The customer's own rhythm
Recency is divided by the customer's usual interval between visits, so a weekly shopper and a quarterly shopper trigger the alarm at different times.
The customer lifecycle answers one question: which customers are leaving before they have left. The system sorts customers into 12 segments, calculates 60 metrics and compares the gap between purchases with the customer's own rhythm, not one threshold for everyone. The calculation runs in the ERP module.
How it differs from a typical CRM
Visit instead of receipt
Two receipts on one day count as one visit.
A ladder with a reason
Each customer keeps the reason for the segment in words, plus the history of transitions.
Split by trading chain
The calculation runs per pair "customer and chain": in one chain a customer can be regular, in another lost.
The ladder of checks
Conditions are checked from top to bottom, and the first match wins. Special cases go first, then the rhythm.
Special cases
No visits, a return after leaving, a single visit. This is where "Dormant", "Reactivated", "New" and "One-time buyer" are decided.
Recency and few visits
More than 365 days without a purchase: "Lost". Fewer than three visits a year: "Developing".
Rhythm
First the irregularity of intervals is checked, then the deviation from the rhythm: above 1.5 is "At risk", above 2.5 "Leaving", above 4 "Lost".
Loyal or regular
A yearly total in the top 20% of the chain: "Loyal", otherwise "Regular".
12 segments and an action for each
- NewRuleone visit, recency up to 30 daysRecommended actiona welcome offer, without overdoing discounts
- One-time buyerRuleone visit, recency 31–365 daysRecommended actionan incentive for a second purchase in the favourite category
- Dormant one-time buyerRuleone visit, recency over 365 daysRecommended actioncheap channels such as e-mail, no discounts
- DevelopingRule2 visits, or exactly 3 with a normal rhythmRecommended actionhelp them reach 3–4 visits
- RegularRule4 or more visits, normal rhythm, total below the thresholdRecommended actiongrow the basket, not the frequency; no discount needed
- LoyalRulethe same, total in the top 20% of the chainRecommended actionprotect them, do not flood with discounts
- At riskRulerhythm deviation 1.5–2.5Recommended actiona light message, not a discount right away
- LeavingRuledeviation 2.5–4Recommended actionan offer and a call from the candidates list
- LostRuledeviation above 4 or recency over 365 daysRecommended actiona strong offer, track the return
- ReactivatedRulecame back after leaving, 60 daysRecommended actionthanks and an offer for a second purchase
- UnpredictableRule3 or more visits, interval spread above 0.8Recommended actionwork by events and category, not by schedule
- SleepingRulehas a card, no purchasesRecommended action"activate your card"
All thresholds (30 days for new, 1.5 / 2.5 / 4 for alarm, leaving and loss, 365 days and others) are defaults and can be configured for each trading chain.
RFM and ABC/XYZ
Customer value in one matrix
RFM gives scores from 1 to 5 for recency, frequency and amount, by the chain's quintiles. ABC shows the customer's share of revenue (limits 80% and 95%), and XYZ shows purchase stability (limits 10% and 25%). A separate report shows who moved up or down in value since last year.
60 customer metrics
A profile you can verify
The metric groups: history and rhythm, money and value, basket and categories, windows and trends. Any metric can be recalculated from the customer's receipts.
12 reports
- Customer base by segment
- Segment transition matrix
- Reactivation candidates
- Customer segment history
- Cohort retention
- Reactivation response
- Offer effect
- RFM matrix
- Customer profile
- Category preferences
- Customer metric dynamics
- ABC migration
An honest word about the start: the reports on transitions, cohorts, response, offer effect, dynamics and ABC migration fill up as calculation history accumulates. In the first month they may be incomplete, and some segments, for example "Sleeping", may stay empty.
From segment to action
A segment change is an event that starts a rule: a reactivation campaign can be built without a programmer, with a message or bonuses. A segment also works as a condition in any discount rule.
A control group is built into the offer: a customer is assigned to a group by a hash of the identifier and always stays in it. The "Offer effect" report compares the offer with the control group.
Deactivating lost customers comes with safeguards: a customer is deactivated only if lost in all chains, preview is on by default, the job is switched off, and an unset term stops the process.
Frequently asked questions
Why are segments better than a single "90 days without a purchase" threshold?
The threshold takes the customer's rhythm into account. Someone who shops once a quarter does not land in "Leaving" after a month's pause.
Can I change the thresholds?
Yes, thresholds are configured for each trading chain.
Will all the reports be filled in from the start?
No. Some reports show dynamics and fill up as calculation history accumulates.
How do I launch a campaign for customers who are leaving?
Create a rule with a segment condition and the "segment change" event, add the message, and take the audience from the "Reactivation candidates" report.