At Klar, we use advanced data models to understand the purchase behavior of your customers.
This article explains how we categorize customer status: Frequency State and Recency State.
1. 📈 Customer Frequency State: How Often You Buy
The Frequency State measures two things: how many times your customer has purchased and the statistical probability of them making their next purchase, based on historical customer data.
Frequency State Categories
State | Definition & Criteria | What it Means |
Evangelist | 8 Orders AND the probability of the next purchase is > 75%. | Your most engaged and consistent buyers. |
Loyal | 8 Orders OR (4 Orders AND the probability of the next purchase is > 50). | Highly frequent buyers with a strong commitment. |
Repeat | At least 2 Orders (but doesn't meet Loyal/Evangelist criteria). | Customer purchased more than once. |
One-Time Buyer | Exactly 1 Order. | Single purchase customers. |
How Probability is Calculated
The most sophisticated part is the Time-Adjusted Repeat Order Probability. This goes beyond a simple order count by considering time:
Median Lag: We find the typical time (in days) between purchases for customers who have reached the same number of orders.
Lapsed Customers: We compare customers who successfully made another purchase versus those who didn't, even after waiting longer than the Median Lag.
The Result: The probability score indicates the statistical likelihood that your customer is on track to make their next purchase. This score is what helps us differentiate a high-volume, active Loyal customer from a high-volume, lapsing one.
2. ⏳ Customer Recency State: How Timely You Buy
The Recency State measures how recent the last purchase was, but it’s always measured relative to the Frequency State Group.
It answers the question: Compared to other customers who buy as often, what is the time in between orders here?
Recency State Categories
State | Definition & Criteria | What it Means |
Active | The time since the last purchase is within the expected buying window of the frequency group. | Your customer is currently buying on or ahead of your shops typical schedule. |
At Risk | The time since the last purchase is slightly past the typical buying window (between 0.2 and 0.8 Standard Deviations past the median). | The customer is slightly behind your shops expected purchase schedule. |
Defected | The time since your customers last purchase is significantly past the typical buying window (more than 0.8 Standard Deviations past the median). | Your customer has lapsed past the point where most customers in the group would have purchased again. |
Reactivated | Your customer was previously Defected but has since made a new purchase, returning the customer to an Active status. | Your customer has returned after a significant break! |
How We Set the Recency Benchmark
Recency is calculated using statistics unique to your group (e.g., all 'Loyal' customers):
We calculate the Median Time Lag and the Standard Deviation of the time between purchases for all customers in your shops Frequency States.
These statistics define the normal buying window for your shops groups.
The longer the time since the last order is past the median time lag, the higher the risk classification (At Risk or Defected).
This model ensures a customer who typically buys every 30 days is flagged as At Risk much faster than a customer who typically buys every 90 days.