Learn to see the three types of waste

Lean asserts there are three types of waste:

  • Mura, or unevenness: waste due to fluctuations in demand
  • Muri, or overburden: waste due to trying to do too much at once
  • Muda, or in-process waste: the traditional target of “process improvement,” e.g. having 5 steps in your process when only 2 are needed

To see these types of waste and their effects, you must open your eyes and look for them. It’s easy for your brain to become accustomed to “the way things are” and to filter out the evidence of these wastes. Here are some tips for how to find these three types of wastes in your IT department:

Mura (unevenness)

Sometimes you have 100 requests; other times you have none. When there is unevenness, you plan for the peaks in demand: infrastructure used for 2 days a year to handle registration demand; staff or temporary workers hired to handle the surge in demand the first week of classes. This capacity is wasted the rest of the year.

One of management’s jobs is to “level-set” work to the extent possible. (See also ITIL’s demand management process, which specifically focuses on how to understand and influence demand.) Level-setting may seem at odds with Lean’s reputation for “Just In Time,” but evening out demand is a critical component to ensuring flow and optimal resource levels throughout the day, week, month, and year.

When you have leveled demand, you can employ “takt time.” Takt time looks at the anticipated demand for the day and levels it into even time segments. For example, if you anticipate that 240 support tickets will come in today over 8 working hours, and there are 5 people who will work those tickets in parallel, the takt time would be 10 minutes/ticket (240 tickets/5 people/8 hours). These units of time–the “takt time”–is the pacing for the day. You try to work to the takt time–neither faster nor slower–to meet the demand exactly without rushing and reducing quality. Process improvements can ensure that work fits into the takt time. The takt time is affected both by the anticipated demand and the staff allocated to performing the process: management can pull one person out of the process to reduce takt time.

Unevenness comes from customer requests, but we also create our own demand. New services bring additional work, especially in the “early life support” phase when a service is first rolled out. How do you decide when to roll out a new service? Do you roll out services evenly over the course of the year to level the demand created by these new services?

Muri (overburden)

I heard a great line last year: if you ask people to do 15 things, don’t think they’ll get more done than if you asked them to do 5 things.

There is a cost to trying to do too much. Computer science has the concept of a “context switch”–the overhead of switching from doing one thing to doing another. When people try to do too much, they spend more of their time switching between work and less time doing the work.

Whenever you hear people talk about prioritization, they’re talking about the effects of overburden. When people ask for priorities, they want to be able to make calls about what’s most important of the too many things they are asked to do. How can we instead keep from asking people to do too much, so they don’t need a priorities list?

IT governance is an IT management capability that helps here: IT governance helps campus have conversations about what they most need. The more effective the IT governance structures, the less overburden there should be due to new work. IT governance should also make high-level decisions about existing service levels to manage workloads.

Overburden is a huge problem in IT. Here are a few of the many factors that contribute to IT staff overburden:

  • People’s time assumed to be infinite–i.e. “level 0 resource management,” a lack of awareness that resource management could be done
  • IT governance systems are not in place–there’s no one in an appropriate position to decide whether big requests from the Provost should be sequenced before or after requests from the Investment Office
  • IT governance systems do not control IT funding–departments receive funding for IT projects and then go to IT governance
  • Project management unable to predict implementation costs–inadequate scope definition processes or inappropriate methodology used to govern project implementation costs (e.g. iterative techniques and checkpoints needed)
  • Project scope control issues
  • Maintenance is assumed to be free–implementation costs are considered but not ongoing maintenance costs
  • Service management unable to predict maintenance costs–processes not built for understanding ongoing support costs
  • Service level issues–customers want service levels that require more resources than available
  • Existing maintenance costs are unknown–yes, a new service will require 10 hrs/week maintenance, but you can’t show how busy people are now

Additionally, many of the approaches used to understand resource allocation are difficult to implement either culturally or practically. People tend to bristle about tracking their time. Support people may flip between a dozen services in the course of a day, and none of their time cleanly rolls up. There is a lot of overhead in building good estimates for projects.

Muda (in-process waste)

Muda is the waste that’s created as a side effect of doing something. Thinking about muda, at any given time you are creating one of three things:

  • value
  • necessary waste
  • unnecessary waste

“Value” is strictly defined as work that immediately adds concrete value for the end customer. For example, if the customer wants a report to have a certain field, and you are adding that field, that’s adding value.

“Necessary wastes” are the supporting activities needed to add value. For example, you may need to start up your computer so you can get to your report editor program. The computer must be started, but you are not adding value while your computer is starting up.

“Unnecessary wastes” are performed activities that don’t really need to be done to add value. For example, you may need to log into the report editor program every time you start it, after already having logged into your local computer. Arguably that duplicate login with the same credentials is a form of unnecessary waste.

Using this mental model for in-process waste, three high-level activities for muda-related process improvement are:

  1. Maximize added value time
  2. Remove unnecessary wastes
  3. Make necessary wastes unnecessary

There is a huge amount of material about how to conduct Lean-based improvements related to all three of these wastes, particularly muda. Please see the Lean Enterprise InstituteLean IT‘s appendix C also has a really good list of “information wastes” to help think through examples of waste.