# Walkthrough: Calculate normalized scores for KPIs

To calculate an overall performance score for an objective KPI, PerformancePoint Monitoring Server computes a normalized, weighted average. Normalization converts individual KPI scores to a common scale, and makes it possible to create overall performance ratings for a group of KPIs, regardless of how different the individual KPIs are.

This article presents a walkthrough of how to calculate the overall performance score for an objective KPI. To roll up individual KPI scores into a normalized, weighted average for an objective score, follow these steps.

## Step 1 - Understand the scenarios for objective KPIs

Typically, a business scorecard includes multiple KPIs, grouped together in categories under an objective KPI. The objective KPI represents overall performance visually with an indicator graphic.

The problem for objective scores

An objective KPI rolls up individual KPI scores into a value that represents the overall score for a group of KPIs. However, unless the KPI scores represent the same kinds of values, a score for an objective KPI is useless.

Imagine, for example, an objective KPI that rolls up the following KPI scores:

 KPI Score Resolved customer service requests. 76% Number of service requests resolved in 15 minutes or less. 22 Number of service requests that require escalation. 3

One KPI score, Resolved customer service requests, is a percentage (76%), and the other KPI scores indicate the number of service requests. Because the scores are different kinds of values, we cannot combine them to make an estimate of overall performance.

To evaluate overall performance, individual scores must represent the same kinds of measurements. When scores represent the same kinds of measurements, Monitoring Server can calculate aggregate values and summary performance ratings.

Comparable values for individual KPIs

To create comparable values for KPI scores, Monitoring Server converts all the raw KPI scores to a single common scale. In financial applications, the process that converts numbers to a common scale is referred to as normalization; the scaled, comparable value is called a normalized score.

After Monitoring Server converts individual KPI values to normalized scores, it can calculate a meaningful summary value for the objective. To do the calculation, Monitoring Server performs a series of adjustments to the related values. In particular, it adjusts the raw KPI score and also adjusts the thresholds that make up the scoring pattern. To understand the values that Monitoring Server displays in the scorecard, use this walkthrough to learn how to perform these calculations manually. Then, you can use that knowledge to plan and customize the configuration settings for your scorecard KPIs.

About the scenario for this walkthrough

This walkthrough shows how to calculate normalized scores in a scorecard that represents the performance of a customer service organization.

The objective KPI is Customer Service; the scorecard uses two individual KPIs.

• Resolved customer service requests.

• Service requests resolved in 15 minutes or less.

The following table shows the elements that might be in a scorecard for such a scenario. It includes normalized scores for individual KPIs, and shows a summary score for the objective.

 KPI Actual Target Normalized score Objective: Customer Service -- -- 97.72% » Resolved customer service requests 1759 1500 95.4% » Service requests resolved in 15 minutes or less 1328 1000 100%

## Step 2 - Calculate the raw scores

The raw score is a simple percentage of the Actual value compared to the Target value:

Raw score = 100* (Actual/Target)

The following table shows the raw scores:

 KPI Actual Target Raw score Resolved customer service request 1759 1500 117.3% Service requests resolved in 15 minutes or less 1328 1000 132.8%

## Step 3 Locate the threshold boundaries and bands that contain the raw scores

Scorecards represent performance by using a variety of graphical indicators to represent the calculated value of each KPI score. To select the indicator that represents a particular score, Monitoring Server must determine which band contains that score.

A band is a range of values between two endpoints called Thresholds. When a KPI score falls between two threshold values, Monitoring Server can identify the associated band.

This scenario has three bands:

• Band 1     This band represents poor performance. KPI scores in this band range from Worst (0.0%) to Threshold 1 (50%). In PerformancePoint Dashboard Designer, the Worst value is always Threshold 0.

• Band 2     This band represents moderate performance. KPI scores in this band have not reached target values. The scores in this band range from Threshold 1 (50%) to Threshold 2 (100%).

• Band 3      This band represents excellent performance. KPI scores in this band equal or exceed the target value. The scores in this band range from Threshold 2 (100%) to Best (120%).

Note: To set values for thresholds, including values for Best and Worst, configure a scoring pattern when you Set indicator thresholds for KPI target values.

The following table shows the band location for each KPI raw score in this scenario:

 KPI Raw Score Band location Resolved customer service requests 117.3 3Threshold 2 (100%) to Best (120%) Service requests resolved in 15 minutes or less 132.8 3Threshold 2 (100%) to Best (120%)

## Step 4 - Convert the threshold values

To convert the band thresholds to a common scale, Monitoring Server calculates a proportional adjustment for each threshold that maps the original values to a new scale.

For each KPI, Monitoring Server calculates a threshold factor for conversion. This factor is the length of the band that contains the KPI raw scores, multiplied times the number of bands in the scenario. The scorecard in this example has three bands. Therefore, the formula that converts the band thresholds is

Threshold factor = (Upper threshold – lower threshold)* 3

In this example, both KPI scores fall in the band between Threshold 2 (100%) and Best (120%). Therefore, both KPIs require the same threshold adjustment factor, 0.6.

## Step 5 - Convert individual KPI scores

To convert individual KPI scores to the common scale, first calculate how far the raw score is from its lower threshold. Then, divide that length by the associated threshold factor. Finally, convert the result from a percent value to a decimal value.

The following formula shows the calculation that converts raw KPI scores to the new scale.

Converted score = .01 * (Raw score – lower threshold)/(threshold factor)

The following table shows the converted KPI scores for the KPIs in this example.

 KPI Converted score (decimal ) Resolved customer service requests 0.2883 Service requests resolved in 15 minutes or less 0.5467

## Step 6 - Adjust the location

The next step determines how much to adjust the converted score so that it is in the correct position relative to the lowest possible value (the Worst value). To determine the amount of adjustment that is required, first determine how many bands separate the raw score from the original Worst value. Then, divide that value by the total number of bands.

Since this example scorecard has three bands, the formula that calculates the band adjustment amount is

Band adjustment amount = (Number of bands from Worst)/3

The band adjustment for the KPIs in the scorecard example is 0.667.

## Step 7 - Calculate the normalized scores

The final step calculates the normalized scores. The normalized score is the sum of the converted raw score and the band adjustment amount, divided by the number of bands. There are three bands.

The following formula shows the calculation for this scorecard.

Normalized score = Converted score + band adjustment amount

The following table shows the normalized scores for the KPIs in the scorecard example. Notice that when the normalized score exceeds 100%, Monitoring Server uses a value of 100% in calculating the objective score.

 KPI Converted score Band adj Normalized score Objective: Customer Service --- --- 97.72%. » Resolved customer service requests 0.2883 0.667 95.4% » Service requests resolved in 15 minutes or less 0.5467 0.667 121.3%

## Where to go for more information

The information in this topic helps develop an understanding of how scorecards work. For more background information about scorecards, see Scorecard overview.

To learn how to use this information in a scorecard, See all steps that are required to build a scorecard.

Share