Most e-commerce brands are running their email marketing programs based on what has already happened. They look at past purchases, past open rates, past revenue. While this is all useful data and worth paying attention to, Klaviyo offers something beyond the rearview mirror: predictive analytics that use AI-driven machine learning to forecast what is likely to happen next.
Understanding and using Klaviyo’s predictive features is one of the most underutilized advantages available to brands on the platform. Most businesses have the features turned on without realizing it. Very few are actively using those predictions to change how they segment, when they send, and which customers they prioritize.
Here is what Klaviyo’s predictive analytics can do and how to put them to work for your email marketing program.
What Predictive Analytics Actually Means in Klaviyo
Predictive analytics in Klaviyo refers to a set of machine learning models that analyze historical behavior across your account to make forward-looking predictions about individual customers. These predictions are calculated at the profile level, meaning Klaviyo assigns a predicted value to each contact based on that person’s specific history and behavioral patterns.
This is different from aggregate reporting. Aggregate reporting tells you what happened to your list as a whole. Predictive analytics tells you something about what is likely to happen with a specific customer, which is far more actionable for segmentation and targeting.
Klaviyo‘s predictive features are built on the behavior of thousands of Klaviyo accounts and millions of data points, which means even a brand with a relatively small customer history benefits from the broader patterns the model has identified.
Predicted Customer Lifetime Value
Predicted customer lifetime value (CLV) is one of the most valuable outputs in Klaviyo’s predictive suite. It projects the total revenue a given customer is likely to generate over the next 90 days, 180 days, and one year.
This prediction is calculated based on the customer’s purchase frequency, order value, recency, and how those patterns compare to similar customers in Klaviyo’s broader dataset. A customer who bought twice in the last 60 days at above-average order values is likely to have a much higher predicted CLV than a one-time buyer who purchased six months ago.
The practical applications of predicted CLV are significant. You can build a segment of your highest predicted-value customers and treat them differently: give them early access to new products, create VIP campaigns just for them, or apply extra care to their win-back sequence if they start to show signs of disengagement. You can also use predicted CLV as a filter for your lookalike audiences on Facebook and Google, sourcing your paid ads from only your most valuable predicted customers rather than your full list.
Churn Risk
Klaviyo’s churn risk prediction estimates the likelihood that a given customer will not make another purchase. It is expressed as a score from low to high, and it updates dynamically as each customer’s behavior changes over time.
Churn risk is particularly useful for win-back strategy. Rather than waiting for a customer to go six months without purchasing before triggering a win-back flow, you can proactively identify customers who are showing early churn signals and intervene before they have fully lapsed.
Building a Klaviyo segment of customers with a high or very high churn risk and triggering a targeted winback flow for that segment is a proactive approach that tends to outperform traditional time-based win-back triggers because you are reaching people while they still have some likelihood of converting, rather than after they have fully checked out.
Combining churn risk with predicted CLV adds another layer of prioritization. A high-CLV customer with rising churn risk is someone worth sending a strong retention offer to. A low-CLV customer with high churn risk may not warrant the same investment.
Predicted Next Order Date
Klaviyo predicts when a given customer is likely to place their next order, based on their past purchase cadence and how that compares to similar customers. This is particularly powerful for brands with replenishable products, where a large portion of customers have a predictable reorder cycle.
Here’s an example. If you sell skincare products and the average customer reorders every 45 days, you can build a flow that triggers for customers approaching their predicted next order date. A personalized reminder that lands in their inbox a few days before they would naturally run out of product can capture that repurchase before they have to think about it, and before they wander off to search for a competitor.
Even for brands without predictable replenishment cycles, the predicted next order date can inform when to push a cross-sell or upsell offer. If Klaviyo predicts a customer is likely to purchase again within the next two weeks, that is a good window to introduce them to a complementary product.
Expected Number of Orders
Alongside predicted CLV, Klaviyo provides a prediction for the expected number of future orders a customer will place over a defined time horizon. This metric is useful for understanding the depth of a customer relationship, not just the dollar value.
A customer predicted to place five orders in the next year and a customer predicted to place one order might have similar predicted CLV if the single-order customer has a much higher average order value. But they represent very different types of customer relationships, and the strategies for retaining them might look different.
Looking at expected number of orders across your customer base can help you identify which segments are most likely to become loyal, multi-purchase customers versus which are showing patterns of one-time buyers. That insight should inform where you invest in loyalty programs, VIP treatment, and high-touch retention efforts.
How to Access Predictive Analytics in Klaviyo
Klaviyo’s predictive data is available in several places in the platform. Individual customer profiles show predictive scores, so you can pull up any profile and see their predicted CLV, churn risk, and next order date.
More practically, all of Klaviyo‘s predictive properties are available as segmentation conditions. In the segment builder, you can filter by predicted CLV above or below a certain amount, churn risk level, predicted next order date within a certain window, or expected number of orders. These conditions can be combined with behavioral data, purchase history, and email engagement to build highly precise audiences.
Predictive properties are also available as flow filters and triggers. You can build a flow that triggers when a customer’s churn risk crosses into the high-risk threshold, or when their predicted next order date is approaching.
Common Mistakes to Avoid
The most common mistake with predictive analytics is treating the predictions as certainties. These are probability models, not guarantees. A customer with high predicted CLV may still churn. A customer with high churn risk may still come back on their own. The predictions are signals to act on, not definitive outcomes.
The second common mistake is over-relying on predicted CLV to justify ignoring lower-value customers. Even customers with modest predicted values are worth retaining through good email practice. The predictive data is most useful for resource allocation and prioritization, not for abandoning entire segments.
The third mistake is not re-evaluating your predictive-based segments regularly. Because predictions update dynamically as behavior changes, a segment that made sense six months ago may need to be revisited. Customers who were high CLV may have shown declining engagement. Customers who were high churn risk may have reactivated. Regular review keeps your strategies aligned with current data.
Klaviyo’s predictive analytics represent a meaningful advantage for brands that take the time to understand and use them. They let you act earlier, target more precisely, and allocate your marketing resources where they are most likely to generate a return.
If you want help building predictive-based segments or incorporating Klaviyo’s AI insights into your flow strategy, we would be glad to help you get there.
Continue reading: RFM segmentation in Klaviyo: A practical walkthrough and Understanding your Klaviyo dashboard: The metrics that actually matter.