.avif)
TL;DR: Klaviyo predictive analytics turns your order history into forward-looking metrics like predicted CLV and churn risk. Once your store has 500 buyers, 180 days of history, and some three-time customers, you can build segments that flag who is worth more and who is about to leave. This guide covers the metrics, the setup rules, five segments to build, and the flows that turn those scores into repeat revenue.
Klaviyo predictive analytics is a set of machine-learning metrics that forecast what each customer will spend and when they will stop buying, so Shopify brands can act on retention before the revenue quietly walks out the door. Most stores already sit on this data. They just never turn it into a segment.
Here is the problem. You can see that sales dipped last quarter, but you find out which customers slipped away only after the reorders stop. By then it is a win-back job, not a retention one.
Predicted CLV and churn risk fix that. They score every buyer on how much they are likely to spend next and how likely they are to leave, and they refresh every week. Build a few segments on top of those scores and you can reach the right people before they are gone.
This guide breaks down the predictive metrics, the data you need to unlock them, five segments worth building, and the flows that turn a churn score into recovered revenue. No fluff, just what we run across the Shopify stores we manage at CartStrings.
What is predictive analytics?
Klaviyo predictive analytics uses machine learning to score every customer on future behavior. It reads order history, timing, and engagement, then predicts lifetime value, churn risk, and the next order date. Klaviyo retrains the model at least once a week, so the numbers stay current as buyers act.
Think of it as Klaviyo doing the math you never have time for. Instead of guessing who your best customers are, you get a running estimate of future value and risk on every profile. Those estimates live on the Metrics and insights tab of each contact and, more usefully, become filters you can segment on.
One catch worth knowing up front: predictive analytics sits inside Klaviyo's paid plans. It is not part of the free tier. If you are already paying for Klaviyo, though, the feature is included, so most brands are leaving it on the table rather than paying extra for it.
When do predictions unlock?
Klaviyo needs enough data before it shows predictions. You need at least 500 customers who placed a real order, 180 days of order history with orders in the last 30 days, and some customers who have bought three or more times. Miss any one of these and the predictive box stays blank.
A few details matter here. Those 500 orders have to be real: not cancelled, not refunded, and non-zero in value. You also need an ecommerce integration like Shopify feeding placed-order events, or the same events sent through the API. And the three-order threshold is not random. Klaviyo needs a pattern before it can predict an individual, and one or two orders rarely make a pattern.
That last point ties to a well-known retention stat: after a third purchase, a customer's odds of buying again jump to roughly 62%. The same repeat behavior that makes a customer loyal is what makes them predictable. You can read Klaviyo's full qualification rules in its guide to predictive analytics.
If you are not sure your account qualifies, or your order data is messy from a migration, a Klaviyo audit is the fastest way to find out where you stand.
What is predicted CLV?
Predicted CLV is Klaviyo's estimate of how much a customer will spend with you over the next year. It sits next to historic CLV, the total of every past order minus refunds and returns, and total CLV, which adds the two together. Together they show where your revenue has been and where it is headed.
Klaviyo builds the number from purchase frequency, average order value, and time between orders. A profile might read $401 in historic CLV, $99 in predicted CLV, and $500 in total CLV. The historic figure is fixed. The predicted figure is the one that tells you how much to invest in keeping that customer around.
This is where segmentation earns its keep. You can group customers by predicted CLV and treat a shopper likely to spend $300 next year very differently from one likely to spend $20. Klaviyo lets you build these groups with the Predictive analytics about someone condition, and its CLV segmentation guide walks through the exact filters.
What is churn risk?
Churn risk prediction is the probability a customer will not buy from you again, scored from 0 to 1. A 0.21 score means a 21% chance they are gone. A 0.90 means 90%. Every new order drops the score. Every silent week pushes it up.
The score leans mostly on the number and frequency of a customer's orders, then factors in signals like browsing drop-off and email disengagement. In the profile view, Klaviyo colors it green when risk is low, yellow when it is climbing, and red when a customer is close to lost. When you export the data, churn risk comes through as a plain number between 0 and 1, so a 0.45 is a 45% risk.
The practical move is to stop treating churn as a surprise. A rising score is an early warning you can act on weeks before the customer would have quietly stopped buying.
5 segments to build
The metrics only matter once they become audiences you can message. Here are five predictive segments worth building first.
