⚙️How Customer Score Calculation Works
This application calculates customer scores based on historical sales data, taking into account various customer actions like completed purchases, returns, refunds, and resends.
Last updated
This application calculates customer scores based on historical sales data, taking into account various customer actions like completed purchases, returns, refunds, and resends.
Last updated
When calculating the customer's score, the system gathers all historical orders linked to the customer’s postal code. It skips customers who have only unprocessed (open) orders, as they don't yet provide meaningful insights.
Each order can have multiple events or statuses, including:
Processed (completed) – Positively affects the score.
Returned – Negatively affects the score.
Refunded – Negatively affects the score.
Resent – Negatively affects the score.
Open (pending) – Not included in the final calculation.
Recent customer activity is more valuable in assessing their current trustworthiness. The app reduces the impact of older orders gradually. The older an order gets, the less influence it has on the customer's current score.
For example:
Orders placed recently (e.g., in the past few weeks) significantly impact the score.
Orders from several months or years ago have a smaller influence.
This ensures the customer’s current behavior is most influential, providing a fairer and more relevant score.
The system calculates scores separately for each event type:
Processed Score – Total value of completed orders (positive impact).
Returns Score – Total value of items returned (negative impact).
Refunds Score – Total value refunded to customers (negative impact).
Resends Score – Total value of orders resent (negative impact).
Each score considers the age of orders, meaning newer events contribute more to the final customer score than older events.
You can control how negatively different customer behaviors affect their score. In your app settings, you set the "weight" or significance of each negative action (returns, refunds, resends).
Setting a low weight (e.g., 10%) means the behavior has only a slight negative impact.
Setting a high weight (e.g., 90%) greatly increases the negative impact on the customer’s score.
To calculate a customer's final score, the system takes the total positive value of recent processed orders and reduces it by the adjusted negative impacts from returns, refunds, and resends.
This means:
Customers with mostly positive recent actions will score high.
Customers who frequently request refunds or return items will score lower.
The final score is presented as a percentage (0%–100%), which is easy to understand:
100% indicates consistently excellent customer behavior.
0% indicates problematic behavior (frequent refunds, returns, etc.).
The percentage score is translated into clear ratings for easy interpretation:
Score Range
Rating
Meaning
0–25%
Bad
High frequency of negative customer behavior.
26–50%
Mid
Moderate concerns with some negative behavior.
51–75%
Good
Generally positive customer behavior.
76–100%
Great
Excellent and trustworthy behavior.
If there's not enough historical data, the app marks the customer as "Unknown."
No data
Unknown
Insufficient data to assess behavior.