Non-working KPIs in IT: what they can cause and how to choose the right ones. Algorithm for creating KPIs

Non-working KPIs in IT: what they can cause and how to choose the right ones. Algorithm for creating KPIs

The second article about non-working KPI. We will look at how to create KPI’s that actually improve performance.

Algorithm for creating KPIs

As I already mentioned, KPIs should be aimed at improving one or more characteristics of the project, team, or process. I propose the following algorithm for creating the correct metrics:

  1. Following the strategic goal. The strategy is to increase customer satisfaction and loyalty to the existing IT product. Its quality is one of the key success factors. Therefore, despite the addition of new functionality, it is necessary to increase the coverage with unit-tests
  2. Definition of the current and target values. Example: the current coverage level of code by unit-tests — 30%, whereas the target, recognized industry standard — 80%
  3. Obtaining metric data. Metric values should be calculated as easily as possible and preferably automatically. In the example with unit-tests, this should be done with SonarQube, correctly configured and included in the general CI/CD-process of the project;
  4. Determination of the KPI weight. For a coherent process, an employee or team need to work on several goals at once. For example, the product quality, the time between the generation of a business idea and its implementation, the introduction of innovative ideas, etc.
  5. Frequency of updates. It is reasonable to revise targets and the weight of KPIs in the overall set at regular intervals. Depending on the business situation, it makes sense to update these indicators once a month or a quarter, not more often. Employees should be notified about such changes in advance, not ex post. Moreover, this information should not be hidden
  6. Life time. If the strategic goal of the company has been achieved or the KPI has ceased to bring the expected benefit, or maybe it simply has been chosen incorrectly, this metric should be abandoned and another one should be introduced; the one, which will show the business situation with a plan for the future.

Non-working KPIs in IT what they can cause and how to choose the right ones. Algorithm for creating KPIs.jpg

Examples of the right KPIs for the IT industry

If we are talking about applications, platforms and solutions that allow a company or clients to make money, then one of the most important metrics will be Uptime – the uptime of an information system or application.

It is measured as a percentage for a quarter or for a year. Depending on the reliability class of the application, it may have these targets:
  • 99.9% — for a year (365 days) the application in total did not work a little less than 9 hours
  • 99.5% — just under two days
  • 99.0% — less than four days

To achieve such KPIs, the manager responsible for the platform will have to work very hard together with their team and other teams.

There are a few more high-level KPIs:
  • Reduction of financial costs (per year) for support of the application
  • The benefits of moving to a new platform or technology

This will require calculation and comparison of implementation costs versus maintaining the existing solution, including support costs.

KPIs also include more technical or IT process metrics.

Next, I will give an example of several indicators, the implementation of which will help to increase the efficiency of the team and add employees’ responsibility for the success of the entire project. As a result, it will make the process more transparent and predictable.

Let’s assume that all IT specialists are smart and ambitious employees. Consequently, the metrics that help the company and employees should also be smart.

Examples of such KPIs for a software developer:

"Premium" KPIs "Penalty" KPIs

The number of completed tasks (new or fixed ones), measured in Story Points (not in hours)

Number of repeatedly (!) found defects multiplied by the complexity of the defect

Number of fixed errors multiplied by the complexity of the defect


Now I will give examples of KPIs for a QA automation engineer:

"Premium" KPIs "Penalty" KPIs

Number of automated tests created multiplied by the test complexity

 Number of defects detected after the product release multiplied by the complexity of the defect

Number of tests running multiplied by the test complexity


And these are just two roles of the many within the IT team.

You can apply an additional unified metric if several teams are involved in the project: customer/client/end user’s feedback. It will increase the overall interest in the result.

If the next release has been successful, users are satisfied with the new functionality, and the application speed has increased or the interaction has become more friendly, a multiplying factor is applied. But if, after the production, users send new serious defects, the application is unstable, obviously, the whole team did not work well.

So, what are the conclusions from all of the above?

First, without metrics, you can start, but it is difficult to survive and develop. This is ineffective and very risky.

Second, the right metrics bring benefits and improvements, while the wrong ones bring harm, rejection, and losses.

Third, KPIs help to solve the issue of transition from the present to the future with less strength, nerves, and finances, but with greater benefit. At the same time, they record in history every step taken.

Fourth, nothing is permanent. You should not cling to once accepted KPIs. They need to be updated regularly, to achieve permanent benefits.

Interested in upgrading your skills? Check out our trainings.

Mikhail Gedzberg
Project Management Consultant
Still have questions?
Connect with us