Put People on One Project
Individuals assigned to work on multiple projects inevitably get less done. Multitasking—attempting to work on two projects or two things at once—is one of the biggest drains on project team performance. Yet it has unfortunately become one of the busy manager's most frequently used tools. The reason for this, I believe, is that multitasking creates the illusion of progress and gives the manager the feeling that a problem has been solved. Really, though, in many cases the problem has been made worse.
Consider the case of Jon, a director of database engineering who managed a staff of database administrators (DBAs) who were woefully outnumbered by the programmers, testers, and other types of developers in his company. Jon was faced with allocating himself and his staff of five across more projects than they could handle. His solution was to create a spreadsheet like the one shown in Figure 10.5. Jon's spreadsheet allowed him to allocate DBAs across the various projects, which he did down to the 5% level. Five percent of an 8-hour day is 24 minutes. Through this spreadsheet Jon was telling Bill he could spend 24 minutes each on the Napa and PMT projects, Ahmed could spend the same on PMT and Spinwheel, and so on.
Figure 10.5 A portion of Jon's project staffing spreadsheet.
Did Jon really think that Bill would stop working on the Napa project after 24 minutes each day? Of course not. But he probably did think that Bill had enough control over his schedule that he could be close to 24 x 5 = 120 minutes in a week. What Jon was really doing in this situation was taking a problem (the correct allocation of resources) that he couldn't solve and pushing it down to the members of his team. What Jon should have done instead was push this problem up to his own manager.
Pushing problems toward the team is often a wonderful strategy. In fact, delegating problems to the team is at the heart of Scrum. However, when a problem is pushed toward the team, the team needs to be given the authority to solve the problem. In the case of Jon and his DBAs, it was obvious that one solution to consider was doing fewer concurrent projects. Without being empowered to enact that solution, they were put into an impossible-to-solve situation.
And they didn't solve it any better than Jon did. They invoked the age-old policy of "work on the project of whoever is screaming the loudest."
Time on Task Decreases with Too Many Tasks
Kim Clark and Steven Wheelwright studied the impact of multitasking on productivity. Their findings, shown in Figure 10.6, indicate that the total amount of time on task goes up when a person has two tasks to work on. After that, however, Clark and Wheelwright found that time on task decreased. In fact, with three tasks the amount of time on task decreased so much it was less than when an individual had only one task to work on (1992, 242).
Figure 10.6 The amount of time spent on value-adding tasks decreases with three or more concurrent tasks.
If you have only one task to work on it is almost a certainty that you will occasionally be unable to work on that task. You will become blocked by waiting for someone to return a phone call, answer an e-mail, approve the design, or so on. And so it makes sense that the Clark and Wheelwright study shows that a person with two tasks to work on spent more time on task than did someone with only one task. However, consider that Clark and Wheelwright did this research in the early 1990s.
What's changed since then? For starters how about e-mail, instant messaging, the proliferation of mobile telephones, and any number of ways in which we communicate? My theory is that the bars in Figure 10.6 need to be shifted one space to the left to reflect today's faster pace. I remember clearly the job I had back in 1992 when Clark and Wheelwright published their results. I remember times back then when I was at my desk and thought, "I'm caught up; I have nothing to do right now." Of course, I haven't thought that since 1992.
The pace of the world has accelerated dramatically. Just being a good corporate citizen takes more time now than it did in 1992. There's more to read, more to process, and more for each person to do. Merely being an employee should count today as a first task for each of us. The first project we are on counts as a second, and we are then already optimally productive. Any further projects we are assigned just make us less productive.
One of the main reasons that multitasking is so horrible is the task-switching cost involved. There is tremendous overhead in getting started on one task, switching to another, and then switching back to the first. The more tasks or projects we are involved in, the more likely we are to be interrupted while working on them. One study of members of a software development team found that team members are interrupted every 11 minutes (Gonzales and Mark 2004). If you're reading this chapter at the office, it is likely that you were interrupted at least once while reading.
When Multitasking Is OK
All of this is not to say that we should never allow multitasking on our projects. It is sometimes helpful. The key is to remember that a person who is multitasking and shared across multiple projects is likely to get less total work done than if she had been dedicated fully to just one of those projects.
Let's again consider Jon and his DBAs. Suppose each DBA could complete "20 database tasks" per day assuming that all database tasks are the same size. A DBA fortunate enough to work on only 1 project would achieve this level of performance. However, a DBA on 2 projects might complete only 16 database tasks per day. And a DBA on 3 projects might complete only 14 database tasks per day.
Although these reduced levels of productivity may look quite bad, they may not be. Suppose 1 of our DBAs is assigned to 2 projects and is to split her time equally between them. She will be able to complete 8 database tasks on each project. This may be the optimal use of her time if neither of the projects needs 20 database tasks done in a day. If neither project needs more than 8 database tasks a day from her, then she is better split between both projects than dedicated entirely to 1. From this we can extract the following guidelines:
- In general, and for the majority of a project's team members, multitasking is to be avoided.
