Metrics_ How to Improve Key Business Results - Martin Klubeck [27]
Conduct Surveys
Surveys are probably the most common data-gathering tool. They are used in research (Gallup Polls), predictive analysis (exit polls during elections), feedback gathering (customer satisfaction surveys), marketing analysis (like the surveyers walking in shopping malls, asking for a few minutes of your time) and demographic data gathering (the US census). Surveys are used whenever you want to gather a lot of data from a lot of people—people being a key component. Surveys, by nature, involve people.
The best use of surveys is when you are seeking the opinions of the respondents. Any time you collect data by “asking” someone for information, the answer will lack objectivity. In contrast to using automated tools for collecting (high/total objectivity), surveys by nature are highly/totally subjective. So, the best use of the survey is when you purposefully want subjectivity.
Customer satisfaction surveys are a good example of this. Another is marketing analysis. If you want to know if someone likes one type of drink over another, a great way to find out is to ask. Surveys, in one way or another, collect your opinion. I lump all such data gathering under surveys—even if you don’t use a “survey tool” to gather them. So, focus groups, and interviews fit under surveys. We’ll cover the theories behind the types of surveys and survey methods later.
Use People
So far I’ve recommended avoiding human provision of data when accuracy is essential. I’ve also said that when you want an opinion, you want (have) to use humans. But, how about when you decide to use people for gathering data other than opinions? What happens when you use people because you can’t afford an automated solution or an automated solution doesn’t exist?
I try to stay fit and get to the gym on a regular basis. I’ve noticed that a gym staff member often walks around the facility with a log sheet on a clipboard. He’ll visually count the number of people on the basketball courts. He’ll then take a count of those using the aerobic machines. Next, the free weights, the weight machines, and finally the elevated track. He’ll also check the locker room, and a female coworker will check the women’s locker room.
How much human error gets injected into this process? Besides simply miscounting, it is easy to imagine how the counter can miss or double-count people. During his transition between rooms, areas, and floors of the facility, the staff member is likely to miss patrons and/or count someone more than once (for example, Gym-User A is counted while on the basketball court, and by the time the staffer gets to the locker room, Gym-User A is in the locker room, where he is counted again). Yet, it’s not economically feasible to utilize automated equipment to count the facility’s usage by area.
We readily accept the inherent inaccuracy in the human-gathered form of data collection. Thus we must ask the following:
How critical is it to have a high degree of accuracy in our data?
Is high accuracy worth the high cost?
How important is it to have the data at all? If it’s acceptable to simply have some insight into usage of the areas, a rough estimate may be more than enough
Many times you collect data using humans because we need human interaction to deal with the situation that generates the data. A good example is the IT help desk. Since you choose to have a human answer the trouble call (vs. an automated system), much of the data collected (and later used to analyze trends and predict problem areas) is done by the person answering the phone. Even an “automated” survey tool (e-mails generated and sent to callers) is dependent on the technician correctly capturing each phone caller’s information.
Another Example
I want to provide you another example of how to develop a