I had just gotten out of a discovery meeting with a client when she said “I’m sure we’re the worst you’ve seen”. The goal of the meeting was to better understand the process they were using, the data source, and how they were consuming that data. Turns out that data was entered into Share Point forms, fed into MS SQL Server, dumped into Access, cleaned up with VBA, and accessed with Tableau and Excel. I’ve seen plenty of scenarios like this in my time as a consultant, and in a previous life, I helped build a solution similar to this. These scenarios are not uncommon, and usually, the business has a problem that needs to be resolved and they figure out a way to do it with the resources they have on hand. Personally, I have no beef with these types of systems. I even like seeing them because it shows how people can problem solve with minimal resources; ingenuity can be a beautiful thing. Sure, like most people in the BI/DI/analytics/(insert all other related buzzwords here) space, I like to come into a client where everything is clean and segmented, but I suppose one of the fun things of this job is unraveling the ball of yarn and untangling the knots.

Now, back to what one of the clients said to me: “I’m sure we’re the worst you’ve seen…” I think it is troubling to hear that kind of self deprecation on such a regular basis. Why? I think that for a few reasons: Business Intelligence and Analytics (BIA) is a spectrum, data is necessary for modern businesses, and most businesses are not like the ones seen in the latest blogs or with the sexiest, newest tech. There are plenty of clients I have worked with that think that since they do not fit into these boxes, they are worst in class. Most need help, which is why I am there in the first place, but even the “best in class” usually need help in more ways than just a technical solution.

The BIA Spectrum

I think that part of the issue is based on semantics. To most people, BI and Analytics are defined like this:

  1. Business Intelligence: An analysis in which data is viewed post business activity to assess the business via metrics and indicators.
  2. Analytics: An analysis which uses past data to make projections as to what could happen.

The biggest difference here is that BI is backward-looking (into the past), and Analytics is forward-looking (into the future). This is a Boolean point of view, and frankly, unnuanced.

It also does not take into account the “unmentionable”: Operational Reporting. Yuck, right? Who the hell wants to do that? And that is (usually) the end of the discussion. The unsexiness of Operational Reporting means that it is forever pushed to the side, resulting in belabored sighs from clients claiming that they are “the worst you’ve seen”, just because they still have stuff running with VBA code to make sure the lights stay on and the orders are filled.

I think the duality of the current definition of BIA is wrong. Here is how I perceive the BIA spectrum:

In the wild, things are much more fluid than this.

This captures the entirety of what is happening in the business with reporting data; data that is being used to solve problems. Whether those problems are in the present, past or future is irrelevant, because if we want to have higher quality data, higher quality decisions and higher data accuracy, why should we only care about 2/3rds of the use case for curated data?

Why the Modern Business Needs Nuance

I think that we can boil down the definitions of those concepts into three simple questions.

Operational Reporting: What is happening?

Business Intelligence: How are we performing?

Analytics: Why do we care?

Here is an example to illustrate the above concept. Let’s pretend that a company called Company A manufactures charging equipment for smartphones, and that they have two product lines: Android, and iOS. When company A sees that the shipments from the warehouse starting to slow because the line workers headed out to the food truck for lunch, that is Operational Reporting. When Company A sees that shipment numbers are down compared to last month and customers are not getting orders by the promised date, that is Business Intelligence. When Company A determines that shipments are down because an increase in customers are not buying Android smart phone chargers, that is Analytics.

You may argue that this is all semantics, and you may be right. But I think that semantics matter. And, I think this example highlights another thing that is not usually discussed: if operational data can be used with curated data sets seen in BI and Analytic settings, the possibility for insight is greater. With the advent of Internet of Things connectivity, streaming with services like Kafka, and the ability to use nontraditional data types like JSON, integrating this data is a no brainer.

Is the Future Now?

Many times I come into organizations and see a mindset that what “we really need [is] a new tool”, and that this new technology will magically fix everything. While technological changes do need to happen, and many organizations need to adapt to current technological standards and methods, the tools themselves do not change the underlying issues (more on that later).

Technologists love the mantra that the future is now. Better yet, they love to say that the future arrives in waves (meaning that the future arrives at different times in different places). That may be true, but I think that simplistic mindset perpetuates a sense of urgency to be an early adopter of the latest technology. Buying and building a foundation for these concepts (or even BIA) takes considerable time, money and effort. This isn’t like heading down to the Apple store and buying a new iPhone every year. These initiatives are usually million dollar efforts that require multiple teams across multiple years. For comparison, how long does it take to buy an iPhone? An afternoon? Even by percentage terms, the amount of effort is not even comparable between these two tasks. Let’s stop pretending that buying a tool will fix the fundamental issue at hand, because in most cases it will not.

