How to Use Klaviyo Predictive Analytics for Smarter Campaigns

Most ecommerce email marketing is reactive. A customer places an order, so you send a confirmation. They abandon a cart, so you send a reminder. They stop buying, so you send a win-back email. Every action is a response to something that has already happened.

Klaviyo's predictive analytics flips this model. Instead of reacting to past behaviour, you can anticipate future behaviour — predicting which customers are likely to churn, when they are likely to order next, and how much they are likely to spend over their lifetime. This lets you intervene before problems occur and invest disproportionately in your most valuable customers.

We use predictive analytics in every Klaviyo account we manage, and the results speak for themselves: more precisely timed flows, better-targeted campaigns, and higher revenue per email sent.

This guide walks through each predictive metric, how to access it, and practical ways to use it in your email marketing strategy.

What is Klaviyo predictive analytics?

Klaviyo predictive analytics is a machine learning system built into the platform that analyses your historical customer data to predict future behaviour. It runs automatically in the background — you do not need to configure anything beyond ensuring your ecommerce integration is properly set up.

The system examines patterns across your entire customer base: purchase frequency, order values, product preferences, engagement patterns, and timing. It then generates individual predictions for each customer profile that you can use in segments, flows, and campaigns.

Think of it as having a data scientist who has studied every customer on your list and can tell you, with reasonable accuracy, what each one is likely to do next.

Key predictive metrics explained

Klaviyo provides several predictive metrics for each customer profile. Here is what each one means and why it matters:

Metric	What it predicts	Why it matters
Predicted CLV	Total revenue a customer will generate over their lifetime	Identifies your most valuable customers for VIP treatment
Expected date of next order	When a customer is likely to place their next order	Enables perfectly timed replenishment and re-engagement emails
Churn risk	Probability that a customer will not order again	Flags at-risk customers before they are lost
Predicted gender	Likely gender based on name and behaviour	Enables gender-specific product recommendations
Average days between orders	Typical gap between a customer's purchases	Helps set flow timing and replenishment reminders
Historic CLV	Total revenue generated to date	Combined with predicted CLV, shows full customer value picture

Data requirements

Predictive analytics is not available immediately on new Klaviyo accounts. The machine learning model needs sufficient data to generate reliable predictions:

• Minimum 500 customers who have placed at least one order
• At least 180 days of order history
• Active ecommerce integration syncing order data to Klaviyo (Shopify, WooCommerce, BigCommerce, etc.)

If your account does not yet meet these requirements, focus on building your customer base and ensuring your integration is capturing all order data correctly. The predictive features will activate automatically once the thresholds are met.

Improving prediction accuracy

The more data Klaviyo has, the more accurate its predictions become. You can improve accuracy by:

• Ensuring all order channels sync to Klaviyo (online, POS, wholesale if applicable)
• Maintaining clean customer data — merge duplicate profiles, correct obvious errors
• Running your Klaviyo account for a longer period — predictions improve significantly after 12+ months of data
• Keeping your product catalogue synced so Klaviyo can track product-level purchase patterns

Step 1: Access predictive data

Predictive data appears in several places within Klaviyo:

Individual profiles

1. Go to Profiles in Klaviyo
2. Click on any customer profile
3. Scroll to the Predictive Analytics section
4. You will see predicted CLV, expected next order date, churn risk probability, and other metrics

Segment builder

1. Go to Lists & Segments > Create Segment
2. In the condition builder, select Predictive Analytics as the property source
3. Choose the metric you want to segment by (e.g., Predicted CLV greater than £500)
4. Combine with other conditions to create precise segments

Flow conditional splits

1. In any flow, add a Conditional Split
2. Select Properties about someone > Predictive Analytics
3. Choose the metric and set your threshold

Step 2: Build predictive segments

Predictive segments are where the real power lies. Here are the most valuable segments to create:

High-CLV customers

Condition: Predicted CLV is in the top 20% (or above a specific value)

