Person-centered metrics and self-organizing teams

Arseni Mourzenko
Founder and lead developer
177
articles
February 20, 2016
Tags: management 34 quality 36 productivity 36

We can­not re­li­ably mea­sure de­vel­op­ers' pro­duc­tiv­i­ty, and this is the ma­jor el­e­ments which push­es many man­agers to fo­cus on less rep­re­sen­ta­tive as­pects of de­vel­op­ers' con­tri­bu­tion to a pro­ject and a com­pa­ny.

Some com­pa­nies do a re­al­ly bad job, choos­ing per­son-cen­tered met­rics: LOCs, time phys­i­cal­ly spent at the work­place, etc. Those met­rics usu­al­ly lack any cor­re­la­tion with the suc­cess of the pro­ject and the com­pa­ny it­self, and are ei­ther ir­rel­e­vant or sim­ply harm­ful. They are good to give lousy ex­pla­na­tions to the CEO about the rea­sons the pro­ject is six months late or to fire peo­ple with­out rea­son, and in long term, lead to best peo­ple leav­ing the com­pa­ny.

Oth­er com­pa­nies do a bet­ter job by us­ing proc­cess-cen­tered met­rics which are ac­tu­al­ly rel­e­vant and of­ten in­dica­tive of both the qual­i­ty of the fi­nal prod­uct and the cul­ture of the com­pa­ny: the num­ber of bugs which hit pro­duc­tion, the time be­tween an idea and its in­te­gral im­ple­men­ta­tion, the time be­tween the mo­ment the fea­ture is im­ple­ment­ed and the mo­ment the users start ac­tu­al­ly us­ing it, the tech­ni­cal debt, etc. Those met­rics ac­tu­al­ly make it pos­si­ble to im­prove over time, achiev­ing faster re­leas­es, high­er qual­i­ty prod­ucts and, un­sur­pris­ing­ly, high­er team pro­duc­tiv­i­ty.

When pre­sent­ing those met­rics to man­agers, they are of­ten puz­zled how would those met­rics help them find­ing who in their team does a great job and should be pro­mot­ed, and who should leave the com­pa­ny or change her role. This is an in­ter­est­ing con­cern, es­pe­cial­ly giv­en that those same man­agers suck at hir­ing staff in the first place and suc­ceed at en­cour­ag­ing the best de­vel­op­ers to leave fast. What is even more in­ter­est­ing is that this con­cern makes them for­get the real is­sues which should be ad­dressed.

Man­agers in fac­to­ries do need to pro­mote and fire work­ers. This is their role for two rea­sons. Most im­por­tant­ly, no­body else could do that; if work­ers' jobs are pure­ly in­di­vid­ual, one can, in the­o­ry, work at a fac­to­ry for years with­out talk­ing to any­one, mean­ing that work­ers them­selves don't have to know or ac­tu­al­ly don't know peo­ple they work with. The fact that there is usu­al­ly com­mu­ni­ca­tion among pairs out­side the job is ir­rel­e­vant. You won't be able to de­ter­mine whether your col­league de­serves a pro­mo­tion based on a few philo­soph­i­cal dis­cus­sions with him in a bar every Fri­day evening. Also, for some types of work­ers, in­di­vid­ual pro­duc­tiv­i­ty can be mea­sured rather re­li­ably.

This doesn't work for de­vel­op­ers. In IT, it's not the in­di­vid­ual pro­duc­tiv­i­ty which is key, but the co­he­sion of the team, the com­mu­ni­ca­tion, the cre­ativ­i­ty. Not only you, as a man­ag­er, can't re­li­ably de­ter­mine pro­duc­tiv­i­ty of a de­vel­op­er, but also the team has much more to say about their mem­ber than you. They know the per­son; you don't. They work with this per­son dai­ly; you don't.

This, in turn, means that man­ag­er's pref­er­ences for a de­vel­op­er when it comes to a pro­mo­tion, or ha­tred to­wards an­oth­er one when it comes to lay-offs, is used in­stead of facts about how much a per­son is pre­cious for a com­pa­ny. This could very eas­i­ly lead to dis­crim­i­na­tion, ha­rass­ment, and oth­er things which are not specif­i­cal­ly in­dica­tive of a healthy work­place. In or­der to mit­i­gate lit­i­ga­tion, larg­er com­pa­nies some­times move those de­ci­sions to hu­man re­sources de­part­ment, which doesn't nec­es­sar­i­ly im­prove things. Hu­man re­sources still need to rely on man­ag­er's feed­back and on per­son-cen­tered met­rics which can be gamed by the man­ag­er and, as I ex­plained, are ir­rel­e­vant any­way, mak­ing their de­ci­sion all but smart.

The so­lu­tion to this prob­lem is, how­ev­er, quite straight­for­ward. Teams should be self-or­ga­nized, and de­ci­sions to join or leave a team should be­long to the con­cerned per­son and the mem­bers of the team. Valve Soft­ware does that, and so does Face­book, and it makes per­fect sense. In­stead of re­ly­ing on doubt­ful de­ci­sions from the man­age­ment or hu­man re­sources, why not let­ting peo­ple chose who they work with?

The great­est thing about this is that is solves a huge amount of prob­lems which can't eas­i­ly be solved oth­er­wise. When I worked in a small French com­pa­ny in 2013, we had a group of a half dozen per­sons in charge of de­vel­op­ing a new prod­uct. Peo­ple were quite mo­ti­vat­ed, as far as they could have been in a dis­cour­age­ment-cen­tered work­place, ex­cept one guy. He came at work, but re­mained there do­ing noth­ing. Not only was he par­tic­u­lar­ly un­help­ful, but he ac­tu­al­ly wast­ed our time, so we all agreed that get­ting rid of this per­son would be a good choice. The only prob­lem was that the guy was our man­ag­er's friend, and go­ing to the man­age­ment and telling that the per­son should leave our team wasn't an op­tion. And so he stayed.

When peo­ple don't chose who they work with, team syn­er­gy is lost. They un­con­scious­ly in­ter­pret it as that man­age­ment doesn't care about team pro­duc­tiv­i­ty, and so they wouldn't ei­ther. When peo­ple chose who they work with, they usu­al­ly make smart choic­es which in­crease co­he­sion and make team pro­duc­tiv­i­ty rock.

Un­like some man­agers be­lieve, self-or­ga­niz­ing teams don't im­pede man­age­ment's job, but im­prove it. The choice is among the peo­ple who are best placed to make it, and those peo­ple should be rather good at mak­ing good choic­es, much bet­ter than their man­agers. This is easy to ex­plain. For the team, it is cru­cial to en­cour­age the most suit­able per­sons to re­join the team, and to en­cour­age the per­sons who don't fit to leave the team. More in­ter­est­ing­ly, for the per­son her­self, it is im­por­tant to lean to­wards teams where the per­son could ac­tu­al­ly make the dif­fer­ence, and leave teams which are not a good place for her: nat­u­ral­ly, a per­son wants to be in a place where her work is en­cour­aged and where self-es­teem can be high.

This leaves man­agers with hard choic­es about peo­ple who don't fit in any of the teams, but here again, self-or­ga­niz­ing teams have their ben­e­fits. A per­son who was un­able to find her place in any teams would prob­a­bly guess that the com­pa­ny is not a good place for her, and ac­cept more eas­i­ly to leave. If not, well, the man­age­ment has a strong point by point­ing out to the per­son that she was un­able to fit any­where.