Permalink
Fetching contributors…
Cannot retrieve contributors at this time
129 lines (80 sloc) 4.92 KB

Title

Introducing Metrics in InnerSource

Patlet

Involve all stakeholders in designing and interpreting metrics to measure the current status in terms of health and performance of the InnerSource initiative.

Context

An organization is planning to apply or this is in the early stages of applying InnerSource. This would like to measure the current status in terms of 'health' and performance of the initiative, and if the expected outcomes such as an increase in the level of cross-divisional and cross-location collaboration are actually taking place. This pattern applies to early stages of the initiative or are small in their scope, but they may be mature in their initial process and steps.

Problem

This organization may already have qualitative feedback from the involved teams, but desire more objective information focused on development activities.

The organization does not really know where to start measuring things or what are the key parameters to measure.

Changes in the top level initiatives may affect the InnerSource program as they rely in the good will of some executive from the organization.

You may have a problem justifying the InnerSource effort when there is a change in business priorities or business leadership. Then you need something concrete to justify the program. A future problem you're guarding against.

If there's a change in the C-level, metrics might be helpful to convince them that InnerSource is useful.

Forces

People do not like to be tracked or measured.

There is not a proper monitoring infrastructure for the software development process and this is hard to build or to get funding for this.

There is not a culture of software development metrics.

Metrics are usually misunderstood if people have not received any training on those.

Organizations collect vanity of any other type of metrics that do not track business objectives' sucess or failure over time.

Metrics tend to become goals, will subsequently be gamed and thus meaningless.

Merging existing team metrics if another team has a way to measure how teams are doing; these new measurements could potentially conflict.

Some countries in some organizations may face extra complexity when introducing metrics as they may not allow to track individuals.

Tools and how they use them. There might be a learning curve in the discussion about metrics. And perhaps the tools do not support the metrics we're looking for.

Solution

Bring developers, middle managers and C-level to have a discussion about metrics. And consider other roles out of the usual development process such as Human Resources, legal departments, product management, and others.

Let developers and middle managers know that these metrics or KPIs are not focused on tracking their personal performance, but to compare if the initiative is currently working as expected.

Consider third party and neutral player to produce such metrics.

Have specific training on the topic of metrics and good practices to use them. An example is to have a methodology to follow metrics such as the Goal-Question-Metric approach or the Objetives-KeyResults one. On the other hand, try to update the metrics used to the short-term and medium-term goals.

Bring specific discussions on the metrics to be used avoiding per developer granularity or at least anonymizing that info.

Produce a characterization of metrics as this might be helpful for others to understand and follow.

  • Nearly always InnerSource is not a goal in-and-of itself but a proposal of how to improve some larger problem that the company is having. One class of metrics is around that larger goal (e.g. quality, interrupt-driven work, duplicated code, etc.) and how it is going for the company.
  • Another class of metrics is around how much InnerSourcing is happening. A raw definition of InnerSource based on code submission to a repository not owned by the submitter's team could be measurable in a few ways.
  • A third class of metrics are KPIs that we believe will improve the raw amount of InnerSourcing happening (e.g. mean-time-to-review, automated test run on PR, etc.).

Note: any proposed metric are just examples and not the ones that should be using. Depending on the selection of business goals to track, those will match with specific set of metrics.

Resulting Context

The organization builds a monitoring infrastructure where the agreed metrics will be the ones to be used.

Those metrics will help C-level to understand the current situation of the project and this will help to compare the status before applying inner source and after applying the inner source initiative.

Those metrics will help middle management to understand how the initiative evolves and help them with their daily work.

Those metrics will help developers to understand how the initiative evolves and help them with their daily work.

Known Instances

Authors

  • Daniel Izquierdo
  • Tim Yao
  • Clint
  • Russ Rutledge
  • Tom

Acknowledgement

  • Georg
  • Bob

State

Early Idea