forward-looking brand, we couldn’t base our decisions on what past customer cohorts had done, but on what our current cohorts - an entirely different mix of customers - could be expected to do. But how?
I decided to try what so many other D2C brands were already doing; I dumped everything into Excel, and by analyzing purchase frequencies, sales channels, product types, and dozens of other factors, I tried to come up with LTV estimates for our customers.
But developing a consistent way of applying so many factors was an impossible task. And to really forecast customer behavior, we wanted something that was:
So that’s what we set out to create!
We knew it would be a long process. First, we needed a foundation for assembling and analyzing customer data and calculating LTV. We built that with the current iteration of Lifetimely. The next step was to recruit a team of data scientists.
We decided that we wanted our model to be different from what was already out there. Instead of projecting just a single LTV number, we built our model to forecast the monthly sales of each individual customer over their lifetime. This gives you a big advantage in estimating CAC payback times and creating financial models. But of course, it also added several layers of complexity.
For over a year, we’ve been developing and testing the model, feeding it more data, even bringing on a PhD to help refine it. And today, it’s finally live!
All of our predictive modeling integrates directly into the existing Lifetime Value report.
For PLUS plan members, your complete LTV report (activated by a simple switch) will look something like this:
The darker cells on the top-left half of your report make up your existing LTV report based on historical data only. The lighter cells in the bottom-right half are the work of the predictive model, forecasting the future monthly sales of the customers that make up your newer cohorts.
Each new cell is a projection of the true value of an average customer in the cohort at any point in time.
Our model projects LTV for every individual customer, meaning you can apply any of normal LTV report filters - marketing channel, customer/order tags, product type, etc. - to dig into your customer segments at any level of detail you want.
Estimating CAC payback times is now automatic, since the CAC payback tool works with projected figures on even the most recent cohorts.
Aggregated LTV Projections
Directly beneath the LTV matrix is a new section called “Predicted LTV” that gives you average customer LTVs on timelines of 3 months, 6 months, 12 months, and 24 months. These projections aggregate both measured and forecasted data for all customers included in the time frame of the report.
The LTV model right now projects accumulated sales per customer, cohort sales, and accumulated sales / CAC. The accumulated sales margin metric will be added soon, and we’ll continue to adapt the model to project other metrics.
All PLUS plan users have access to an exclusive Slack channel to receive priority customer support, give direct feedback, and get early looks at new features and improvements.
To access the new predictive modeling features, you will have to sign up for our new $149/month PLUS plan. For now, we only recommend this plan only for stores with at least 1000 customers and 12 months of sales data (though we’re working to produce reliable projections for smaller and newer stores).
Sign-ups for Beta access are limited and are by request only. You can request access from the Change plan page in the Settings tab.
We're excited to continue to build out this model, and we'll be adding plenty more features to the PLUS plan in the coming months!