Software Metrics: Classification and Guidelines

“Perhaps the worst software technology of all time was the use of physical lines of code [for metrics]. Continued use of this approach, in the author's opinion, should be considered professional malpractice”. 

— Capers Jones

 

What are Software Metrics?

 

Software Metrics are the measure of the characteristics of the software products and software development processes. These features are quantifiable. While measuring the software products the metrics look for-

 

1. Size

2. Design Features

3. Complexity

4. Performance

5. Quality 

 

By using the software metrics we can improve the overall performance and quality of the software development. Current quality is determined and improvements are done to make it more efficient.

 

All the phases of the software development can be targeted using metrics. These metrics are identical to the management functions therefore, you can combine both and streamline the projects.

 

Also Read | Marketing Metrics

 

Advantages of Software Metrics

 

1. Multiple softwares can be compared on the basis of their design methodology.

 

2. The features of programming languages can be analyzed and compared through software metrics.

 

3. Using the metrics we can design the specifications of the software.

 

4. Whether all requirements and specifications are compiled or not can be verified using the software metrics system.

 

5. The overall code complexity can be determined through software metrics.

 

6. Allocating the resources for code testing can be done using software metrics.

 

7. Different phases of the development of software life cycle can be tracked using software metrics.

 

8. Comparing maintenance costs, development costs and trade-offs can be done through appropriate software metrics.

 

9. Decisions of dividing complicated modules can be easily made through software metrics.

 

 

Disadvantages of Software Metrics

 

1. One cannot use software metrics in all cases. It becomes really complex and expensive at times.

 

2. Sometimes the source data is not reliable. Validating the historical and empirical sources is not easy. If the source is unreliable then the verification is not justified.

 

3. One can use software metrics to quantify the features of software products and services but the performance of technical staff cannot be measured using software metrics.

 

4. There is no standard derivative or criteria for software metrics. They are customized and developed based on the criteria and needs of the products.

 

 

Classification of Software Metrics

 

According to Galileo, measuring something which is easily measurable is simple but trying to measure the unmeasurable requires effort. In business, metrics help us measure a lot of variables that could not be quantifiable earlier.

 

Software Metrics will give us the information that we need after analyzing multiple data sources. Choosing the reliable sources for data is the duty of the management and extracting, analyzing and projecting information is what machines will do for us.

 

On this basis Software Metrics can be divided into five types which are:

 

1. Product Metrics

 

Product Metrics measure the characteristics of the software product. There are two main features that are focused on- First one is the size and the complexity of the product and the Second one is the reliability and quality of the software.

 

2. Process Metrics

 

Process Metrics on the other hand measure the features of the software development processes. There can be multiple things to focus on like how efficiently faults or errors are detected.

 

Apart from fault detection it also focuses on the  techniques, methods, tools and reliability of the overall software process.

 

3. Internal Metrics

 

All the properties which are important to the software developer are measured using the Internal Metrics. LOC i.e Line of Control is an example of Internal Metrics.

 

4. External Metrics

 

All the properties which are important to the user are measured using the External Metrics. 

 

5. Project Metrics

 

This metric system is used by the project managers to track the progress of the project. Past project references are used to generate data. Time, Cost, Labor etc. are some of the key measurement areas.

 

Also Read | Project Managing Softwares

 

Top 5 Software Metrics Examples

 

There are multiple examples of Software Metrics, but the given 5 are the ones that are directly related to the user.

 

1. Customer Satisfaction

The most used Software metric for customer satisfaction is Net Promoter Score. In this a score range is pre-decided i.e from -100 to +100. Getting -100 means that no customers will refer you further as they are not satisfied.

 

+100 on the other hand means the customers are extremely happy. One of the examples of Net Promoter Score metric is SurveyMonkey. It offers free services as well. Let us take an example to understand NPS better.

 

Suppose you got 100 responses in your survey and 15 of them ranged from 0-6, 25 were in the range 7-8 and 60 in 9-10. The first 15 are called detractors, next 25 are passive and the renaming are promoters.