Another concept that we should take a good, hard look at is the “data lake”. I mentioned above that Internet of Things data can be streamed, captured and analyzed to benefit the business. But just because we can stream it and hold on to it, does that mean it has value? I hear many industry talking heads talk about data lakes, big data, algorithms and the like, but what is the use if there is no use case? Most clients I have worked with don’t understand what data they have or how to use it — is it really feasible to think that these clients will directly benefit from a solution like this? Long term, perhaps they will. Short term, there are better initiatives that can bring higher benefits for lower cost and effort.

A better approach is to focus on how we can use, and most importantly think about data. I dare you to go to HBR (a favorite business publication of mine, so I’m not hating here), and search for the term “analytics”. When I conducted this experiment, I had 1,521 results returned. So, clearly, people are talking about this topic. What are they saying? Here are some of the titles: “Figuring Out How IT, Analytics, and Operations Should Work Together” (Berkooz), “How an Analytics Mindset Changes Marketing Culture” (Sweetwood), “The Reason So Many Analytics Efforts Fall Short” (McShea, Oakley, Mazzei), and… I could keep going with examples. Each of these articles, and many others seen have great points describing what mistakes were made. I think that many of these mistakes continue to be made because businesses refuse to change their thinking about analytics, reporting, BI or whatever you’d like to call it. Einstein’s definition of insanity is apt here: “doing the same thing over and over and expecting different results”. I would be remiss if I placed the blame squarely on businesses, but consulting partners also have a role to play in this. Being a technical consultant does not necessarily mean only focusing on a technical solution; it is important to advise on the implications of these solutions and how the business should be thinking and acting.

Doctor, It Hurts When I Do This

These attitudes towards data lead to many situations I see with clients. Many times, I feel as though I am the one who needs to broach the topic and feel akin to someone who needs to tell a friend that s/he really needs an Altoid after that falafel pita they had for lunch. Because businesses are so focused on getting things done, they rarely have the time to focus on the self reflection that would lead them to recognizing these issues. These are difficult discussions — but would we rather not have them and let people drown in data quagmire? The items below are the road blocks that I most commonly see as deterrents from accepting a new mindset around data. Until these fundamental data hurdles are jumped, it will be hard for any organization to overcome an old world view and embrace a new one for a new world.

  1. Data is inaccessible. Many times, the people who need the data do not have ready access to it. This may be because data is difficult to extract (maybe they are accountants that do not have database access). Or maybe it is hoarded (someone has access but only sends limited amounts of information). Perhaps the data is disparate (meaning that it is spread thinly throughout the organization, typically in Excel files). These issues prevent people from making decisions and they spend many, many hours finding loopholes, work arounds and writing their own “underground” code and databases to compile this data, when they should be making decisions or correcting it in a different system.
  2. Data is indecipherable. Sometimes data is accessible but the data means different things to different people. For example, Jimmy may calculate “On Time Order Percentage” as ((Number of On Time Orders / Total Number of Orders) * 100), where Jane may calculate “On Time Order Percentage” as ((Number of On Time Orders / (Total Number of Orders — Number of Cancelled Orders) * 100). Who is right? In some respects, they both are; in some respects, they are both wrong as well. Because no one can clearly understand what is happening, the data loses it’s meaning. It always needs a qualifier, and thus, is value is decreased because no one fully trusts it. This is also an issue when the data set is complex. If someone needs to understand different codes that represent business processes (100 = order placed, 102 = order packed, 103 = order ready for shipping, etc), or if the data does not represent the business process, the common language between the end user and the data is destroyed, and not only does the data become useless, it becomes meaningless.
  3. Data lacks vision. Many times I see companies that “just want reports on X”. X could be order management, or accounting, or purchase orders. However, rarely do I see companies create and execute a vision for their corporate data. This requires thinking of data within several tracks. Operational reporting keeps the lights on; what do I need to do right now to keep us moving forward? Business intelligence provides the business with goals, key performance indicators and metrics to track performance over time. Analytics answer the hard questions regarding what is happening in the business and in the general marketplace; these are usually open ended and have a grey area in terms of what the answer is. The lack of vision is a massive detriment to many companies because it means they move disjointedly when it comes to strategy and execution, particularly when it comes to internal resources.

For those of you who are in consulting, I am sure you have more points to add to the list. However, the point isn’t about making a list, it is about recognizing the blind spots many companies have in regards to data. It’s hard for any organization to get on board with data investments when it can’t overcome obstacles like the ones above. It is important for all of us to speak out and help transform the landscape of data from a Boolean point of view into a varchar point of view; it may need some error handling, but at least you can get what you want out of it… most of the time.

An Ending, but not The End

All of the above is great food for thought, but how do we implement a plan to combat the mistakes of the past? How can we start a movement that changes how we think about and interact with data? I’ll be following up this blog with strategies and ideas for winning these battles. Until then, remember that the mindset that we bring to the table when talking about data matters as much as problem we are trying to solve.

You may also like

Leave a comment