Good Management
Management is a User Study
A different perspective
I come from a data science background, and I have designed a lot of user studies over the years. I’ve been managing a team for seven years, and after a while, I started to see a lot of commonalities between how you design and implement a user study and how you manage an effective team. Let’s discuss:
User studies
User studies aim to collect data to better understand a process. In the world of the machine learning that means to collect data to train an algorithm or to test an algorithm.
So let’s take an example: toe recognition
The goal of user study to collect data to train an algorithm to recognize toes. You have multiple variables to look at including the population, age, and ethnicity that are intrinsic to the person. That is collected by recruiting broadly.
Then you have the extrinsic properties like toe pose, toenail length, and nail polish. Additionally, there is lighting, environment, and shadows.
For a Design of Experiment (DOE), you design a Standard Operating Procedure (SOP) to guide what the subject will do, and this where you have two trade-offs:
- Compliance
- Cost (time/money)
You want the user to comply to an SOP, and they want the compensation. However, a bad SOP can make compliance difficult. Often, a moderator is needed to help the subject follow the SOP. Many people have trouble following directions verbatim.
Compliance is not free; compliance costs time and money. You need more moderators to have better compliance, and you need people checking the data after collection to enforce compliance.
Management
A company has employees, and the leadership wants the employees to produce something. They offer compensation to the employees for compliance to a SOP as well as spending time to train an employee to comply to an SOP.
Some jobs require the person to do the same job over and over again, and the SOP calls for consistency. Some jobs like maintenance require fixing problems that aren’t necessarily known, but often there is either an SOP or one is created. Then there are jobs where the SOP is about the culture not necessarily the route to get to the final produce.
A high level of compliance comes at the cost of creativity. The tighter one has to fit to a mold, the less likely they are to think outside of the box, and this diminishes impact especially if the person doesn’t think inside the box at all. The box is the SOP.
The best SOP’s start out generalized and become individualized because everyone works a little differently. The job of a moderator is to help with the individualization of the SOP to guarantee quality. The same is true for the manager
If you want good data, you iterate on improving how you run a user study, and make changes for things that don’t work. If you want good management, you should treat it like a user study.
Are you efficiently getting results you want or are you even able to?
Is the DOE correct?
Is the SOP correct?
Iterate, test, repeat.
