Browse Prior Art Database

# System and Method for Business Performance Benchmarking using Probability Distributions of Multiple KPIs

IP.com Disclosure Number: IPCOM000182993D
Original Publication Date: 2009-May-12
Included in the Prior Art Database: 2009-May-12
Document File: 2 page(s) / 32K

IBM

## Abstract

Disclosed is a computer implemented method for business performance benchmarking that uses probability distributions of multiple performance metrics comprising: (1) Building relative benchmark database by computing probability distributions (2) Calculating relative positions of performance metrics of a target company in peer group using cumulative distribution.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 55% of the total text.

Page 1 of 2

System and Method for Business Performance Benchmarking using Probability Distributions of Multiple KPIs

Benchmarking is one of the most important techniques for business performance analysis and improvement because it provides a guidance for improving business performance from strategic level to operational level. A typical benchmarking process involves comparing actual business

performance with benchmark data (e.g., mean or superior of the sample data). Typical

challenges and pain points for benchmarking a single KPI are due to the fact that difference between actual performance value and benchmark data (e.g., mean value or superior) cannot accurately reflect the relative gap (with respect to other peer companies in the industry). For example, if a firm's inventory turn is 12, industry average is 10,and the best practice is 20, but where does the firm stand with respect to all other peer companies in the industry? Would it be 70% or 80% or 99% (2nd best)? Typical challenges and pain points for benchmarking for multiple KPIs are due to the fact that different KPIs cannot be accurately compared as they have different units and measurement methods (e.g., service level (%), lead time (days), number of FTE per \$billion revenue, transportation cost (\$)

pe

\$billion revenue etc..). Prior Art benchmarking methods typically compute the net distance between the actual value and benchmark data is used as the performance gap. Also, weighted average of all different KPI values are used as comprehensive evaluation result in prior art, and probability distributions of sample data in peer group are rarely used.

A computer implemented system and method is disclosed that first builds a benchmark database by characterizing a probability distribution (with parameters) for each KPI based on the sample data in a peer group, and normalizes the benchmark data (e.g., average,

parity, superior,

advantage) to 0%-100% based on the probability distribution (rather than simple mean value). It then calculat...