An approach for identifying metrics and response measures so that we can connect data with business goals. The Goal-Question-Metric approach (GQM) was introduced by Victory Basili, Gianluigi Caldiera, and H. Dieter Rombach in The Goal Question Metric (GQM) Approach [BCR94]. The goal of GQM is to identify measures we can use to determine whether a goal has is satisfied.

There are three parts to the GQM approach. The goal defines conceptual requirement that must be met. Goals can describe quality attribute scenarios, general software quality, business goals, or other topics. Questions illustrate the means by which we can characterize one or more goals. Metrics defines the measures needed to answer one or more questions.

Benefits

• Emphasizes using stakeholder goals as the basis for measures

• Shows clear lineage from data to stakeholder goals by way of the questions that must be answered to decide whether a goal is satisfied

• Flexible approach that can help teams think about metrics in a variety of situations

Activity Timing

15--90 minutes

Participants

This activity can be done solo or in a small group of 2--5 people. Any mix of stakeholders will work.

Preparation and Materials

• Whiteboard or flipchart papers, markers
• Goals to be explored can be optionally identified before the workshop

Steps

1. Write a goal at the far left of the whiteboard.

2. Prompt participants to provide questions. What questions would you need to answer to know if we’ve met this goal? Write each question to the right of the goal. Draw lines from the goal to the questions to create a tree.

3. Explore each question to identify metrics needed to answer the question. Write each metric to the right of the questions. Draw lines from each question to the metrics required to answer it.

4. Repeat the exercise for any related goals that might reasonably use the same questions or metrics. The end result should be a tree that connects metrics to questions and questions to goals.

5. Identify data required to compute each metric. Write the data needs to the right of the metrics. Draw lines from each metric to the data needed to compute the metric.

6. For each piece of data identified, determine where you can get the data. Write down each data source for each bit of data. Describe the cost of gathering data from each of the data sources.

7. The last phase of the workshop is to prioritize data and metrics. Clearly identify must have metrics. Look for data sources that can provide data needed to compute multiple metrics or metrics that can answer multiple questions.

8. Record the results of the workshop by taking pictures and writing down the discussed goals, questions, metrics, data, and data sources.

Guidelines and Hints

• Be sure to have plenty of space for drawing the GQM tree.

• Look for opportunities for reuse. Metrics can be used to answer more than one question. Likewise, the same data might help compute multiple metrics.

• Data sources and data are likely to influence the architecture, but also look for opportunities to gather data outside of the architecture.

• Record results in a table or spreadsheet to validate with stakeholders later. The identified metrics are revisited throughout the life of the software system.

Example

In this example, the goal is written on the far left, questions are captured in the middle column, and metrics are written in the column on the far right. Data used to compute the metrics was omitted from this particular GQM sketch.