Metrics is used to measure performance with other teammates? What were your experience?
Posted by badboyzpwns@reddit | ExperiencedDevs | View on Reddit | 25 comments
My company started to track metrics a while back and now its brought up on my 1:1 where m being compared to the rest of the team (like how fast I close tickets, how many tickets closed in x period). What were your experiences with these? It feels like stack ranking implicitly?
dnult@reddit
At one time our boss tracked ticket closure counts. The most junior engineer in the team was closing about 5x the number of our most senior engineers because they were working on mostly support requests. The seniors on the other hand were taking days/weeks to close requests that required design, development, and testing. Thankfully the boss recognized what was happening and dropped it from performance reviews.
circalight@reddit
If you're trying to turn your dev org into Survivor with one dev left, this is a great idea.
ninetofivedev@reddit
This feels like a social contagion that has recently become viral.
For years, leadership was all about delicately tracking performance by pretending to be objective but beneath the surface, we all knew it was moreso based on the perception of productivity than actually measuring various metrics and holding engineers accountable to those metrics. And that is actually better. Metrics can and will be gamed and there isn’t really any hard science that says “if someone provides more outputs, they’ll provide more value” which is typically the type of metrics we collect.
That all changed somewhat recently and I don’t know where it originated from. I said this in another post, my team built a platform that basically monitors everything everyone one of our engineers does and we can use that to build what is essentially a summary of what any engineer “accomplished” and how they accomplished it over any timeframe.
Like down to the “then is when they were active. This is how much ai they used, this is the number of PRs they submitted, this is the number of LoC they changed, this is the jiras they created.
And then we boil down all those events and basically you can build anything on top of it. Down to the binary “was this engineer productive today” and the derivatives from those metrics.
It was fun to build. But I think it won’t be used for good and I think the people using it don’t understand how flawed it is.
ButWhatIfPotato@reddit
"I need a source of useless and highly skewable numbers pulled straight out of my ass to use as an excuse for denying promotions" is my experience with this.
codescapes@reddit
Your colleagues are formally now your opponents and your usefulness is measured in manipulable metrics instead of actual delivery. I don't think I need to expand on why that's damaging for collaboration and genuine productivity / value.
More importantly though, once this starts it can brainfuck you so badly it stays with you the rest of your career. By which I mean you start processing all team interactions through the lens of stack rankings, promo slates, end of year grades etc. This sort of thing can utterly ruin a company, everything is downstream of incentives.
This is how you get people doing bizarre shit like closing incomplete tickets, putting it in blocked and blaming another team or messing about with point metrics etc.
GlobalCurry@reddit
Metrics are always bad for building teams, people just game them. I feel like companies forgot the lessons learned in the 80s and 90s and 00s.
WiseHalmon@reddit
What other metrics
badboyzpwns@reddit (OP)
How fast PRs gets merged, story points for velocity, tickets closed and comments :l
yxhuvud@reddit
Time to do super broken up tickets with inflated story points, is what I see.
belkh@reddit
you mentioned 1:1s, is this being brought up by your manager? are they so far removed from work they do not see how these stats can differ wildly?
these metrics could be useful as supplementary context, but you can't use them to judge people, because they're easy to game.
ratttertintattertins@reddit
As a manager, I look at them but don’t normally discuss them with my reports because of:
https://en.wikipedia.org/wiki/Goodhart%27s_law
This is basic stuff and management that don’t know about that ^ are usually idiots.
That said, it gets interesting when you have a report who is both qualativley slow, you’re getting negative feedback about them and they’re also showing up badly in the metrics. It is tempting to mention it to them, simply because you know that if you don’t make them aware of it, they’re going to end up attracting more and more notice and will end up getting fired eventually.
lawrencek1992@reddit
Like if you estimate poorly, take forever to respond, don’t ever meet your estimates, and if I consider both complexity and volume of work, you produce less, metrics will show it. And bringing up the obvious gap is a kind way of giving someone a chance to turn it around.
Also I have hated doing people management as a team lead on previous teams. I can lead technically and handle project management for a team, but managing humans is so much more draining.
originalchronoguy@reddit
I would not call it hunches but I know who performs and who doesn't.
You can just go based on what they demo bi-weekly. You can see the depth of their work. And how they estimate and how well they keep to those estimates. You can look at the complexity of their tasks and how they don't make excuses.
I know the guys who spent three weeks on something I can literally do in 45 minutes, The same ones who always push back on trivial ambiguity. Those who make excuses for blockers.
My leadership is the same. If we really had to, We can look at git commits, jira velocity but we don't. Those things are only needed if you are going to fire them.
My problem isn't with those people. It is with management who let those shenanigans transpire too long as it brings down morale of teams.
lawrencek1992@reddit
Yeah this is how I feel. I know they rank us, but I think it’s more nuanced than just code-based metrics (and those are easy to see for the whole department).
Some people make their whole team more effective. Some people just get their own stuff done and do it well. And some people drag down the team to a lesser or greater extent. Metrics will likely “prove” it, but you can just tell by being on a team with someone for a bit where they fall on that spectrum.
badboyzpwns@reddit (OP)
>We can look at git commits, jira velocity but we don't. Those things are only needed if you are going to fire them.
originalchronoguy@reddit
Never need to get to that point. It is usually obvious -- missed deadlines. Unresponsive to team mates. Just something HR would suggest to make it easier to protect from a HR lawsuit.
lawrencek1992@reddit
Generally using any one metric as a way to measure engineers isn’t effective, also it’s easy to game any one metric/Goodhart’s law. If your manager is focusing on a slew of metrics, and you are on the low side on all of them, then it sounds like a productivity issue on your part.
Honestly a lottttt of companies stack rank. As long as they aren’t publicizing where you are in the rank, I don’t think it’s a big deal. Generally if you are in the top 50% of engineers it shouldn’t be an issue. You probably have an idea of where you stand in the ranking even if it’s not publicized.
nasanu@reddit
My company uses this as one metric. Do more tickets? Well that is obvious proof you simply take the easy tickets. That junior who is taking 5 weeks on their ticket? Damn they are smashing it. Obviously doing something amazing since it's taking so long...
originalchronoguy@reddit
It is way different where I am at. It is the few juniors that take 5 story points, do it in 2 weeks on time as promised while some of the 10-15 YOE guys would take 8 weeks. The 10-15 YOE would build some widget that changes layout from desktop to mobile. The junior is building the entire REST backend and the entire user state management and session handling. But the 10-15 YOE got to make sure that one button goes on the next row when you scale down to mobile.
Dead serious. How those guys have jobs is beyond me.
badboyzpwns@reddit (OP)
Hahaha I love that!
BusEquivalent9605@reddit
basically, metrics don’t work because as soon as people know the metric (other than just build a good product that makes money and is stable and easy to expand/adapt) they start adjusting their behavior, putting effort towards meeting the metric (instead of building a good product that makes money and is stable and easy to expand/adapt).
this does not stop upper management from wanting metrics that they can skim instead of actually getting to know who is doing valuable work on the team
xlb250@reddit
Every company tracks your metrics. At least your manager is being transparent with you.
Murky_Citron_1799@reddit
Game the fuck out of the metrics. Make them have to absurdly compare you. For example, if they care about how many commits you have, make a script that can generate a commit per line of code that you write. So when it's time to compare metrics, yours is 1000x the other guy. And entertain the conversation about it. Hopefully they stop talking about stupid metrics then.
obelix_dogmatix@reddit
That’s one way to pit teammates against one another
DingBat99999@reddit
Bad, bad, bad idea.
At least you've learned what kind of place you work at.