I was recently asked to give a talk on data change-making. I shared a story at a very relaxed pace – probably something to either listen to over morning coffee or to put you to sleep! Anyway, it was aimed at data analysts. I hope it contains some helpful advice. If you’d prefer to skip the video and get the tl;dr, it’s this: to make your analytics more actionable, learn the dynamics of your business from a domain expert before you analyze any data. When you present your analysis, show only the data you can validate.
As hard as I am on analysts and analytics, I sympathize with them almost exclusively. After all, I was an analyst for almost a decade. When I think about how data transformation falls short, I feel business leaders are far more responsible than their analytics practitioners. There’s a tension these leaders introduce that frustrates most every analyst or data professional. I’m not sure I’ll do the maddening nature of this tension any justice in a blogpost, but let me see if I can capture the essence here:
Business Leader: “Analyst, show me new, exciting and strategic insights about my company using your superior analytics.”
Analyst: “Yes, ma’am! I hope you are ready, we have prepared a comprehensive view of our business – marketing, sales, and customer management. We’ve found a trend you need to see!”
(Shows trend & business recommendation)
Business Leader: “Thank you Analyst, but there’s a lot of data I’m not familiar with and I’ve never thought about any of this before. Can you simplify the insights for me?”
Analyst: “We thought you might say that, Business Leader! We’ve prepared a simpler view of this analysis without as much clutter – see here our single key metric in a single bar graph. At the bottom, please read our three-word recommendation.”
Business Leader: “Analyst, I’m reading your analytics here and I’m surprised how little context you’ve included about our business. Can you do more research and take more things into account before presenting your findings next time?”
I hope I got at least one analyst reading this to smile. Maybe even laugh? Cry? If I had a dollar for every time someone sent me on this circle to nowhere as an analyst, I promise you I’d be on my beach right now, not blogging to you about how crazy this all is. It’s preposterous. Want to know what I’ve learned from all of this over the years? Here’s the first thing: there was nothing wrong with the analysis. It didn’t need to be shortened, tightened, abbreviated, and it certainly didn’t need to be “simplified”. Good analytics and insights can’t be simplified. That’s not how simplification works.
I believe simplicity comes from deep understanding.
Think about something you deeply understand – something you studied in school, your favorite sport, or maybe your professional practice or domain. How does it work? How would you describe excellence in this domain? I bet you think to yourself: “it’s simple really – it works like [X], and the best ones do [Y]”.
Now, find something that you know absolutely nothing about. Maybe it’s sports. Maybe it’s the arts. Maybe it’s data science, law or some other domain. Now, explain it to yourself. Go ahead! Try to explain even the smallest example. I bet you can’t. It’s complicated — where do you even begin?
When a client asks me to simplify something, I no longer assume that I’ve overcomplicated the message. I do not feel the datasets are too numerous, detailed, or overbearing. I’ve learned over the years to resist the urge to do the thing I’m being asked to do. Now, whether it be a client, a co-worker or a boss, when I hear “Can you simplify this for me? It’s too complicated,” I now just hear “Please teach me more about this so I can start to understand what’s going on.”
Here’s my advice to business leaders: Stop asking your analysts to simplify their analytics, insights and recommendations. Instead, admit that you do not understand your business’ data landscape or domain well enough, and just ask your analytics leaders to teach you more about the data behind your business. Analytics won’t look so complicated anymore. Your company might actually start to adopt a data-driven transformation after all.
