Churn Prediction Software: Check What Klaviyo Already Does

On Shopify and Klaviyo under 10M a year, you likely own churn prediction software already. Klaviyo scores churn risk on each profile once your store clears its data bar. Buy a separate tool only for a signal Klaviyo cannot see.

Most lists of churn prediction tools put Adobe, Salesforce and Braze next to a 2-person Shopify brand. That is the wrong shortlist. The first question is what you already pay for, and whether your data is big enough to feed it.

What is churn prediction software?

Churn prediction software scores each customer on how likely they are to stop buying. Churn means a customer who stops buying from you. The score lets you act before they are gone, not after.

For a DTC store, most of these tools read the same few things: how many orders a customer placed, how long between orders, and how long since the last one. Klaviyo's help center says its churn risk "is based on their number and frequency of orders." Each order lowers the risk. Each week without one raises it.

Do you need churn prediction software if you use Klaviyo?

Probably not at first. Here is where the confusion starts. Klaviyo's own pages do not say the same thing.

Klaviyo source What it says Marketing page Churn risk is calculated "on every customer profile" Help center Predictions show only if your account meets 3 conditions Klaviyo blog The churn model "begins with some rules": the same 3 conditions

The 3 conditions, from the help center:

1. At least 500 customers have placed an order.
2. At least 180 days of order history, with orders in the last 30 days.
3. Some customers who have placed 3 or more orders.

Miss one and the profile shows "does not qualify." So "every profile" is true only for stores past the bar. If you are past it, you already have a churn score. If you are not, a separate tool will not fix it. Every model needs order history, and you do not have enough yet.

Klaviyo shows churn risk as a percent. Its Helpdesk article groups it as Low (33% or less), Medium (over 33% up to 66%) and High (over 66%).

One gap to know. None of these pages says what time window the score covers. Treat it as a ranking, not a date.

What do your options predict, and what do they need?

Option What it predicts Data it needs Cost Klaviyo predictive analytics Churn risk, CLV, next order date 500 buyers, 180 days, some 3-time buyers Plan not stated on the pages we read Shopify predicted spend tier Future spend: High, Medium, Low Over 100 sales Not stated on the page Churnkey Customer Health Who is likely to cancel in the next 30 to 90 days Billing data from Stripe, Chargebee, Paddle or Braintree From $250 a month billed yearly (pricing) RFM segments Who has gone quiet, by your own rules Order data only Free in Klaviyo or Shopify

Note the Shopify row. Predicted spend tier is a spend forecast. Its help page never mentions churn. Do not use it as a churn score.

Note the Churnkey row too. Churnkey's pricing page does not list Customer Health on each plan. Ask before you buy.

RFM means recency, frequency and monetary value. It sorts buyers by how recent, how often and how much. Insider One sells a churn tool, and even it says brands without 6 months of clean history "often get more reliable signal" from RFM. That is the honest default under the data bar.

When should you buy a separate churn tool?

Our view: buy one when the signal you need is not in your order data.

Klaviyo's model reads orders. It does not read failed cards, skipped boxes or cancel attempts. If you sell subscriptions, those are the signals that come first. That is where a billing tool like Churnkey earns its place. Check that it connects to your billing app before you book a demo. Its listed integrations are Stripe, Chargebee, Paddle and Braintree.

Klaviyo's help center adds one more limit. It says the expected next order date "doesn't consider which products the customer ordered". Your products may run out at different speeds. Then a store-wide score is early for some and late for others. Our customer churn analysis article shows how to set the line per product.

Skip enterprise platforms at this size. The setup assumes a data team you do not have.

How accurate is churn prediction?

No page we read gives a sourced accuracy number for a DTC store. The tools say it depends on how much data you have. So test it yourself before you trust it.

1. Build a Klaviyo segment of High churn risk customers.
2. Split it. Send half your winback offer. Hold the other half back.
3. After 30 days, compare repeat orders in each half.
4. If the 2 halves match, the score is not earning its discount.

This is a holdout test. It costs one flow and some patience. Our winback email flow article covers what to send.

What do you do with the score?

A score does nothing on its own. It tells you who to talk to. What you say comes from the moves in our churn prevention article, which puts the discount last. If you sell subscriptions, count churn the way our subscription churn article does first, or you will not know if the score helped.

For how every piece fits together, start at the ecommerce customer retention hub.

Talk it through

Not sure whether your store is past Klaviyo's bar, or whether you need a second tool? Book 30 minutes. Show us your Klaviyo account and we'll tell you what your data can support, whether you hire us or not.

Book the call