How Should You Use Klaviyo Predictive Analytics for Retention?

Use Klaviyo predictive analytics as a prioritization signal, not an autopilot. Predicted CLV, churn risk and expected date of next order help you spot drifting first-time buyers, size your winback audience and find VIP candidates. They work best across groups of customers, so let real repurchase cycles and product data decide your timing.

Open a customer profile in Klaviyo and there it is: predicted lifetime value, churn risk, the expected date of their next order.

It looks like a crystal ball.

It's not. It's a useful model, and we'd use it. We just wouldn't hand it the keys.

By Emiel Dingemans, Founder of Magicianly

What Is Klaviyo Predictive Analytics?

It's a set of predictions Klaviyo calculates from your order data and shows on each customer profile: how much they've spent, how much they're likely to spend in the next year, how likely they are to churn and when they'll probably order again.

Here's what each field means, based on Klaviyo's guide to understanding predictive analytics, and how we use it.

Klaviyo retrains its CLV model at least once a week, so the numbers move as customers buy.

There's also predicted gender, based on first names. Klaviyo itself calls it an approximation, and we rarely build anything on it.

Does Your Klaviyo Account Qualify for Predictive Analytics?

Only if you have enough order history. Klaviyo needs at least 500 customers who have placed an order, an ecommerce integration or orders sent through the API, at least 180 days of order history with orders in the last 30 days, and some customers with 3 or more orders.

A few details matter here. Those 500 are customers, not profiles, and canceled, refunded or zero-value orders don't count. And if the section shows up on a profile but it's blank, Klaviyo doesn't have enough data on that person to predict anything.

Below that bar, don't worry. You can get most of the value from your own order data. Look at the median days between first and second order, ideally by product, and build timing on that. It's the same approach we use to find the repurchase cycle in our Klaviyo winback flow guide.

How Accurate Are Klaviyo's Predictions?

Accurate enough for groups, not precise for any one person. Klaviyo says it plainly: predictions work best averaged over many customers and aren't expected to be exact for any single individual.

You'll even see numbers like 1.43 predicted orders. That doesn't mean someone will place 1.43 orders. It means one or two is likely, maybe more, maybe fewer. Add up a group of customers and the total starts to make sense.

The bigger limitation for retention: expected date of next order doesn't look at what the customer bought. A 30-day supplement and a 6-month device get treated the same way. If your products have distinct cycles, Klaviyo itself recommends separate Placed Order flows filtered by product, with delays that match the known cycle. That's how we build replenishment emails.

How Should You Use Expected Date of Next Order?

As a gentle second-order nudge for one-time buyers, when you don't have a clear product cycle. Not as a countdown, and not as a replacement for replenishment.

Klaviyo's own guidance lines up with how we'd use it:

• don't count down to the expected date, or repeat customers get the same emails before every order;
• don't let it replace replenishment flows when you know your product cycles;
• if many of your customers already buy again, use it mainly for people who've bought once.

A segment like "expected date of next order in the next 14 days" can work well for a soft reminder or a content-led email for those one-time buyers. Klaviyo even suggests targeting signup forms to people whose next order is expected in the next two weeks.

How Should You Use Churn Risk in Klaviyo?

As a priority flag. High churn risk tells you who's drifting. It doesn't tell you why, and it shouldn't trigger a discount on its own.

How it works: Klaviyo calculates churn risk from the number and frequency of each customer's orders. Every order pushes it down. Time passing between orders pushes it up. When you export it, it shows as a number between 0 and 1, so 0.45 means 45%.

Klaviyo's advanced segmentation reference suggests a churn risk segment for winback. We'd use it, with two adjustments:

• Combine it with engagement. A high-risk customer who still opens and clicks needs a different message than one who's gone quiet everywhere.
• Diagnose before you discount. In winback, we don't hand out discounts automatically. A reminder of what they loved or what's new often does the job.

One more thing. If lots of your customers have only ordered once, your average churn risk will look high. Klaviyo's guide on how to segment by CLV points out exactly that. The fix isn't a bigger winback flow. It's a better post-purchase journey and second order. We go deeper on that in how to reduce customer churn.

How Should You Use Predicted CLV?

To find customers worth protecting. Predicted CLV helps you spot VIP candidates early and stop discounting people who were going to buy anyway.

We define VIPs by value, frequency and recency, not just total spend. Predicted CLV is a useful extra input there, especially for newer customers who haven't spent much yet but look likely to. Here's how we build the VIP customer flow.

The second use is margin. If someone's predicted CLV is high, you probably don't need to discount to bring them back.

Klaviyo also mentions that you can export predicted CLV and add it up for a rough 12-month revenue projection from existing customers. It leaves out customers you haven't acquired yet, so treat it as a planning number, not a forecast you'd bet the budget on.

What Can't Klaviyo Predictive Analytics Tell You?

Why a customer is drifting, what they should buy next and whether your emails caused the next order.

Predictions describe likely behavior. They don't explain it, and they don't prove what your program changed. For that you still need first-to-second purchase rate, cohort revenue over time and, when you can, holdout groups. Our guide to ecommerce retention metrics covers which numbers we'd watch.

The same goes for programs you add on top. If you run a points program, judge loyalty program emails by second orders and margin, not by how members compare with everyone else.

When Should You Use Each Prediction, and When Shouldn't You?

Use predictions to decide who gets attention first. Use real buying behavior to decide what to say and when.

What Are the Most Common Mistakes With Klaviyo Predictive Analytics?

Most of them come from treating a group-level model like an individual forecast.

1. 1Treating one customer's prediction as a promise.
2. 2Building countdown flows to the expected date of next order.
3. 3Using expected date of next order instead of product-based replenishment.
4. 4Sending discounts to everyone with high churn risk.
5. 5Ignoring that lots of one-time buyers push average churn risk up.
6. 6Judging your retention program by predicted CLV instead of actual second orders.

So How Should You Use Klaviyo Predictive Analytics?

As a way to prioritize. Spot the first-time buyers who are drifting, the lapsed customers worth winning back and the high-value customers you shouldn't be discounting.

Then let real buying behavior, product cycles and your own judgment do the rest.

Prediction tells you who to look at.

It doesn't tell you what to say.