When a color-coded Excel table worked as a progress indicator

A proxy metric field story

I learned about the term "proxy metric" in a training about Objectives & Key Results (OKR) from Felipe Castro. They can serve whenever measuring "the real thing" takes too much time or effort to be practical. For this to work, there needs to be a close relationship between both.

Until today, my favorite example of a proxy metric is the story of Van Halen and the brown M&Ms. (I learned about it in the same training.)

David Lee Roth tells the story behind the "no brown M&Ms" legend

For my taste, I do not read enough about successful uses of proxy metrics. So why not do this myself?

Fair warning ahead: It seems to me that proxy metrics suffer the same fate as the lifehacks mentioned by Cory Doctorow:

"One trait all those lifehacks shared: everyone who created a little hack was faintly embarrassed by it, and assumed that others who learned about their tricks would find them trivial or foolish."

In other words, they sound ridiculous when you think about them. How could this ever work?

This story happened in a complicated environment where producing the right output was enough. Nevertheless, creating this output turned out to be challenging enough. A set of data flows had to be built. However, this had never been done within that company using that technology on a target system that was yet to be built. In other words, there was a lot of risk and little transparency. Moreover, there was some preparation work to be done before any functionality could be shown - basically "providing the infrastructure".

After a couple of months, the sponsors became impatient. How could they get an idea whether they would get what they needed within the right amount of time - or even anything useful?

It was obvious that shorter updates would at least create trust. If there were any problems, shorter reporting intervals would at least surface them as early as possible.

But how do you adequately express the whole situation every day in all its nuances in a way so that the updates can still be consumed? In short, you don't. Forget about the details.

The manager working with me came up with a great idea that we worked out as a team: List all the interfaces in one table as rows. List the raw steps or phases for producing them in columns, splitting those in such a way that you can assume they will be doable within one day by one person. Count all the known work as 100% and assign percentages to every phase of every interface. Agree with the team on a set percentage of progress that you need every day to make it on time. Start working and measuring.

We used an Excel table with a color coding that a colleague prepared. Every interface started as red and slowly turned to yellow and finally green based on the percentage of steps done. I made sure to include all the updates and monitor the progress myself. At the end of the day, the manager would ask how far we got. This quickly turned into a routine. We got the whole initiative back into calm waters.

Why did it work? First, we agreed on this as a team. It was not a top-down decision ordered by someone far away from the team. Second, I immediately understood that "having the first phase of X interfaces done" was less valuable than "moving X interfaces in different stages of completion one step closer to done, focusing on those close to done first". I intentionally tried to get the first interface to done as soon as possible to learn about the potential problems as soon as possible. If any surprises were waiting for us down the line, this would not be nice news to share, but at least as early as possible. I assumed that starting with the easiest interfaces would get us some early success while still discovering general problems. At the same time, I monitored that we would not slow down but use the knowledge gained to become better in our way of working.

In other words, I actively worked against gaming the metrics. Plus, the table helped inform our decision-making on what to work on next, and everyone could get the bigger picture at any point in time.

In the end, we made it. There were more factors involved in the success, especially the trust between the people on the team involved, which might have been the condition to use such a proxy metric in the first place. The Excel file did the trick: At the end of every day, we knew where we were, and we did not have to rely on gut feeling.

Van Halen: Jump