Vendor score calculation

The vendor score determines the vendor rank based on four key factors. This is calculated using Gen AI:

  • Volume final score: The volume score represents the total received goods amount a company purchases from a specific supplier within the specified period. The score is calculated using line counts and the total received amount and these values are normalized as:

    Volume Score = 50% of the normalized line counts + 50% of the total normalized amount

    The final score is adjusted to a range of 0 to 200, where a lower score indicates a lower vendor rank, and a higher score indicates a higher rank.

  • Fill rate final score: The vendor fill rate score measures the percentage of customer orders that a supplier fulfills completely and on time. Orders are considered for calculation only if:
    • Transaction type: PO
    • Stage codes: 5, 6, or 7
    • Quantity received < Quantity ordered
    • PO Suffix > 0

    The fill rate percentage is determined by the ratio of Filled Counts to Total Ordered Counts based on the above conditions. The final score is normalized within a range of 0 to 200, where a lower score indicates a lower vendor rank, and a higher score indicates a higher rank.

  • Item quality final score: The item quality score measures how well a product meets customer expectations in terms of performance and durability. A high number of returned goods transactions indicate quality issues, such as defects or customer dissatisfaction. The calculation considers transactions with these conditions:
    • Transaction type: RM
    • Code identity: M
    • Usage flag: True

    The transactions count is compared to the total line item counts, and the values are normalized. The final score is adjusted between the range of 0 to 200, where a higher score indicates fewer reported quality issues, and a lower score indicates potential product concerns.

  • On time delivery final score: The on-time delivery score measures the ability of a vendor to deliver goods or services by the agreed date or within the expected timeframe. The calculation is based on the difference between the due date and receipt date of a purchase order. The delay in the deliveries is classified as follows:
    • Late by 1 day
    • Late by 2–5 days
    • Late by 6–10 days
    • Late by 11–15 days
    • Late by more than 15 days

    The counts for each of these categories is normalized against the total deliveries, and Fibonacci numbers (1, 2, 3, 5, 8...) are applied in case there is a further delay in the order deliveries. The final score is adjusted between the range of 0 to 200, where a higher score indicates better on-time delivery performance, and a lower score indicates frequent delays.

The vendor score is the sum of these four factors, with a maximum of 800:
Factor Weightage Score range
Volume final score 25 0-200
Fill rate final score 25 0-200
Item quality final score 25 0-200
On time delivery final score 25 0-200

Vendors are classified into Excellent, Good, Average, or Below Average ranks, where a score of 0 indicates a low rank and 800 indicates a high rank. The analysis is based on the data of past three years, and vendors are categorized using the K-Means Machine Learning algorithm.