My Team Built a Developer Productivity Platform for our Executive Team - It’s Awful
Inspired by ninetofivedev
The Original Argument
No tokens generated during composition of this post
First off, I don’t mean it’s awful from a design perspective. I’m biased, but I think the team did a good job and it was actually a pretty fun project to work on.
No, I mean it’s awful because of how it’s going to be used.
The easiest way to describe what this platform does is that any system you can possibly imagine that you might interact with on a daily basis, we get event data from.
We get event data from all issues tracker (Jira, Linear, ADO). We get event data from all the VCS platforms (Github, Gitlab, ADO). We get event data from all the CI platforms (GHA, Gitlab, ADO, CCI, Jenkins). We get event data from Zoom, Slack, Teams. We get event data from Claude, Codex, Copilot. We get event data from AWS, GCP, Azure. We get event data from our SIEM, which has when your computer was active and when it was idle.
We aggregate all that data and index it based on the user.
Based on this data, I can build a summary of what any engineer did on any given day, down to the meetings you attended, the conversations you had, the PRs you created, the issues you touched. I can guess when you went and got lunch. And I can do that for any timescale.
When you look at the project itself, it was fun to build. A lot of different integrations. A lot of different constraints. We built “productivity dashboards” which basically just take a stab at “Was this engineer productive today” and then we built derivatives of metrics off of that.
It’s kind of fun.
But it’s awful. It falls for the trap that outputs correlate to impact. I tried to explain to our executive team that it’s about like looking at CPU usage and assuming 100% cpu usage means my computer is being useful.
They didn’t care. They didn’t appreciate the analogy. They love the tool. And now I dread having built it.
At the end of the day, we were all simply doing our job. And people will find out how to game the metrics, and those probably won’t be our best engineers.
I don’t know where this sudden urge to track every little thing every engineer does originated. It feels a bit like a social contagion. I feel like there used to be more common sense in this space, and suddenly that has all gone out the window and we’re back to tracking LoC.
I thought I would share this story and maybe get insight into what others are finding at their place of work around this subject.
Update: Some of you will be pleased to know that a few different teams got together and each individual submitted PRs with 10k commits as well as taking other measures to game the system. And I’m here for it.
Also some of you are acting like building a tool using company data is akin to the 1930s-1940s Germans who quite literally tortured and murdered their fellow man. For those of you that think this is even remotely comparable, kindly go outside and touch some grass.
Our Extension
[Your Riff content here — how StellarView extends this with a working build]
The Build
[Demo, repo link, or code walkthrough]