Use case: VIP treatment, early access to sales, exclusive products, higher-value incentives in win-back flows, loyalty programme enrolment

At-risk customers (high churn)

Condition: Churn risk probability is greater than 0.5 (50%)

Use case: Proactive win-back emails before they fully disengage, satisfaction surveys, personalised offers to prevent churn

Due for reorder

Condition: Expected date of next order is within the next 7 days

Use case: Replenishment reminders, reorder prompts with personalised product recommendations based on past purchases

Overdue for reorder

Condition: Expected date of next order has passed (is before today)

Use case: Gentle nudge emails, "time to restock" messaging, early intervention before they enter win-back territory

For more on building effective segments, see our guide on segmenting your ecommerce email list.

Step 3: Use predictions in flows

Predictive analytics transforms your automated flows from generic sequences into intelligent, adaptive experiences.

Post-purchase flow with CLV split

After a customer places an order, split the post-purchase flow based on predicted CLV:

• High-CLV path: personalised thank-you from the founder, VIP early access to new products, loyalty programme invitation
• Standard path: standard thank-you, product care tips, cross-sell recommendations
• Low-CLV path: focus on building the relationship, educational content, social proof to encourage repeat purchase

Win-back flow with churn risk

Instead of using a fixed time delay to trigger win-back emails, use Klaviyo's churn risk prediction to identify customers who are showing signs of disengagement before they fully lapse. This is more accurate than a one-size-fits-all time delay because it accounts for individual purchase patterns.

For a detailed guide on building win-back flows, see our flow optimisation guide.

Replenishment flow with predicted next order

For consumable products, create a flow that triggers based on the expected next order date:

1. Set the trigger to the Placed Order metric
2. Add a time delay based on the average days between orders (or use a conditional split based on the predicted next order date)
3. Send a reminder email: "Time to restock your [Product Name]?"
4. Include a direct link to reorder the same product

Welcome flow with CLV personalisation

Even in welcome flows, predicted CLV can inform your approach. After a first purchase, Klaviyo begins calculating predictive metrics. By the second or third email in your welcome sequence, you can split based on early CLV signals to tailor the onboarding experience.

Step 4: Use predictions in campaigns

Predictive analytics is equally powerful for one-off campaigns:

Sale and promotion targeting

Instead of blasting your entire list with a sale announcement, use predictions to target strategically:

• High-CLV customers — early access or exclusive preview
• Medium-CLV with high churn risk — the sale could be the nudge they need to stay engaged
• Due for reorder — the sale aligns with their natural purchase cycle
• Low churn risk — these customers will buy anyway, so consider excluding them to protect margins

Content personalisation

Use predicted CLV to determine how much marketing investment each segment receives:

Segment	Campaign approach
Top 10% CLV	Exclusive offers, founder's letter, VIP events
Top 10-30% CLV	Early access, loyalty rewards, personalised picks
Middle 40% CLV	Standard campaigns with targeted recommendations
Bottom 30% CLV	Value-driven content, education, brand storytelling

Budget allocation

When running campaigns that include a discount or free gift, use predicted CLV to allocate your budget. A 20% discount for a customer with a predicted CLV of £2,000 is a worthwhile investment. The same discount for a customer with a predicted CLV of £50 may not be.

Step 5: Optimise based on predictions

Monitor prediction accuracy

Klaviyo's predictions are probabilistic, not deterministic. They are directionally accurate but not perfect. Monitor their accuracy by:

• Comparing predicted CLV against actual revenue over time
• Checking whether "high churn risk" customers actually churned
• Verifying that "due for reorder" timing aligns with actual reorder patterns

Adjust thresholds

The thresholds you set for predictive segments should be reviewed quarterly. As your customer base evolves, the distribution of predicted CLV, churn risk, and order frequency will shift. What constituted "high CLV" six months ago may be average today.

