Customer Segmentation Using LTV: Champions, At-Risk, and Lost

Most eCommerce brands treat their customer base as a monolith. They send the same newsletter to a customer who has bought twelve times and a customer who bought once two years ago and never came back. The result is mediocre open rates, high unsubscribes, and a huge amount of marketing budget wasted on customers who've already churned.

LTV-based segmentation fixes this. By dividing your customers into groups based on their predicted lifetime value and their current engagement state, you can send the right message, at the right time, to the right person — and stop spending money on customers who were never going to buy again.

Why Predicted LTV Is Better Than RFM Alone

RFM (Recency, Frequency, Monetary value) is the traditional segmentation framework. You score each customer on how recently they bought, how often they buy, and how much they spend — then combine those scores into segments.

RFM is backward-looking. It tells you what a customer has done. Predicted LTV tells you what they're likely to do next — which is a fundamentally more useful signal for making marketing decisions.

For example, a customer who bought once for €240 six months ago might have a high RFM monetary score but a low predicted LTV (because high one-time orders that aren't repeated often indicate gift purchases or deal-seeking behavior). A customer who has bought five times for €45 each over 18 months has a lower AOV but a much higher predicted LTV — because the data tells us they'll probably keep buying.

The Four Core Segments

LTV Analyzer produces four segments based on predicted lifetime value and purchase probability — matching the framework used by most retention-focused eCommerce teams.

1. Champions

Your best customers. High predicted LTV, high purchase probability, bought recently. These customers account for a disproportionate share of your revenue — typically 60–80% of total revenue from 10–20% of customers.

What they look like: Multiple purchases in the past 6 months, above-average AOV, high repeat probability (>70%).

Marketing playbook:

• Early access to new collections or products — they feel valued, you get feedback before full launch
• VIP perks: free shipping, extended returns, handwritten thank-you notes
• Referral program invitations — Champions refer higher-quality customers than average
• Avoid discounting: these customers will buy anyway. Discounts train them to wait for sales
• Seed them into Google Ads / Meta lookalike audiences — they're your best acquisition signal

2. Loyal Customers

Solid, repeating buyers. Not your highest-value customers, but reliable. Their purchase probability is good but not exceptional — they respond well to reasons to buy and gentle nudges.

What they look like: 2–4 purchases over 12–18 months, near-average AOV, repeat probability 40–70%.

Marketing playbook:

• New product launch emails — they're warm enough to be early adopters
• Cross-sell: "customers who bought X also love Y" campaigns based on actual purchase patterns
• Loyalty points that accumulate toward meaningful rewards (not 5% discounts — those feel cheap)
• Replenishment reminders if they buy consumables
• Upgrade campaigns: introduce them to your premium product line

3. At-Risk Customers

Previously good customers who are showing signs of churning. They haven't bought recently, but their historical behavior suggests they have real value if you can win them back before the relationship goes cold.

What they look like: No purchase in 60–120 days (depending on your typical purchase frequency), previously had 2+ orders.

Marketing playbook:

• Win-back email sequence: send at 60, 75, and 90 days of inactivity
• First email: "We miss you" — warm, no discount, feature new arrivals or bestsellers
• Second email: soft offer (10–15% off, or free shipping)
• Third email: stronger offer or loss-framing ("Your loyalty points expire in 14 days")
• Suppress from brand awareness campaigns to avoid wasting budget on people already in your funnel

4. Lost (Churned) Customers

Customers who have almost certainly stopped buying from you. Low purchase probability, long inactivity. Trying to win these back with standard campaigns is usually not worth the cost.

What they look like: No purchase in 150+ days, purchase probability <10%, no engagement with win-back campaigns.

Marketing playbook:

• Remove from regular email sends to protect your sender reputation (high unsubscribes and spam reports hurt deliverability for everyone)
• One final "we're cleaning our list" email — these actually perform surprisingly well because they feel honest
• Sunset: move to a monthly digest at most, or suppress entirely
• Do not include in lookalike audiences for acquisition — you'd be optimizing toward customers who didn't stay
• If they're genuinely lost, accept it. The budget is better spent on At-Risk customers who are salvageable.

Building Your Retention Playbook

Once you have segments, the goal is to build automated flows that move customers in the right direction — from Loyal to Champion, from At-Risk back to Loyal, and from Lost to gracefully removed.

How Many Customers Do You Need?

Segmentation is useful from as few as 200–300 customers with at least one repeat purchase. The segments will be small, but the playbooks work at any size — you're just running them manually rather than via automation.

At 500+ customers with repeat purchase history, probabilistic models (BG/NBD) start producing reliable individual predictions. At 2,000+ customers, you have enough in each segment to run proper A/B tests and measure the impact of your retention campaigns.

How Often to Update Segments

Customer segments should be refreshed monthly. A loyal customer can drift into at-risk territory in six weeks; an at-risk customer can snap back with one purchase. Static segments that haven't been updated in three months are almost certainly sending the wrong messages to the wrong people.

Practical cadence:

• Monthly: Refresh segments, update email suppression lists, check win-back conversion rates
• Quarterly: Audit the segment size distribution — if your Champions segment is shrinking or your Lost segment is growing, you have a retention problem that needs to be addressed at the product or experience level
• Annually: Recalibrate what "at risk" means for your brand based on updated purchase frequency data

Using LTV Segments for Paid Acquisition

Your Champions segment is more valuable than any third-party audience you can buy. Here's the acquisition strategy:

1. Export your Champions segment as a customer list (email addresses)
2. Upload to Google Ads Customer Match and Meta Custom Audiences
3. Create a lookalike audience based on Champions (typically 1–5% lookalike)
4. Run acquisition campaigns targeting the lookalike, not cold audiences
5. Exclude your existing customer list from acquisition campaigns to avoid wasting spend

Brands that run lookalike campaigns seeded from high-LTV customers typically see 20–40% lower CPAs and 15–30% higher LTV from the resulting new customers compared to interest-based or broad targeting.

FAQ

What if most of my customers have only bought once?

This is very common, especially for brands under 2 years old or in low-frequency categories. For single-purchase customers, the model predicts whether they're likely to buy again (probability alive) rather than expected future orders. Your key segment in this case is "high-probability repeat buyer" — these are the customers to invest retention budget in before they decide not to come back.

Should I run different creatives for each segment?

Yes, ideally. Champions respond to exclusivity and community, not discounts. At-risk customers often need a tangible reason (offer or new product) to re-engage. Loyal customers respond to FOMO and social proof. The same ad or email to all three segments will underperform what you'd get from even rough personalization.

How do I import these segments into Klaviyo or Mailchimp?

LTV Analyzer exports segment data as a CSV with a segment label column. Import it into your email platform as a list or use it to update a custom property on each contact. In Klaviyo, you can create dynamic segments based on a custom property like ltv_segment = "Champions".

What's the difference between this and RFM segmentation?

RFM groups customers based on what they've done in the past. LTV segmentation groups them based on what they're likely to do in the future. Both are useful: use RFM for understanding your current customer mix, use LTV segments for deciding where to invest retention and acquisition budget going forward.