- Multitasking may be acceptable if a person cannot be fully or nearly fully utilized on a single project. If we look back to Figure 10.5 and Jon's DBAs, we see that the Connect project was allocated three people with a total allocation greater than 100%. A better solution would likely have been to allocate a single person but for 100% of his time.
- Rather than have everyone multitask a little, it is better to have a few people multitask a lot. Figure 10.6 illustrates how the largest drop in time on task occurs after a person takes on the first task too many. In Jon's case, a better solution would have been to do anything possible to have two or three of his DBAs not have to multitask, even if that meant the others had to multitask even more.
The Corporate Form of Multitasking
Individuals feel compelled to multitask because the organizations in which we work attempt to multitask as well. The corporate form of multitasking is pursuing too many concurrent projects. When an organization takes on too many projects, people become shared across multiple projects, which leads to individual multitasking. The detrimental effect of multitasking then causes those projects to take longer, which leads to more multitasking near the end of the project when "we need to get started" on the next project.
An eight-year study of projects at a dozen companies and published in Harvard Business Review concluded that "projects get done faster if the organization takes on fewer at a time" (Adler et al. 1996). Corporate multitasking—attempting to make progress on too many concurrent projects—is what created the situation that Jon found himself in earlier in this chapter when he resorted to allocating his people to the 5% level.
Mary and Tom Poppendieck urge organizations to limit work to capacity. An organization that has more projects running concurrently than can be adequately staffed is attempting to work beyond its capacity. As they write, "If you expect teams to meet aggressive deadlines, you must limit work to capacity (2006, 134, emphasis is theirs).
Stopping the Treadmill
One of the happiest days of my life as a consultant was when I explained the impact of personal and corporate multitasking to the general manager of a large division of a big company. I could tell the message resonated with her. She asked me to follow her as she rose from her desk. We walked to a conference room near her office. She pointed toward a huge number of sticky notes stuck to the widest wall in the conference room and said, "We just made our plan for next year. There it is. Do you think we're doing too much?"
Her division had well over 100 developers but the wall was full. We talked about the plan, the number of concurrent projects, and the ripple effect that would occur if one project was substantially late. She knew they were planning to do too much, and I confirmed this for her. She convened a meeting for the next day of the vice presidents and directors who had made the plan and instructed them to start taking projects off the board. A look of relief (and surprise) went across the faces of everyone present. They had each known that the plan they had created the week before was overly ambitious and would not happen. However, no one had been willing to say so.
I checked back with this general manager a year later and was delighted—but not surprised—to hear that her division had just completed its most successful year ever. Part of that was attributable to the adoption of Scrum and the improvements it brought across her department. But an equal part of the success was attributable to the focus that was brought to each project by having fewer projects in progress at one time.
As this anecdote shows, often the best way to stop multitasking is to stop cold turkey. However, the reason I was so impressed with this general manager is that she is one of the few I have seen with the courage to do that. If you can't stop immediately, or if you're not in a position within the organization to make such a far-reaching decision, there are other things you may want to try.
Don't start a new project until it can be fully staffed
Avoid the temptation to start a new project with just a few analysts and maybe one programmer. Try to get everyone to agree that new projects will be started only when they can be staffed with all disciplines represented. This isn't to say you need to wait to start a large project until all 50 developers are available. Starting a new project only when at least one full team can be fully and appropriately staffed will help adjust the rate at which new projects are started to closer to the rate at which they can be developed.
Include ramp-up and wind-down time in enterprise plans
If, like the general manager in this section's story, you put together a big, annual plan, be sure to include the time necessary to start and stop the various projects. All too often a team provides an estimate of six months, and six months are reserved on an enterprise calendar. However, even on a Scrum project (especially from a new Scrum team), there may be a month or two of wind-down. During this time at least a subset of the team may be needed for high-priority bug fixes or to implement great, new ideas that were discovered only upon release. Failing to plan for some of this will cause unexpected periods of overlapping projects.
Institute simple rules
Gaining agreement on simple rules can help lead to the right organizational behavior. A simple rule such as "No one can be assigned to more than two projects," can work wonders. Johannes Brodwall, chief scientist with Steria in Norway, suggests one simple rule.
- Everyone on the team must be at least 60% allocated to the team. Sixty percent seems to be a magical number, which says to people, "This is the most important thing." With 60%, when one task suffers, it is usually one of those 10% or 20% tasks. So this structure guides people to be more dedicated to their primary team.
Go Slow but go
I can totally respect the leap of faith required to believe that doing fewer concurrent projects will lead to more projects being completed. Even if they believe that completing projects more quickly will ultimately lead to increased productivity, people will be uncomfortable postponing or canceling large-scale projects. So, start small: Remove one project from the first quarter plan and see how it goes.