Alternatives to scaling
Here is an email I recently sent. (Slightly modified.) Said quickly, but some key points.
Just wanted to share these ideas quickly. Some may find them new, and some people may not.
First: In this context I am defining scaling as 3 teams working together on one product.
The reason we want to do this is to deliver it faster. And “it” is big (a year’s worth of work or more).
And there is fairly high coupling and cohesion in the code (or product, if not code).
Do NOT scale with a scrum team that can barely crawl. It is stupid and mean.
Teach them first to do at least decent Scrum. (I define this in one sentence as: They use all of Scrum as defined by the Scrum Guide, and have started to get some results, and at least understand (internalize) many of the basic principles.) I think it takes typically a year to learn decent basic scrum (without a really good coach).
That time frame is related to having a world of impediments, but mainly cultural impediments.
Option A: Divide and Conquer.
Often the big product does NOT have high coupling and cohesion. And it can be divided into 3 main parts. And the connectedness of the 3 parts is defined in a relatively few basic interfaces or APIs (or similar). Then: Let each team work on one part.
It has some similarities to scaling, but much simpler.
Option B: Velocity to 3x
Take one Team and fix their impediments so that velocity goes from 1x to 3x. That is, 3x more than that team’s original velocity (X). (I am assuming this is a “normal” team.)
Personally I think this is easier, cheaper and faster (delivers faster) than 3-team scaling. Seldom do 3 teams get to 3x in velocity.
Option C: Dream Team
Rather than scaling 3 regular teams, get a Dream Team.
Definitely easier, faster, cheaper. (Not cheaper per person, but cheaper in total.)
But not always possible.
Option D: “Best of” Team
Get the “Best of Team”. If you were going to do scaling, you have 21 people (in my simple example). Cherry pick the 21 people and form the “best of” team.
Again, often this team will get more done than the 3 teams working together. To me, this option is likely to be better than a scaled group, but not as likely as a Dream Team.
What I am VERY impressed by is: I also never see options A-D even considered.
Even though the company has ZERO evidence (in its situation) that scaling ever helped. Or the evidence is word-of-mouth. (Like a US general leaving Vietnam in 1972 and declaring victory…what do you expect him to say?).
They need to consider these options prospectively and retrospectively.
I think they will learn that sometimes these options are better than scaling and quite do-able.
Useful Scaling is Possible
I do think it is possible that scaling can be effective.
One more quick word: I would prove to yourself that you can do 3-team scaling well before trying anything larger.
I think it takes a year to do 3-team scrum well (for most companies in typical circumstances). Unless you have excellent coaching. Why these companies do not pay for excellent coaching more often is … it does not make sense to me.
This is my summary. You might like it. Very fast. A lot more to say. Maybe useful to you. Interested in your reactions.
Worth repeating: If you scale, scale down (scale less).