- High predicted CLV. Filter for predicted CLV above a threshold that fits your store. This is your VIP bench: send early access, premium bundles, and your best email campaigns here first.
- High value, high risk. Combine predicted CLV above your threshold with churn risk above 0.7. These are the customers worth saving hardest, so give them a real reason to return, not a generic blast.
- Low predicted CLV, one-time buyers. Target shoppers unlikely to reach your average order value with a small nudge toward a second purchase. Klaviyo's own example uses this exact group.
- Replenishment due. Use average time between orders and expected date of next order to catch buyers in their reorder window before a competitor does.
- Rising churn, still engaged. Churn risk is climbing, but they still open your email. Reach them with content and reminders, not discounts, while attention is cheap.
Layer in an engagement filter, like opened an email in the last 90 days, so you protect deliverability while you send.
Flows built on predictions
Segments tell you who. Flows do the sending on autopilot. Predictive data plugs into the automations that drive most retention revenue.
A replenishment flow is the clearest fit. Klaviyo's expected date of next order and average time between orders let you time a reminder to each buyer, firing five to seven days before they are due to reorder. A churn-prevention or win-back flow can trigger when churn risk crosses a line, catching people before the gap gets too wide to close.
High predicted CLV feeds a VIP flow with early access and rewards, while one-time buyers with low predicted CLV get a second-purchase nurture. These are exactly the kinds of sequences we set up in our email automations work, and you can see how the pieces fit in our flow guides.
One warning from Klaviyo: do not count down to the expected date of next order for repeat buyers, or they get the same sequence before every order and start unsubscribing. Use it to nurture first-time buyers toward a second order, and lean on standard replenishment logic once a cycle is clear.
Read segments, not people
This is the part most guides skip, and it is the one that saves you from bad decisions. Predictive numbers are reliable across a group, not for any single person.
Klaviyo says so directly. A profile might show 1.43 predicted orders, which is not a real number of orders. It means Klaviyo expects one or two, with a chance of more or fewer. Line up five customers at 1.43, 0.25, 3.12, 0.78, and 2.97 predicted orders, and you can reasonably expect about nine orders from the group. The individual noise averages out.
So do not panic over one profile's predicted CLV or churn score. Build the segment, message the group, and measure the group. There is even a clean way to size the payoff: average the churn risk across a segment, subtract that from 1, and multiply by the number of people in it. That gives you the count of customers predicted to come back. Plan campaigns around that math, not around any one customer.
Turn scores into revenue
Klaviyo predictive analytics is only as good as what you send next. The metrics are already in your account, waiting behind a feature you likely pay for. The work is turning predicted CLV and churn risk into segments, then wiring those segments into flows that reach people at the right moment.
Retention is where this pays off. It costs five to seven times more to win a new customer than to keep one you already have, and you have a 60 to 70% chance of selling to an existing customer versus 5 to 20% for a new prospect. Predictive segments point your budget at the people most likely to say yes. Across the Shopify stores we manage at CartStrings, that focus on retention is a big part of why email drives around 32% of revenue.
If you want that set up properly instead of half-used, book a call and we will map the segments and flows to your store.
Frequently Asked Questions
Is Klaviyo predictive analytics free?
No. Predictive analytics is included on Klaviyo's paid plans, not the free tier. If you already pay for Klaviyo, the feature is available at no extra cost, so most brands can turn it on today.
How accurate is Klaviyo's churn prediction?
It is reliable across a segment, not for a single customer. Any one profile can spend more or less than predicted, but the highs and lows cancel out over a group. Accuracy also improves as your store accumulates more order history.
What counts as a high churn risk score?
Churn risk runs from 0 to 1. Many brands treat anything above 0.7 as high and worth a save attempt. Pair the score with a filter like last purchase date so you are targeting real lapses, not brand-new buyers.
Why is my predictive analytics box blank?
Either your account has not met the requirements yet, or Klaviyo does not have enough data on that specific person. You need 500 customers with real orders, 180 days of history with recent orders, and some three-time buyers before predictions appear.
Can I use predicted CLV inside a flow?
Yes. Build a segment on predicted CLV and trigger a flow when someone enters it, or use date properties like expected date of next order to time sends per customer. Both let you automate against predictions instead of sending manually.
