How To Begin
In our opening article, we defined analytics as —
the scientific discipline of developing reliable insight
We also noted that in subsequent articles we will detail How To Break It Down and shared our silly acronym/mnemonic. We also made the claim that –
it takes the analytic process to teach the analytic process
Which both catches you up and brings us back to How?
It is all a bit recursive…
Ideally, great analytics starts with a well-defined question. Practically, it begins by using the analytic process to develop one.
So let’s jump to practically.
Undisciplined analysts ask themselves — what question should I be answering? Disciplined analysts ask — how do I develop a well-defined question? It is a subtle but important difference. How implies a process and that is significant and measurable — two terms we will see again.
A well-defined question often starts with a lot of poorly defined context. This is most often delivered in the form of a client request. The client, typically not a disciplined analyst, does not position there request in well-defined terms or with great knowledge of the data, resources, and tools available. Their request can almost always be summarized as — the business wants to know this… The subsequent interrogative is almost never How. Typically — it is what or why?
The goal then is to use the analytic process to translate a request full of context into a well-defined analytic question. So now you may be thinking…
How Do I?
Perfect! You started with How and you got to I! Great start.
I is for information, inventory, and investigation. What do you know? What do you have? Where can you look? What follow-up questions should you ask? Define the time frames, events, people/populations, etc. We will develop this in more detail in our next article. But you get the idea, right? Just iterate.
Probably? Possibly? Problem is…
Perfect! There is no problem… well there is, but you are using the process. It takes you to P. Once you have the information, you move on to processes, patterns, and testing… of predictions. There is plenty more to detail here, but people with an analytic nature tend to have an intuition for this piece. If you don’t, have some patience — we will get their soon.
Note — you get to ask a lot of questions during this phase. They need not start with how. If they do, you are actually nesting deeper.
So far — so simple?
Only… Once I do the process? Oh, boy…
Okay, no worries! It is time for order, organization, and structure. We often refer to this phase in question development as framing. It is about structuring your information, ordering your processes, and organizing your outcomes. I have always argued that this is the step in the process where truly gifted analysts excel.
At this point in the practical question forming process, we have arrived at form. You should have a solid idea of what your question is forming up to be. Ready to move on to defining the details?
Details? How do I determine those?
Decisions, decisions. We set out to develop reliable insights and create a well-defined question. We gathered information. We processed it — looking for patterns and making predictions. We organized it and gave it form. At this point, we are well informed. Which is the perfect time to define and detail our question.
A great analytic question and process is disciplined and reliable. The only way to assure that is to know that it is measurable. You need to be sure the question frames a decision and that the decision once decided can be measured. There is much more to say… we will leave that TBD.
Storytelling time! Time to share the question. This is an easy step for most folks to understand, but analytic storytelling has some subtleties. Analyst are teachers by necessity. You don’t just return to the client and say — you gave me a lot of random context so the question I am going to answer is this…
You need to check your analyst ego. Synthesize your process. And share it!
This is the information you gave me (succinctly). This is the process I went through (simplify). This is the way I organized it (visualize). These are the decisions and details we are looking to measure to determine if we chose the right approach (statistically). Sound like a solid strategy?
I told you it was a bit recursive. Next article, we will focus on I. If I had to predict, we will likely repeat ourselves… good processes often do. But we will also elaborate and develop the framework further. So stay tuned. And thanks for reading!
Gurupriyan is a Software Engineer and a technology enthusiast, he’s been working on the field for the last 6 years. Currently focusing on mobile app development and IoT.