Why dedicated & stable teams matter. A case study.

Every Scrum trainer and every Scrum Master teaches us that stable teams with full-time team members are much more productive than unstable teams with part-time members. And every client or any given organization tells us that this isn't possible in their world. There are too much obligations, too few experts, the numbers are not relevant for this industry and so on. But this is damaging not only motivation and team spirit. Actually it creates a quantifiable financial loss.

Scrum makes visible what does not work

Photo by Luis Villasmil on Unsplash

This is a case study from our client work in Germany. We anonymized the data. We had one Product Owner for several Scrum Teams. The Scrum Teams worked for months and they still need to work many more months to finish the work. This is an agile island within a siloed company. The Scrum Teams improved the technical practices of this company a lot. Though we made significant progress compared to non-agile project work we still need much more productivity.

The Scrum Teams told us between the lines and several times that team members are obligated to work in other projects, too. We decided to quantify this financially to fight for our dedicated full-time team members.

Part-time membership makes results unpredictable

The first recognizable effect was unreliable delivery. We had committed availabilities of let's say 80% or 30% of several team members. The 100% people could finish their planned work, the others could not. They were called for immediate help that of course took longer than expected. This was a repeated pattern that blocked delivery. 

One might argue that this only prolonged the project, because we don't need to pay people when they are not working for our project. What's the deal here? Five hours booked, five hours paid. But this pattern reduces the team effectivity. We calculated that in effect five hours booked were only three hours worth of work. We referred to the stable teams pattern, but the stakeholders did not except this "assertion". We needed to provide different data.

How much loss produces part-time, unreliable membership?

Let's assume that we have 23 people with varying availability (20%-100%) in 4 teams. In this case 23 real persons are equivalent to 13 FTEs:

• 5 persons with 100% availability (no other projects): 5 full time equivalents
  
• 2 persons with 80% availability (at least another project): 1,6 FTEs
• 8 persons with 50% availability (at least another project): 4 FTEs
• 4 persons with 40% availability (at least two other projects): 1,6 FTEs
• 4 persons with 20% availability (at least two other projects): 0,8 FTEs
  

First we tried it with Weinberg's quantification of losses due to context switching. People with other projects have 20%-40% less productivity than those working only in our project. That reduces the 13 FTEs to 11 FTEs. The 13 FTEs have only a productivity of 84%. 

Most managers would accept that loss for juggling all obligations in a company. This less seems small, but the numbers add up: 5,000 person days with an internal day rate of 500 EUR (2.5 Mio EUR) bring only value of 2.1 Mio. EUR of work. 400,000 EUR are lost. How many developers can you employ for that money? But it's worse.

There is a study that looked on 160,000 projects managed in ca's ALM platform. /1/ The authors conclude that stable teams are 60% more productive than unstable teams. The 13 FTEs have only a productivity of 63%, or the equivalent of 8 FTEs.

Given that our above mentioned project budget of 2,5 Mio EUR is only 1,6 Mio EUR worth (2,5 * 1/(1+60%)), a loss of 900.000 EUR. Again: how many developers can you employ for that money?

Is it better to take 8 instead of 23 people, but have them full-time in our project? Do the save time?

Smaller teams need less time to coordinate and to communicate

Let's have a look on the communication paths. The number of communication lines between a given number of nodes is n * (n-1)/2.

• 23 people have 253 communication lines.
• 13 people have 78 communication lines.
• 11 people have 55 communication lines.
• 8 people have 28 communication lines.

If you double the number of people, you'll need roughly as many as four times the number of links between the people. The number of links is a substitute measure for necessary conversations to get work done.

One question is open: WHY? Why are stable (and small) teams more productive?

Mutual responsibility and shared knowledge drive productivity

Solving complex problems means that all team members need a shared understanding of the problem and possible solutions. We saw that part-time members only feel responsible for their part of the work. A team member of a strong team would always say: "I don't know what the others are talking about. But I need to learn it now for helping them. There are no other people than me here." 

Bottom line: 23 people, most of them part-time members in several Scrum Teams have the same productivity like 8 full time members. The company has to pay 2.5 million EUR for the work. They pay 100% but only get 63% of productivity, a loss of 900 thousand EUR. What will happen with people in your company if they steal 900 thousand EUR?

References

• /1/ CA Technologies: The Impact of Agile. Quantified., published 2017, available at https://docs.broadcom.com/doc/the-impact-of-agile-quantified