Marketing consultants tell about the shoe company that sent two salesmen to Africa in the 1800s. One sent back this message: “There is no market here; nobody wears any shoes!” The other wrote back: “There’s a tremendous market here; nobody wears any shoes!”
Beyond our standard goals (such as “how well we meet the service objectives we promised in the SLA”), we have our non-standard, unstructured goals such as “what could I be doing for my customers and what needs do they have that I could meet, which I’m not even aware of?”
Do none of your end users wear shoes? How do you find out whether they need shoes, or think they do? How could the concept of “shoe” be modified to give them something useful?
- Ask your end users. This can work if they know what they need, but won’t work if they don’t realize a need.
- Do your homework analyzing their operations, looking for ways your specialists in data analysis, functional analysis, cross-platform interaction, and other IT disciplines can help identify opportunities. (We discussed how to go about this in the November/December 2009 column.) This can work if the unrealized opportunities lie within the standard realms and dimensions of the IT disciplines.
- Find ways to look outside these boxes. To develop this ability, consider these scenarios and what causes the possibilities to be hidden:
First scenario: You send a postcard to 100,000 people selected at random from the phone book. Half of your postcards predict that the Republicans will win the next election; half predict the Democrats. Fifty thousand people will subsequently believe that you predicted correctly.
Send those 50,000 another postcard. This time tell half of them that the weather, the stock market, interest rates, or something else will be one way. Tell the other half the opposite. Half of these 50,000 will soon believe you made two correct predictions in a row.
Continue the process until you have a small number of people who believe that you predict anything reliably. They will be eagerly awaiting your next postcard. Ask them to pay for your next prediction.
Second scenario: A friend buys a new sports car and brags about its great mileage. Once a week someone sneaks into his garage at night and adds a few gallons of gasoline to his tank. He will brag even more about his incredible mileage. After several weeks of this, the same person starts siphoning off a few gallons of gas per week from his tank. His bragging may be replaced with a puzzled look.
What’s happening here? In each case, someone is presented with information that falls outside normal bounds. For a variety of reasons, our eagerness helps us ignore that something is out of bounds. Statisticians call out of bounds “outside the control interval”; that is, outside the pre-established range of normal. This is the basis of statistical quality control. (Think of Dr. Deming and how Japanese automakers raised automotive quality standards.)
Factories, software developers, semiconductor fabs, and other installations use this concept to maintain extremely high quality. They determine what to measure, set “normal” control boundaries, and set their control mechanisms to inform them only when some measurement is out of bounds. Investigation of out of bounds conditions often leads to unexpected opportunities and unforeseen causes of problems.
However, the measurements involved often aren’t the main story. An abnormal streak of postcard predictions or wildly fluctuating car mileage may indicate a variety of possible causes. It’s the investigation of these causes that leads to unexpected opportunities.
Discuss with your managers what measurements you might want to use to define normal boundaries. Set up control mechanisms to flag “out of control interval” instances. Do this for measurements you already monitor, but include some you’ve never monitored before. Do this with measurements for your own operation as well as for your end users’ operations. The better the choice of things to measure, the more interesting the opportunities you’ll uncover.