- Klaviyo shows historic CLV, predicted CLV, total CLV, churn risk, average time between orders and expected date of next order on eligible profiles.
- To qualify, Klaviyo says you need 500+ customers with an order, an ecommerce integration or API orders, 180+ days of order history with orders in the last 30 days, and some customers with 3 or more orders.
- Klaviyo says predictions "work best when averaged over many customers" and aren't expected to be exact for any one person.
- Klaviyo itself says an expected-next-order flow shouldn't replace replenishment flows if you know your product cycles.
- Build the fixed replenishment flow first. Use predicted CLV and churn risk to sort people, not to time every send.
This is for Shopify founders who've seen the "Predictive analytics" box on a Klaviyo profile and want to know whether it's worth building anything on.
What is Klaviyo predictive analytics?
It's a model Klaviyo builds on your own order data. Per Klaviyo's help center, it builds a CLV model for your account and retrains it at least once a week. The results show in a "Predictive analytics" section on each eligible profile:
- Historic CLV: the total value of the person's past orders, accounting for refunds and cancellations.
- Predicted CLV: what Klaviyo expects them to spend over your prediction period, which is 1 year by default.
- Total CLV: historic plus predicted.
- Churn risk: the probability they stop buying, based on how many orders they've placed and how often.
- Average time between orders: in days.
- Expected date of next order: when Klaviyo thinks they'll buy again.
Klaviyo also sets a predicted gender from first names and census data, but it doesn't show it on the profile, and Klaviyo says it's an approximation.
What do I need before predictions show up?
Klaviyo lists four conditions for CLV predictions to appear:
- At least 500 customers have placed an order. Cancelled, refunded and zero-value orders don't count.
- You have an ecommerce integration, like Shopify, or send placed orders through the API.
- You have at least 180 days of order history and orders within the last 30 days.
- You have at least some customers who've placed 3 or more orders.
Even then, an individual profile can come up blank. Klaviyo says that means it doesn't have enough data on that person to make a prediction.
So if you're a newer store, this isn't your problem yet. Get the core flows live and come back to it.
How does Klaviyo calculate expected date of next order?
Per Klaviyo, it looks at that customer's own order pattern and at how all your customers behave. If someone's orders follow a pattern, Klaviyo uses it. If they don't, or there isn't enough data, it makes a "reasonable prediction" based on your other customers. One-time buyers get a date based entirely on your customer base.
Two limits Klaviyo calls out:
- It doesn't consider which products the customer ordered.
- Profiles who've never ordered don't have this date.
That first point is the big one. If you sell a 30-day supply and a 90-day supply, the prediction doesn't know which one someone bought. A replenishment flow filtered by product does.
Should I use expected date of next order instead of a replenishment flow?
No. Klaviyo says this flow "shouldn't replace replenishment flows if you know the general cycles for most of your product categories," and it recommends separate Placed Order flows with trigger filters and time delays for product-specific cycles.
That lines up with what the numbers say about replenishment. Across 4 consumable brands I work with, replenishment flows earn $2.23 per person, about 28x the best campaign, and only 2 of the 4 had one. A subscription wellness brand's "your order is coming up" reminder made $47,321 in one month, 74% of all its flow revenue. That's a fixed, product-aware reminder doing the work.
Klaviyo also warns against counting down to the expected date for repeat customers, because they'd get the same sequence before every order, which may lead to unsubscribes.
If you haven't built the fixed version yet, start with how to set up a Klaviyo replenishment flow and replenishment email timing.
Where predictive analytics is actually useful
Once the fixed flows are live, predictions help you decide who gets what. My take: use them to sort people, not to time every email.
1. A VIP segment by predicted CLV. Klaviyo lets you build segments on "Predictive analytics about someone," using historic, predicted or total CLV. Pair a high predicted CLV with engagement conditions and you've got a list worth treating differently: early access, a better thank-you, fewer discounts. My VIP email segment post covers what to send them.
2. A low-value filter for campaigns. Klaviyo's own CLV segment example uses a condition like predicted CLV "is at most" a small number. That's useful for keeping heavy promo sends away from people who rarely buy, or for giving them a different message.
3. A catch-all for people outside your cycles. Klaviyo says an expected-date-triggered flow checks existing and new profiles against the date automatically when its messages are in Manual or Live status. That can be a reasonable net for customers whose products don't fit a known cycle, as long as it's not doubling up on your replenishment flow.
4. A gut check on lifetime value work. Historic and predicted CLV give you a quick read on whether your retention work is moving the right people. If you're working on that, read how to increase customer lifetime value.
For replenishable products, I also like a simpler rule alongside any prediction: customers with no order in 60 days go into an educational cycle that brings them back. That doesn't need a model.
How accurate are Klaviyo's predictions?
Accurate enough in aggregate, not for one person. Klaviyo says predictions "work best when averaged over many customers and are not expected to be exact for any single individual." It even notes you may see an impossible number, like 1.43 predicted orders for a customer.
So use them for segments of hundreds or thousands of people, where the averages hold. Don't build a one-to-one promise on a single profile's date.
FAQ
What is Klaviyo predictive analytics?
A set of model-based estimates on each eligible profile: predicted CLV, churn risk, average time between orders and expected date of next order, plus historic and total CLV.
Why don't I see predictive analytics in Klaviyo?
Your account likely doesn't meet the requirements yet: 500+ customers with an order, 180+ days of order history with orders in the last 30 days, an ecommerce integration or API orders, and some customers with 3+ orders.
How often does Klaviyo update predictions?
Klaviyo says it retrains the CLV model at least once a week.
Can I trigger a flow on expected date of next order?
Yes. Klaviyo supports a flow triggered by expected date of next order. It says not to use it in place of replenishment flows when you know your product cycles.
Can I change the predicted CLV period?
Predicted CLV covers 365 days by default. Klaviyo says Custom CLV, included with Personalization and Advanced KDP, lets you change it.
Founder of Bardown Digital, a Chicago-based Klaviyo email and retention agency for Shopify brands. He's worked with brands doing $500K to $100M a year, with $100M+ in email-attributed revenue across them.