Barriers to Added Value in Data Analysis

There are also Five Primary Barriers to value add.

NR - Not Relevant
NA - Not Actionable
NS - Not Sufficient
NE - Not Efficient
NI - Not Interested

Imagine, for example, that your automobile has 100 sensors and each sensor creates a datastream of 1GB per mile. That's going to generate BIG data. But what do you really care about when it comes to your car? What you might care about is how many chicks you can pickup. In that case, none of the big data adds value. = NR

You might care about how quickly you can stop in the rain. You have brake sensors and you have wet weather sensors. So you can use your big data to calculate your braking distance in wet or sunny weather. The problem is that you are 500 miles away from anyplace (or anybody) that can install new brakes or tires. = NA

You might care about trunk capacity. You have sensors on all four wheels but they all combine into one number. You can't disambiguate between weight in the front seat from weight in the back seat or shifting weight in the car.  = NS

You might care about gas mileage on steep roads as compared to straight roads. However to get an adequate sample size, you have to drive so many steep roads (and you live in Kansas) that it costs more to find the steep roads and then calculate the savings, (and add sensor 101, an inclinometer), than you could possibly save. = NE.

Notice in all of these, I am talking about what you care about. If you don't care, then you certainly aren't going to pay somebody to care for you. If your answer to all of this is 'so what?', or 'maybe later', then you clearly are not going to add value. This happens more often than you might expect. A BI project might just be seen as something tangential to the business, or as a special gift to the dorks in accounting and IT. 'Buy him an iPad to shut him up'. Such systems do not contribute value. = NI

---
Understand the goals of the business. Find out who has responsibility for what that can be expressed in dollars. Everybody has to know *something* to make the organization work. Determine what the computer can know.

 

Comments

[Cobb]

Yeah I modified it on April 16th.