Toolkit · Playbook
Classic RFM tells you who has been valuable. Our Value × Intent (RFMB) model adds what each person is doing right now, so you know who to target first, with what message, and whether they need a discount at all.
Why add intent
RFM analysis scores customers on Recency, Frequency and Monetary value. It's one of the most useful ways to split a customer list, because it separates your best customers from one-time buyers and lapsing ones using data every store already has.
What it can't see is intent. Two customers with the same RFM score get the same message, even if one of them started checkout an hour ago and the other hasn't visited in weeks. And everyone who hasn't bought yet scores the same, so your warmest prospects sit next to people who bounced from a blog post.
The Value × Intent model keeps RFM for who the customer is and adds a Buying Stage score (B1 to B5) for what they are doing right now, based on the furthest step they reached in the last 30 days. Together they turn audience segmentation from a report into a list of actions.
Customer segmentation model
RFM tells you who the customer is. Buying Stage tells you what they are doing right now. Together they decide who to talk to, and when.
Hexagon
IntelligenceeCommerce
R, F and M describe the relationship so far. B describes intent in the last 30 days. Monetary is calibrated to each store's average order value.
| 1 | 2 | 3 | 4 | 5 | |
|---|---|---|---|---|---|
| RRecencydays since last session | 365+ | 180–364 | 90–179 | 30–89 | 0–29 |
| FFrequencytransactions | 0 | 1 | 2–3 | 4–5 | 6+ |
| MMonetarytotal revenue, × store AOV | No revenue | Up to 0.5× | 0.5× – 1.5×the AOV sits here | 1.5× – 3× | Above 3× |
| BBuying Stagefurthest step in 30 days | Session onlyToFu · Visitors | view_itemMoFu · Product viewers | add_to_cartMoFu · Add to cart | begin_checkoutBoFu · Checkout stage | purchaseAfter purchase |
F ≥ 2 · bought at least once→rules run in priority order; the first match wins
Grow
R4–5 · F≥3 · FM4–5Best customers, buying recently and often.
R4–5 · F≥4 · FM3Buy often at moderate value. Retain and upsell.
R4–5 · F3 · FM3Recent repeat buyers. Turn the habit into loyalty or a subscription.
Develop
F2 · R3–5One recent purchase. The job is order number two.
R3 · F≥3 · FM3–5Good customers starting to slow down.
F2 · R2One purchase, going quiet. Last call before they go.
Win back
R1–2 · F≥3 · FM3–5Used to buy often, now gone quiet.
R1 · F≥3 · FM4–5High value, long absence. Own reactivation budget.
F2 · R1One purchase, 365+ days. Low-cost reach-out, then stop investing.
Priority for repeat buyers: Champions → Can't Lose Them → At Risk → Needing Attention → Loyal → Potential Loyalist.
F = 1 · never bought→segmented by R × B
Acquire
R3–5 · B3–4Never bought, but filled a cart or started checkout.
R3–5 · B2Viewed products recently.
R3–5 · B1Recent sessions, no eCommerce events yet.
R1–2 · any BCame once, long ago. A read on traffic quality.
The segment says who. The Buying Stage says when. The combination decides what to send, and whether a discount is needed at all.
Champions / LoyalB3–B4
Loyal and in the middle of a purchase. Cart or checkout reminder, high priority, no discount.
Champions / LoyalB5
Just bought. Ask for a review, suggest a complement, exclude from discounts and acquisition ads.
At Risk / Can't LoseB2–B3
A dormant customer is back on the site. Reactivation offer now, while intent is live.
At Risk / Can't LoseB1
Back, but no buying signal. Low-pressure reactivation by email or SMS.
New / PromisingB3–B4
Second order halfway there. Second-order nudge with the next logical product.
Potential Loyalistno subscription
Repeat buyer, not subscribed. Subscribe-and-save flow showing the yearly saving.
Hot ProspectsB3–B4
Never bought, filled a cart. First-purchase incentive.
Warm ProspectsB2
Viewed products. Remarketing with the exact products viewed.
Every user gets a code:5554= Champions · B4a best customer with checkout started and not finished.