 

NPS is calculated by subtracting the detractors from the promoters. Passives are ignored in most cases. In this case NPS is 60-15=45.

 

2. Team Velocity

 

Team velocity is about the value that team is offering to the users. With team velocity we measure the story points, number as well as the size of the product characteristics etc. All this is calculated for a particular period of time.

 

Let us take an example. Suppose in three sprints the team has completed 15, 20 and 13 points. From this we can calculate the average velocity which is 16. This average will be considered as a standard in the upcoming iterations.

 

3. Release Burndown

 

Release Burndown is a chart which helps us to understand the progress after the software release. Basically, it indicates the overall functionality and answers what remains and what can be done.

 

4. Escaped Defects

 

As the name suggests, this metrics checks the bugs and errors that were identified after the software was in production. If we can keep a check on the quality then we can maintain the overall quality of the system and offer the best to the users as well.

 

5. True Test Coverage

 

It is a True test coverage metric not a regular one as it checks how well your codebase is covered by all types of tests. Tests like Units, Integration Testing, manual tests, UI automation, End-to-end encryption tests etc are some examples.

 

Also Read | Customer Lifetime Value (CLV)

 

Guidelines for Appropriate Use of Software Metrics

 

The whole software development team should collectively design the software metrics for better and efficient work. An ideal software metric should posses the following features:

 

• Simple and Reliable

 

• Consistent 

 

• Unambiguous

 

• Uniform units for measurement

 

• Easy calibration

 

• Cost Effective

 

• Accurate

 

But in order to obtain the best results, we must understand how to use these software metrics. Given below are the guidelines for appropriate use of Software Metrics.

---

Guidelines for appropriate use of Software Metrics

---

 

1. Link Software Metrics to Goals

The software metrics should be linked to the goals of the software development team to unite the purpose. If the focus becomes common then work will be efficient.

 

• Reduction in the line of codes.
• Reduction in the bugs reported.
• Increasing the iterations
• Speeding up tasks

 

If metrics become targets then developers will find it easier to improve the overall use of software and user experience.

 

While the overall goals of the enterprise are being discussed by the management, it is necessary to involve the software development team to choose the relevant software metrics. These metrics will later on derive the overall performance of the enterprise.

 

2. Track Numbers not Trends

 

Software Metrics are useful for management as they provide numeric data in place of complex data. These simple numbers are easier to compare with other standards. Whenever the metric target is met, developers know it is a success. 

 

On the other hand when it is not met, it means somewhere something is missing and better work must be done for the next turn. When we use metrics we measure trends. Single data set is of not much relevance.

 

Trends show how much we have progressed since before. If the target is not achieved it is a failure or vice-versa. But if we measure trends then we can understand what to do to make it a success. We can understand the reasons for failures. 

 

3. Set a Shorter Measurement Period

Software team always wants to get work done instead of waiting and setting targets or measuring them. Doing is more important. Therefore, using an approach in which repeated target setting is not required is better and less worrisome for the team.

 

Only when the final outcome is ready, the team should check if it is a success or a failure. If the whole measurement period is broken into smaller segments then the team can monitor the metrics better.

 

They can analyze the trends and keep a check on the progress. The broken segments will offer better data points to analyze the outcome and it will be better for achieving the goals as well.

 

4. Stop using Metrics that do not lead to Change

 

It is a known fact that doing something repeatedly when the end result will be a failure is a waste of time. In fact it is called Insanity. But making relevant adjustments that can change the outcome is what we need.

 

But when we talk about software metrics, doing the exact same thing repeatedly is called managing metrics. Why do developers do it? Why do they keep repeating the same thing again and again?

 

The answer is that developers are focussed on those software metrics that do not measure the goals. There are certain metrics which are irrelevant for measuring the goals or objectives or workflow.

 

Management and the developers must focus on those software metrics that will help them achieve the goals and bring them closer towards success.

 

Also Read | Performance Testing

 

We have no fixed standard or definition of metrics. Different software development teams have their own different standards and definitions. It all depends on the goals or requirements.