Understand custom CLV
Every Klaviyo account comes equipped with AI-powered calculations called predictive analytics. Klaviyo takes all of the customer data stored in your account, and generates predictive metrics to help you market smarter and more proactively.
Account pre-requisites
In order to see predictive analytics in your account, your account needs to meet certain requirements in order for the model to make predictions. Those requirements are:
- An active ecommerce integration or API that is sending placed order events to Klaviyo.
- At least 500 unique customers who have placed an order.
- At least 180 days of order history in your Klaviyo account.
- Orders that have been placed within the last 30 days.
- At least some customers who have placed 3 or more orders.
Predictive analytics metrics
Once your account has enough data to fuel the model, you’ll see the following metrics in your account:
- Historic CLV (customer lifetime value)
- Predicted CLV
- Total CLV
- Churn risk prediction
- Average time between orders
- Predicted gender
- Expected date of next order
The default CLV calculation in Klaviyo is based on a future prediction window of 365 days. That means all predictive metrics are in the context of 1 year from today. With Klaviyo CDP, the Custom CLV feature allows you to adjust the prediction window for all CLV metrics in Klaviyo which can be helpful for forecasting. Learn more about how this feature works.
Feature walkthrough
Note: Custom CLV is available to users in 2 ways: through Marketing Analytics and the intelligence section of Klaviyo Advanced KDP. Since you have purchased Marketing Analytics, you will find this feature in your own account under Analytics > Advanced.
This demo video shows the feature located under CDP > Intelligence. The location in the side panel is the only difference; all feature details and capabilities outlined in the video should match what you see in your account.
Adjust your prediction window
Similar to the RFM analysis thresholds, your prediction window should be based on your business structure and goals. Is it important to you to be forecasting multiple years ahead (like 730 days)? Or do you want to see predictions in a shorter window of time (like 180 days)?
If you have existing data around customer churn patterns, this may help you determine what your prediction window should be set to. For instance, if you know that customers are most likely to churn if they don’t purchase anything for 200 days, you may want to set your prediction threshold to 200 to easily identify people who appear to be a churn risk.
Use CLV metrics to convert more customers
Predictive CLV metrics (also called predictive analytics) give you insight into likely future behavior, which means you can proactively incentivize people who are likely to purchase, as well as attempt to save riskier customers with special incentives.
Build segments using predictive analytics
Include CLV-related metrics in your segment criteria to proactively reach out to customers who are likely to follow specific behavior. Determine the dollar values and segment criteria based on averages and overall trends that you know about your audience. Here are 3 suggested segments to build:
- Predicted to place at least 1 more order: Predictive analytics about someone > Predicted number of orders is at least 1
- Predicted to be high-value: Predictive analytics about someone > Predicted CLV is at least $XX.XX
- Low cart value: Predictive analytics about someone > Average order value is less than $XX.XX
Send targeted messages based on predicted behavior
Take the segments you built into consideration, and craft targeted campaigns with messaging that makes sense given their anticipated behavior. Here are a few strategies to try:
- Predicted to place at least 1 more order: Round up your best-selling and/or newest products into a curated trend email.
- Predicted to be high-value: Craft a campaign that includes a personalized product feed with recommendations a customer may like based on past purchases.
- Low cart value: Encourage these customers to try more products by offering them a buy-one-get-one deal, or an incentive to bundle products at a discount, encouraging a higher average order value
Build a flow based on Expected Date of Next Order
Configure a date-based flow that uses a profile’s Expected date of next order. You can set parameters around how far in advance of the date you’d like the flow to start. In the example above, the flow starts 2 days before the expected order date. This allows you to proactively prompt customers to restock or make a purchase when they are primed to buy.