In â€œStop Obsessing Over â€˜The Teams
John Cutler argues “we shouldnâ€™t focus on â€œThe Teamsâ€, until the conditions for effective teams are in place.”
Most Agile methodologies/frameworks focus myopically on â€œthe teamsâ€ (front-line teams of individual contributors). Meanwhile, organizational dysfunctions (those causing 80% of the drag) persist.
And that you have to solve organizational problems first.
In a tweet, Melissa Perri adds
Companies usually ask me to â€œtransform teamsâ€ but many of the problems they need fix stem from the org level. Local optimization over global rarely works. You have to work on both.
Everyone’s right — the trick is, sometimes it’s agile methodologies are what let you find the global problems, and make wide organizational change necessary.
The dreamland utopia of â€œThe Businessâ€ and â€œThe Developers / The Teamsâ€ just working it out over mugs of coffee â€¦ is ultimately unfeasible in any orgâ€Šâ€”â€Šwith >1 teamsâ€Šâ€”â€Šthat simply tries to graft something like Scrum on to their existing structures, dependencies, and project culture.
When faced with something that seems intractable, you start with what you can control, improve what you can, and force the organizational change that seems unfeasible.
I worked at Expedia way back during a rough patch for the company, and if you’d asked “what’s going on here?” you’d get a litany of entirely different and well-supported answers from different organizations. Or teams. Or people at adjacent desks on the same team with different roles. Top-down “let’s unlock transformation” attempts went nowhere. The business deteriorated and everyone knew why, but no one could agree on the reasons, much less what to do.
Three of us Program Managers (Adam Cohn, Yarek Kowalik, me) got permission to turn our teams to Scrum from the Microsoft process Expedia kept after spin-off, which went:
1) Strategy comes from somewhere
2) Business crunches numbers, hands a direction to team in Business Requirements Document
3) Program Manager turns that into a full specification down to a really deep technical level
4) Spec is reviewed and iterated on until it gets stage-gate approvals
5) Development goes until it hits “Dev complete”
6) QA until it hits “QA complete”
…the whole thing.
There were all kinds of organizational and business problems, all the stuff Cutler points out. The teams weren’t really autonomous, the Program Managers were effectively Product Owners, Project Managers, and deputized Product Managers, it took forever to ship, what did ship didn’t do well — all of it.
We had to hand-lobby leadership and make ourselves pests until we got permission to try Scrum just on the team level, just on those teams. Adopting was rough, and there was a time where transparency and the ability to produce stats meant we were the only teams admitting we weren’t making progress.
But it worked. It transformed everything.
It started with the immediate team and radiated out:
- showing the teams’ work created a sense of increasing success where there’d been frustration
- seeing the effect of gates on productivity let us remove the “toss over the wall” steps and get everyone closer
- reporting how having design do one set of “red-lines” in the spec and then only respond to critical issues delayed iteration got us more and better collaboration throughout
- documenting the problems with releasing code helped drive efforts towards continual delivery
With the “engineering can’t ship” out as an excuse, it kicked off organizational change. Once, no one could agree on why things weren’t working in our organization. Now we could prove them, and when people agreed on the causes, we could solve them.
For instance, tracking the work done and the cost of it meant we could show executives the productivity-draining effect of business-level randomization. I could go and say “as the business, you told us to prioritize quick-wins in response to this metric, nothing happened and we made no progress on what you want for four weeks” and the leadership of that organization had to grapple with what that meant.
That started a conversation first about how much time we put into tech debt, and on-call, and soon it was a business-level reckoning of “are we doing the right things at all?”
Solving that drove change through the once removed business organizations: it required Product Managers who could act quickly, pragmatically, who were available and collaborative (I wrote about what I learned from working with one of them ). And for the first time, you could tell who was an effective Product Manager, and that organization transformed.
All of those changes were painful and it took way longer than anyone wanted. And if you’d started omniscent, knowing what all the barriers were going to be, sure, it’d have been way easier to make a list. But we’re not. Instead agile methodologies let you see and prove the barriers to the team’s work and force action, and then you get to see the next one.
When large organizations are faced with holistic questions, you don’t get progress. At best you get a weighty consultant-produced report on how you’re deviating from best practices, and each of the groups tasked with work finds some way to argue their bit of the findings don’t apply, or feet are dragged and no progress is ever made.
Agile methodologies allow you to flip that: to not have to solve the whole thing at once. To find the things that are small, and fixable, and get better as you build clear and incontrovertible cases for the next implementable organizational change. You don’t know what the effect of a set of wide-ranging best practice announcements will be, but you know having a designer embedded on product teams will make everyone more effective.
Obsess over teams. Not only is it the often the best way to create change, in large organizations beginning with with the team is the only place it can originate.