Content:
SHARE:
Subscribe and don't miss anything important
Don't worry, we don't spam 😍
Klaviyo retention playbook · Shopify
Most Shopify segmentation looks backwards. It groups customers by what they bought, how much they spent or when they last opened an email. Klaviyo’s predictive analytics adds a forward-looking layer: who is likely to spend more, who is drifting away and when a customer may be ready to order again.
The model is useful, but it is not magic. A predicted value is most reliable across a group, not as a promise about one person. The strongest programmes use predictions to choose timing and treatment, then keep behavioural guardrails around every segment.
What Klaviyo Predictive Analytics Includes
| Field | What it means | Best use |
|---|---|---|
| Historic CLV | Past order value after refunds and returns. | VIP recognition and historical value bands. |
| Predicted CLV | Estimated spend over the next year. | Prospective value, acquisition audiences and service levels. |
| Total CLV | Historic plus predicted CLV. | Broader customer-value tiers. |
| Churn risk | Probability of churn based on order number and frequency. | Prioritized winback and retention. |
| Expected next order date | A forecast based on the customer’s and account’s order patterns. | Timing replenishment or repeat-purchase messages. |
| Average time between orders | Average days between that customer’s purchases. | Cadence checks and custom timing windows. |
Klaviyo retrains its CLV model at least weekly. It also warns that individual predictions are not exact. Use a predicted CLV threshold to shape a segment, not to decide that one named customer is “worth exactly $184.”
Before You Build: Check Eligibility and Data Quality
Klaviyo currently requires:
- At least 500 customers who placed non-zero, non-cancelled, non-refunded orders
- An ecommerce integration such as Shopify, or valid Placed Order events sent by API
- At least 180 days of order history and orders within the last 30 days
- At least some customers with three or more orders
If a profile’s predictive panel is blank, Klaviyo may have enough account-level data but not enough information about that person. Before using CLV, confirm refunds and order value sync correctly. If Placed Order events come through a custom API, the actual order value must be passed in the $value field.
How to Choose Your Thresholds
Do not copy a dollar threshold from another store. Export customers with CLV fields, then review the distribution. A sensible starting point is to create bands around your own percentiles:
Recalculate bands quarterly or after major changes to price, assortment or acquisition mix.
7 Klaviyo Predictive Segments to Build
1. High-Potential New Customers
Find recent first-time buyers with predicted CLV in your top band. The goal is to earn the second purchase before the relationship cools.
Use education, product pairing, post-purchase care and social proof. Avoid leading with a large discount when the customer already has high predicted value.
2. At-Risk VIPs
Combine strong historic or total CLV with elevated churn risk. This prevents the winback budget going equally to a lapsed one-time discount buyer and a genuinely valuable repeat customer.
Lead with service, newness, early access or a personal recommendation. Use a discount only when margin and prior behaviour justify it.
3. Next Order Due Soon
Reach customers shortly before Klaviyo’s expected next order date. This works well for stores with repeat behaviour but without rigid product-level replenishment cycles.
Use a reminder, new arrivals or a convenient reorder path. Keep the message useful; Klaviyo advises against a repetitive countdown sequence before every predicted order.
4. Overdue High-Intent Buyers
Separate customers whose predicted date has passed from people who are merely inactive. They have a clearer expectation gap.
Test a short winback sequence: reminder, objection handling, then a controlled incentive. Exit immediately after purchase.
5. Predicted VIP, Not Yet a VIP
This audience has high predicted CLV but has not crossed your historical VIP threshold. Treating them better early can be more effective than rewarding value only after it arrives.
Offer priority support, personalized recommendations, loyalty education or early access. Measure repeat purchase and margin, not open rate alone.
6. Predictive Cross-Sell Audience
Use expected next order timing with actual product history. The prediction chooses when; purchase data chooses what.
Show the connection between products, not a random catalogue. Exclude unavailable items and customers already in a higher-priority service or winback flow.
7. Low-Predicted-Value Efficiency Segment
Prediction can improve restraint as well as targeting. Identify low-predicted-value customers with weak engagement so you do not spend your richest incentives or highest send frequency on them.
Reduce cadence, exclude from expensive offers and test lower-cost channels. Do not suppress solely because of one score; consent, deliverability and recent behaviour should remain part of the rule.
How to Activate the Segments
Use segments in four places:
- Flows: trigger second-purchase, due-soon and winback journeys.
- Campaigns: vary offer, message and product selection by value tier.
- Forms: show appropriate onsite messages to due-soon or CLV groups.
- Paid audiences: sync valuable or at-risk groups for coordinated retargeting where consent and platform rules allow.
Cross-channel coordination matters. Our guide to using paid social and email together shows how retention audiences can support media efficiency without flooding the same customer.
How to Test Predictive Segmentation
- Choose one audience and one business outcome.
- Create a rule-based comparison group where possible.
- Hold out a random share from the predictive treatment.
- Keep creative, offer and send time consistent.
- Measure conversion, revenue per recipient, contribution margin and unsubscribe rate.
- Wait long enough to capture the expected purchase window.
A high open rate is not evidence that the prediction created value—especially with Apple Mail Privacy Protection inflating opens. Use clicks and purchases, and review our guide to interpreting and improving email engagement.
Common Mistakes
Also remember that better segmentation cannot rescue weak fundamentals. Compare platforms and setup requirements in our Klaviyo vs Mailchimp for Shopify guide, then make sure consent, deliverability and core flows are sound.
Frequently Asked Questions
Does every Shopify store get predictive analytics?
No. Klaviyo requires sufficient customers, order history, recent orders and repeat-purchase data. The predictive panel can also be blank for an individual profile.
How often does predicted CLV update?
Klaviyo says its CLV model is retrained at least weekly.
Can expected order date replace replenishment flows?
Not always. If product categories have known and different replenishment cycles, Klaviyo recommends product-triggered flows with appropriate delays.
