In my last blog posting, I provided some advice on how an organization could go about building a data map, in order to establish a baseline of the data assets underlying their enterprise. I had suggested that we’d follow that up with a column on organizing those assets for analytic purposes, but I want to take a short detour, and talk about a different challenge that is really foundational to everything else we are discussing –and that is the topic of “becoming data driven”.
You hear this phrase a lot –I think it’s safe to say we can enshrine it as one of the building blocks of analytics Bingo these days, but you don’t necessarily see a lot of useful discussion about what it means, in practical terms. So, let’s talk about that.
Let me start with my own, personal, core belief about this –and it is that becoming data driven has a lot less to do with software, tools, BI, analytic techniques and wonder weapons, and a lot more with your organization’s culture, leadership, and philosophy about decision making, and decision support. Things that do not make your company data driven:
- Buying an expensive Business Intelligence tool, the necessary usurious licenses, and installing it on multitudinous desktops.
- Hiring Data Scientists and creating CDO titles.
- Creating a data mart, warehouse, or lake and accumulating metric whack-tons of data.
Most of these things may have a place in a company evolution towards being data driven, of course –they may be necessary(at some point) but they are definitively not sufficient(at any point), to drive the transformation. Put another way, in my years as an analytics consultant I have worked with many smaller companies whose toolkits were basically just Excel based and were far, far more data driven than giants that had spent millions on the items illustrated above, and were failing to address the core changes necessary to actually utilize them
.In the absence of commitment by the leadership team to make the necessary, harder changes to their operational processes, then these kinds of investments become, as an earthy friend of mine likes to say, examples of “putting lipstick on the pig”. I can testify, as I have seen a lot of pretty pigs in my travels.That came out kind of weird, but you get the point.
The trap that companies fall into, of course, is not utilizing the capabilities that these kinds of toolkits deliver, in a “data driven” manner. You fundamentally have to a) trust the data and b) be willing to have courage of your convictions to drive the outcomes that the analytics illuminate –and that is often torpedoed by leadership-centric issues of politics (not wanting to make hard choices, make somebody mad, or deliver unpopular messages), expediency (avoiding conflict by following inefficient paths), procrastination (kicking it down the road to a “better” decision point), or cults of personality (another name for management by decree or by I-know-best). I have worked FOR companies where that list constitutes the entire operational methodology. We laugh, but everybody reading this knows it’s true, and probably sees some of it at their own company, every day that they come to work. If you’re a leader, and your reaction is “not at my company!”, well, good luck!
Because let’s not kid ourselves –these are very common problems, to a greater or lesser degree, at many companies. The kinds of organizational behaviors and dysfunctions are the biggest barrier to “becoming data driven”, not the lack of shiny tools and cool software.
So, this begs the question –what is the CEO, CMO or COO that is truly committed to making this happen supposed to do?
At its very core, becoming data driven means being fact-driven. It means making more efficient, informed decisions. It means having what fighter pilots call maximum situational awareness–striving for near perfect clarity on the state of your operation, and relentlessly seeking highly informed insight into what is likely to come in the near, intermediate, and longer-term future. It means embracing measurement, and celebrating the results –both “good” and “bad”. Those qualifiers, by the way, are probably holding you back, right now. What you want, as a manager, is accuracy –and if you want to get your people in the habit of thinking that way, you should be substituting “accurate” and “inaccurate” as your key descriptors for your numbers. Don’t punish people, at all costs, for bringing you numbers or analysis that you don’t like –if it is accurate. Reward honesty in measurement, regardless of the relative interpretation.
I should take a moment here to interject –for the sake of history, if nothing else, that the concept of being data-driven is not new, is not unique to advanced analytics, and has been a goal-state of excellent business practice from the time the first market was created. If you’re a manager, and haven’t read W. Edward Deming’s “Elementary Principles of the Statistical Control of Quality” –put it on your list. The first components of it were published in the mid-1940’s.It describes a methodology that was used to revolutionize manufacturing and ensuring quality by –you guessed it –embracing measurement and statistical tools to define when a process was in control, and when it was not.
What IS new are the capabilities that advanced analytic toolkits deliver –and the mechanization, via software, of data handling tasks that can be leveraged into better understanding future states, as well as current states.
If you are committed to reaching this goal of becoming data driven ,then you are likely to going to take your company on a journey through the three levels of analytics:
Descriptive Analytics –Focused on maximizing the utility of the datasets generated by your current operation, supplemented with other data sources, to maximize efficiency.
Predictive Analytics –Deploying tools that will allow your operation to anticipate customer needs, and to model forecasts and scenarios of possible business scenarios (product launches, for example).
Prescriptive Analytics –Currently much debated in definition (and outside the scope of this article), but grounded in the implementation of advanced AI and machine learning techniques to address complex, multi-variate questions. Characterized by a state of maximum automation, it can be thought of as the point where algorithms begin to manage much of the operational decision making in an enterprise.
So, how to begin that journey? Issue an RFP for an advanced BI tool, right? WRONG! You have homework to do, my friend. Developing the roadmap that will eventually guide you through this journey means coming back to the core of what being “data driven” means –understanding the current state of your operation, and what the priorities are for you to achieve the reasons we’re all here, and you’re reading this:
1)Drive revenue and growth
2)Reduce expense and grow margin
3)Increase the efficiency of the operation (in many cases, cost avoidance rather than cost reduction)
To make these things happen, in a data-driven way, then you have to begin by understanding what barriers are blocking progress across these strategic goals. I advise a company to start with a very straightforward exercise –the Business Threat Assessment.
Becoming Data Driven: Step 1 -The BTA:
This is the foundational step to all that follows –it establishes the priorities that analytic solutions need to address, and, the good news is that it’s entirely in the control of the company to achieve. The company needs to answer three fundamental questions:
1)What are the three greatest tactical (next 1 to 2-year horizon) threats to the operation’s success?
2)What are the three greatest strategic (next 3 to 5-year horizon) threats to the operation?
3)What are the three greatest, persistent operational issues that the company seems to face, year after year?
Some words of advice about this –if, on reading this list, your first impulse is to reach for the phone and call a high-powered (expensive) business consultant to come in and execute this –you’re already off the rails. This is an exercise that any company should be able to accomplish without any external help –and if you can’t, you have bigger problems that analytics can’t fix. Get the bright leaders in your company to spend a day on this –and if your feeling is “I can’t trust this to be done right”, then do pick up the phone, call recruiters, and get on top of your real issue.
Let’s be honest –anybody in a leadership position should have a pretty good idea of the answer to these questions. If you’re not talking about them today, in a regular fashion, then your first step on the way to becoming data driven is to institutionalize this list, refresh it on a monthly basis, and focus the leadership team on addressing it.
Why is this first step necessary? Because being data driven means committing to a process of continuous improvement. As leaders and managers, you are prioritizing your people, their assets, and their efforts. If you’re not clear on the size, pressure and importance of the challenges facing the enterprise, you’re not able to task anybody effectively. You should be developing plans that deliver the maximum return to the business along the three key metrics we’ve discussed (growth/cost/efficiency), and those plans are NOT one and done –they are a continuous, reinforced, optimized set of decisions that are consciously selected to deliver the maximum, measurable return.
To be continued: In our next post, we’ll talk about how to take the outputs of the BTA, and do the exercise of asking the question –how can my current data assets and KPIs help me address these challenges? We’ll be in the land of descriptive analytics, and talk about taking a hard look at your current reporting infrastructure –before you spend a dollar to change it.