Hexagon Intelligence · eCommerce
Implementation
You don't need a data warehouse to run this model. You need order data, onsite behavior tied to a known email or phone, and a place to send from. Here's the order we build it in.
view_item, add_to_cart, begin_checkout and purchase, or Klaviyo's onsite metrics (Active on Site, Viewed Product, Added to Cart, Started Checkout, Placed Order). These give you R, measured as days since the last session, and B, the furthest step in the last 30 days.The scales in the model are a starting point. Level 1 stays fixed (365+ days, no transactions, no revenue) so it keeps reading traffic quality the same way over time. Levels 2 to 5 should follow your data:
Score R, F, M and B from 1 to 5. Combine F and M as FM = round((F + M) ÷ 2) and keep B outside the average, because intent changes daily and value doesn't. Then run the segment rules in priority order so each person lands in exactly one segment: buyers (F ≥ 2) into the nine buyer segments, people who never bought (F = 1) into the four prospect segments by R × B.
add_to_cart in the last 30 days with no purchase) and share them with Google Ads. Sync buyer segments from your ESP to Meta and Google as customer lists, for exclusions as much as targeting.Buying Stage moves every day, so B should update daily or run as live segments. R, F and M can refresh weekly. Review the thresholds every quarter and after your biggest sales peak, and track how many people move between segments each month. Movement from New / Promising to Potential Loyalist is the clearest sign the program is working.
Turning segments into decisions
Each group has one job. Judge it on the metric that matches that job, not on open rates.
Protect margin and raise customer lifetime value. Early access, complements, subscriptions and review requests. Keep them out of discount campaigns and acquisition ads.
Win the second order, where most of the profit sits. Post-purchase education, the next logical product and timing based on your reorder gap, before any coupon.
Spend where value is high. Give Can't Lose Them their own reactivation budget, act fast when an At Risk customer shows intent, and stop paying to reach the long-lost.
Match the push to the stage. A first-purchase incentive for Hot Prospects, product remarketing for Warm ones, and nurture for Visitors. Dormant Visitors tell you about traffic quality.
Channel plan
A starting plan. Adjust it to your consent rates, margins and what each channel costs you per send.
| Group | SMS | Paid audiences | |
|---|---|---|---|
| Grow | Product news, early access, review and referral asks | Only for high-intent moments, such as a started checkout or a restock | Exclude from prospecting; use as a seed for lookalike audiences |
| Develop | Post-purchase series and second-order nudge | One timed reminder near the reorder gap | Light retargeting with the next logical product |
| Win back | Reactivation series, new arrivals, what changed | Short offer when they return to the site (B2–B3) | Customer-list retargeting for At Risk and Can't Lose Them only |
| Acquire: Hot and Warm | Browse and cart abandonment flows | Cart reminder for subscribers with SMS consent | Retargeting with the exact products viewed |
| Acquire: Visitors and Dormant | Welcome series for new subscribers | Not used | Broad prospecting; exclude buyers |
SMS only goes to people who opted in to SMS. Confirm your consent setup with counsel.
FAQ
RFM segmentation groups customers by Recency (how recently they were active), Frequency (how many orders they placed) and Monetary value (how much they spent). Each is scored from 1 to 5, and combinations of scores become segments such as Champions, At Risk or New customers. eCommerce teams use it to decide who gets which email, SMS, offer or ad audience, using order data they already have.
It adds a Buying Stage score, B1 to B5, based on the furthest step someone reached in the last 30 days: a session, a product view, an add to cart, a started checkout or a purchase. RFM says who the customer is. Buying Stage says what they are doing now. The combination tells you when to act and whether a discount is needed, and it also segments people who haven't bought yet.
Start with the bands in the model, then check how your customers spread across each level and adjust levels 2 to 5 to your own percentiles and reorder cycle. Keep level 1 fixed so it keeps measuring traffic quality the same way. Set Monetary in multiples of your average order value, so a typical single order lands in level 3 whatever your price point.
Update Buying Stage daily, or use live segments, because intent changes from one day to the next. Recency, Frequency and Monetary can refresh weekly for most stores. Review the thresholds quarterly and after big sales peaks such as Black Friday, when a burst of new customers can shift your distribution and push people into the wrong levels.
Yes. In Klaviyo you can approximate each rule with built-in conditions such as order count, revenue and recent activity, or sync calculated scores to profile properties and segment on those. Shopify's customer segment editor filters on number of orders, amount spent and last order date. Intent from GA4 events can feed ad audiences for visitors who haven't identified themselves yet.
New playbooks go to subscribers of The Retention Brief first, with one practical retention or profit idea per issue. No more than twice a month. Unsubscribe anytime.
The Free Profit Audit report includes a "Who to target first" section built on this model, alongside your biggest profit leaks and a 90-day plan.