Hacker News Reader

Comment by darkerside | Hacker News Reader

Ever hear someone complain about big company bureaucracy? The bigger an organization gets, the more impossible it gets to manage it and keep people aligned towards effective and compounding goals. OKRs promise to help improve that impossible process. I'm not surprised at all that people are interested.

View on HN

> OKRs promise to help improve that impossible process.

You mean like Google devs who kill projects faster than they roll them out

A lot of people are mistaking my explanation for why people are interested in OKRs for an endorsement of OKRs. I think having goals is good, understanding how they type into the broader company mission is good, but being able to adapt is also good. OKRs:goal-setting::Scrum:agile

Kinda disagree. All the times I have seen OKRs implemented has been as part of a "transition to agile" with several other measures, and often what happens is they get slapped on top or replace other measures, without doing anything real for fixing company bureaucracy. You need much more fundamental changes to fix that, and that would involve pissing off a lot of people with a lot of power.

OKRs can in theory make sense but they create a new kind of tension. Because they compete with the BAU work, the niggly things that come up mid quarter that you need to get done. Individuals need to resolve the dilemma, do what has to be done, or do what makes me look good.

I think a new thing is needed at least for smaller teams. Something like adaptive goals. Roadmap a year, vaguely and plan the next 6 weeks. Measure stuff where it makes sense but not everything needs a measure (or things that don’t might be 0 or 1). Plan based on velocity that takes into account that you wont have planned everything.

Measures are useful but also bullshit. There is no correlation between the measure and business success without intelligence. Even revenue is not a measure of success (I could sell half price bitcoins and have a lot of revenue!)

The existence of bad measures doesn't preclude the possibility of good ones, even though they won't be perfect. BAU activities, e.g. devops, can be measured in useful ways[0] as well.

[0] https://dora.dev

Oh yes. But intelligence needs to be applied. OKRs turn measures into something like a sport, like soccer where those measure become the goal (beat the KR and get a bonus, or even just pressure on the KR and nothing else).

> But intelligence needs to be applied

I think this is an almost universal caveat.

> OKRs turn measures into something like a sport

No, I don't think that's true. People might do that, but they also do that to things that aren't OKRs. It's great to critique OKRs, but not by comparing them to a hypothetical perfect world.

BAU work should be covered in resource planning, not OKRs. The amount of time available to work on KRs should never be 100% -- it should be in the range of 60-70% to account for overhead, trainings, vacations, and unexpected issues that must be addressed.

Reply on news.ycombinator.com