Separating the Signal from the Noise

Say I gave you a hammer, some nails, a mitre saw, and a tape measure and said, “Go build it and make the most use out of it” and then sat back and waited for your response. You’d probably look at me expectantly, waiting for me to finish. But what if I were finished? What if there were no more instructions? You’d have a lot of questions, wouldn’t you, and you probably wouldn’t get very far if you decided to go and build whatever it was you thought I was talking about — or if you did, it most likely wouldn’t be what I wanted.

This example is exaggerated, but essentially, it’s the kind of demented dance that a lot of companies go through when they build analytics systems. They acquire analytics technology without an adequate understanding of their information needs and without having realistic expectations in about the benefits the system will be expected to return.

When companies pursue analytics initiatives, they’re typically looking for one thing: better information to improve decision-making capabilities. What many don’t realize is that they already have a wealth of information. What they need is the ability to parse that information — to sort out what they need from what they don’t — to determine what new information they need. They need the ability to separate the signals from the noise.

The purpose of analytics is not simply to collect information. It is to get the information needed to understand which problems need to be solved and to understand what decisions that must be made to achieve the solutions. That information might not reside in your information systems, and it might have to be augmented with new information, but the bones of intelligent decision-making are typically there — resident in the experience, gut instincts, and inside industry knowledge of your decision makers.

Still, many organizations struggle to understand what information they need to augment what they have. They struggle to ask the right questions. The thinking often goes, “If I only knew what questions to ask, I’d know what information I needed.” That’s true, but no one is going to tell you the questions. No one can.

Your business is unique, as are the questions you need to ask and answer. However, you can use your experience, and the knowledge of your environment, to develop those questions, and determine what decisions you need to make. The process requires you to go outside the box, to think big and believe that there’s nothing that is unknowable. It goes something like this:

1. If I understood:

2. I could:

3. To achieve:

There is already so much knowledge resident in your organization. Good analytics is about using the information you have to ask the right questions about your business, so that you can use your analytics tools to gather additional information you need, and determine what decisions you need to make to meet your goals. It’s about helping you understand what information is relevant — a good signal that bears listening to and acting on — and what is merely noise that you should ignore.

a. Analogy of college and foreknowledge of questions on the exam

i. Learn the principles of the system (the business environment)

ii. Systematically go through the problem and determine what information you have, what you need, and what you don’t

iii. Ask the questions you need to solve the problem. How do you recognize what you need?

1. Look at the hard numbers

2. Then apply the experience, gut instinct, and industry knowledge to understand the missing pieces you need to solve the problem

Bottom line: If you know which questions to ask, you can use your analytics, coupled with the ‘thick’ data in your organization — the experience, gut instinct, and inside industry knowledge — to understand what decisions you need to make, and what additional information you need to make those decisions. You can separate the signal from the noise.

Anu Jain is the Chief Growth Officer and Analytics/Data Partner at Nexus Cognitive. Prior he was SVP, Teradata and General Manager IBM. anu@nexuscognitive.com