I argued against multitasking a number of times. In fact, not that long ago I argued against it in the context of portfolio management too. Let me have another take on this from a different perspective.
Let’s talk about how much we pay for introducing too many concurrent initiatives in our portfolios. I won’t differentiate here between product and project portfolios because for the sake of this discussion it doesn’t matter that much.
Let’s imagine that the same team is involved in four concurrent initiatives. Our gut feel would suggest that this is rather pessimistic assumption, but when we check what organizations do it is typically much worse than that. For the sake of that discussion and to have nice pictures let’s assume that all initiatives are similarly sized and start at the same time. The team’s effort would be distributed roughly like that.
The white space between the bars representing project work would be cost of multitasking. Jerry Weinberg suggests that for each concurrent task we work on we pay the tax of 20% of the time wasted on context switching. Obviously, in the context of concurrent projects and not concurrent tasks the dynamics will be somewhat different so let me be optimistic with what the cost in such scenario would be.
If we reorganize the work so that we limit the number of concurrent initiatives to two we’d see slightly different picture.
Suddenly we finished faster. Where’s the difference? Well, we wasted much less time on context switching. I assumed some time required for transition from one project to another yet still, it shouldn’t be close to what we waste on context switching.
In fact, we can move it even further than that and limit the work to a single project or product at the same time.
We improved efficiency even more. That’s the first win, and not the most important one.
Another thing that happened is we started each project with the exception of the first one in presence of new information. We could have, and should have, learned more about our business so that we are better equipped to run another initiative.
Not only that. It is likely that technology itself or our understanding of technology advanced over the course of running the first project and thus we will be more effective building another one. These effects stack up with each consecutive project we run.
The total effect will be further improvement of the total time of building our projects or products. This is the second win.
Don Reinertsen argues that the longer the project is the longer the budget and schedule overrun. In other words, if we decided to go with all the concurrent initiatives we’d likely to go longer that we assumed.
In short it means that we do end up doing more work that we would do otherwise. Projects are, in fact, bigger than we initially assumed.
The rationale for that is that the longer the project lasts the bigger the incentive to cram more stuff into it as the business environment keeps evolving and we realize that we have new market expectations to address.
Of course there’s also an argument that with bigger initiatives we have more uncertainty so we tend to make bigger mistakes estimating the effort. While I don’t directly refer to estimates here, there’s an amplification effect for scope creep which is driven by overrunning a schedule. When we are late the market doesn’t stand still. To make up for that we add new requirements, which by the way make the project even later so we add even more features, which again hit the schedule…
A bottom line is that with bigger projects scope creep can get really nasty. With fewer concurrent initiatives and shorter lead times we get the third win.
Let’s assume that we’ve had deadlines for our projects.
What happens when we’re late? Well, we pull more people from other teams. Well, maybe there was one guy who said that adding people to the late project makes it later but, come on, who reads such old books?
Since in this case all our projects are late we’d pull people from another part of an organization. That would make their life more miserable and their project more likely to be late and eventually they will reciprocate taking our people from our future projects in a futile attempt to save theirs. That would introduce more problems in our future projects. No worries, there will be payback time when we steal their people again, right?
It’s a kind of reinforcement loop that we can avoid with fewer concurrent initiatives. That’s a fourth win.
Finally, we can focus on economies of delivering our products or projects. A common sense argument would be to bring time to market as an argument in a discussion. Would we prefer shorter or longer time to market? The answer is pretty much obvious.
To have a meaningful discussion on that we may want to discuss Cost of Delay. How much it costs us to delay each of these projects. It may translate to the situation when we don’t generate revenues or the one when we lose the existing ones. It may translate to the situation when we won’t optimize cost or fail to avoid new costs.
In either case there’s an economic value of delivering the initiative later. In fact knowing the Cost of Delay will likely change the order of delivering projects. If we assume that the last project had the biggest Cost of Delay, the first the smallest (4 times smaller) and the middle ones the same in the middle of the spectrum (a half) we’ll end up building our stuff in another order.
The efficiency of using the teams is the same. The economic effect though is vastly different. This is the biggest win of all. Including all other effects we roughly cut down the total delay cost by two thirds.
The important bit of course is understanding the idea of Cost of Delay. However, this couldn’t have been enabled if we’d kept running everything in parallel. In such a situation everything would be finished at the same time – at the latest possible moment. In fact, if we avoid concurrent work even the ultimately wrong choice of the order of the projects would yield significantly better economic results than building everything at the same time.
What we look at is a dramatic improvement in the bottom line of the business we run. The effects of limiting a number of concurrent initiatives stack up and reinforce one another.
Of course, it is not always possible to delay start of specific batch of work or limit the number of concurrent projects to very low number. The point is though that this isn’t a binary choice: all or nothing. It is a scale and typically the closer we can move toward the healthy end of it the bigger the benefits are.