Combine with behavioural data

Predictive analytics is most powerful when combined with behavioural signals:

• High CLV + low email engagement — valuable customer who is disengaging, needs immediate attention
• Low CLV + high engagement — potential for growth, nurture with upsell and cross-sell content
• High churn risk + recent browse activity — still interested but not buying, may need a nudge or incentive

For more on combining data points for smarter marketing, see our managed Klaviyo services overview.

Advanced use cases

Predictive product recommendations

Combine Klaviyo's predictive analytics with its product recommendation engine. For high-CLV customers, recommend premium or higher-AOV products. For at-risk customers, recommend their previously purchased favourites to trigger a comfort repurchase.

Dynamic discounting

Use predicted CLV and churn risk together to determine discount levels dynamically:

• High CLV + high churn risk — highest priority, generous offer (20% or free gift)
• High CLV + low churn risk — no discount needed, focus on exclusivity and recognition
• Low CLV + high churn risk — modest offer or no offer, may not be worth the investment
• Low CLV + low churn risk — focus on growing CLV through upsells and cross-sells

Lifetime value forecasting for business planning

Export Klaviyo's predicted CLV data to forecast future revenue. If your top 1,000 customers have a combined predicted CLV of £2 million, that gives you a revenue baseline for the next 12-24 months. This data informs inventory planning, marketing budgets, and growth projections.

Customer acquisition cost evaluation

Compare predicted CLV against customer acquisition cost (CAC) by acquisition channel. If customers acquired through Instagram have a higher predicted CLV than those from Google Ads, that informs where to allocate your advertising budget.

Predictive churn prevention

Build a dedicated "churn prevention" flow that triggers when a customer's churn risk crosses a threshold (e.g., 40%). This flow sits before the traditional win-back flow and uses softer messaging:

1. Send a "just checking in" email with personalised recommendations
2. Share new content or blog posts relevant to their interests
3. Offer early access to an upcoming product or collection
4. If no engagement, escalate to a more direct re-engagement approach

Common mistakes to avoid

1. Treating predictions as certainties

Predictive analytics provides probabilities, not guarantees. A customer with a 70% churn risk is not definitely leaving — there is a 30% chance they will stay. Use predictions to guide strategy and prioritise resources, not to make absolute decisions about individual customers.

2. Setting and forgetting segments

Predictive metrics change as customer behaviour evolves. A segment based on "predicted CLV greater than £500" may contain different customers next month. Review and adjust your predictive segments quarterly to ensure they remain relevant.

3. Ignoring the data requirements

If your account barely meets the minimum requirements (500 customers, 180 days), the predictions will be less reliable. Do not make major strategic decisions based on thin data. Wait until you have a robust data set before heavily relying on predictive features.

4. Over-segmenting

It is tempting to create dozens of micro-segments based on every combination of predictive metrics. This creates unnecessary complexity and makes it difficult to manage your email programme. Start with 3-5 key predictive segments and add more only when you have a clear use case.

5. Not combining with other data

Predictive analytics alone is powerful, but it is most effective when combined with behavioural data (email engagement, browse activity, cart abandonment), demographic data, and purchase history. Use predictions as one input among many, not the sole basis for your strategy.

---

Klaviyo's predictive analytics is one of the platform's most underutilised features. Most brands are sitting on a goldmine of customer data that Klaviyo has already analysed and turned into actionable predictions — they just have not built the segments, flows, and campaigns to take advantage of it.

The shift from reactive to predictive email marketing is significant. Instead of waiting for customers to disengage and then trying to win them back, you can identify at-risk customers early and intervene. Instead of giving every customer the same experience, you can invest proportionally based on predicted lifetime value.

Start with the basics: build a high-CLV segment, create an at-risk churn segment, and add predictive conditional splits to your most important flows. Refine from there based on what the data tells you.

Need help implementing predictive analytics in your Klaviyo account? See our Klaviyo services or get in touch for a free account audit.