Good KPIs

One of the challenges any organization, including drug safety has, is designing good Key Performance Indicators known as KPI.

Any organization must challenge the way things are done (and for what results). Designing good KPIs

1. Objectives: what are we trying to achieve?

• May be more than one indicator for each objective
• Each objective will have strategies on how to achieve them

2. Indicators: what are you going to measure?

• used to assess the present state of progress and to suggest an appropriate course of action.

3. Measures: how are you going to measure it?

• can be qualitative or quantitative data related to inputs, processes or outputs.

4. Targets: what is the desired result?

• can be minimal targets, stretch targets or a combination.

5. Results: what have you actually achieved?

This creates a good starting point but it does not provide the separation between various KPIs.

What you need then to do is to categorize the KPIs in various groups:

Organizing KPIs

As a recommended practice, I found useful to get this segmentation in place:

1. “Approach” KPIs

• How do we have/need change things?

2. “Deployment” KPIs

• How do we have/need implement things?

3. “Results” KPIs

• What we have/need obtained?

4. “Improvement” KPIs

• What we have/need to improve?

Scaling KPIs

Once everything is in place, I would use on a 1 to 5 Likert-type scale for both the KPI and the actual target.

A Likert-type scale consists of a series of declarative statements. The subject is asked to indicate whether he agrees or disagrees with each statement. Commonly, five options are provided: "strongly agree," "agree," "undecided," "disagree," and "strongly disagree." Other Likert-type scales include four or six steps rather than five, excluding the undecided position.

Scales developed by the Likert method will ordinarily include from six to thirty declarative statements. Some of these statements will be worded in a positive manner and other will be worded in a negative manner. For example, two statements which might be used in a scale to measure attitude toward capital punishment would be, "Capital punishment is nothing but legalized murder," and "Capital punishment gives the murderer just what he deserves." A person who is in favor of capital punishment would be expected to disagree with the first statement but agree with the second statement. Of course, the person opposed to capital punishment would be expected to give opposite responses. The individual responses "strongly agree" through "strongly disagree" are assigned numbers, usually 1-5. In this manner the responses to the various items are quantified and may be summed across statements to give a total score for the individual on the scale. It is necessary, of course, that the assigned numbers are consistent with the meaning of the response. For example, the first statement above could be scored 1-5 and the second statement scored 5-1. In this way a person with a strongly favorable attitude toward capital punishment would receive a score of 10 for these two items while a person strongly opposed to capital punishment would receive a score of 2.

Attitude scores produced by Likert-type scales have been found by Likert and others to yield data which may be analyzed by "normal curve" statistics. Because this type of data is desired by researchers in most cases, the Likert-type scale provides a very useful and relatively uncomplicated method of obtaining data on people's attitudes for persuasion research.