Let me start off with this fact. I am not a data scientist. I have studied data science. I have a degree that says I could be considered a data scientist, but I am not a data scientist. If data science were a language, I speak the equivalent of resort Spanish. I can order a meal, ask where the restroom is, but if you wanted me to be a translator in a high-stakes negotiation, it wouldn’t go well!
What I am is a business professional that has always been fascinated by the search for unique ways to solve complex problems. I love applying technology and organizational change to gain insights that were previously very challenging to access. To that end, I have worked at and with companies to help bridge the gap between technical personnel and the business leaders that are trying to gain a competitive advantage. As part of that effort I have seen a few patterns emerge within companies that seem to be obstacles (sometimes real, but often imagined) in trying to work intelligent technology into their businesses.
Here are a few of my observations:
You’re not too late! Using data science is still a relatively new phenomenon ... be curious.
Sometimes industries that are not intrinsically rooted in technology believe that data science-based tools like artificial intelligence (AI) or machine learning (ML) don’t apply to them. “We aren’t Uber!” These are considered “old line” industries, such as traditional manufacturing or distribution. Leaders in these companies tend to feel they are behind the curve when it comes to applying science to their business, or that there isn’t a practical application for what they do or the customers they serve. This couldn’t be further from the truth.
It’s important to understand that incorporating this type of technology is relatively new to everyone. The term “data science” wasn’t even coined until 2001. Supply chain management went through a similar evolution. Modern supply chain tools and techniques began to appear in the early 1980s, gained steam in the mid-1990s and today are considered an integral part of any well-managed organization. Bringing new approaches into a business usually starts gradually as awareness grows at a senior leadership level and capabilities are developed as new blood is brought into a company.
“The important thing is not to stop questioning. Curiosity has its own reason for existing.”
― Albert Einstein
I think for any organization, even those that seemingly have done things a certain way for a long time, an important element is to remain curious and open-minded about the potential of new technologies. Healthy skepticism can be a good thing and prevents lunging at every new trend that surfaces. But as evidence grows around the benefits gained from AI/ML tools, exploration of how or where they can fit into a business is a worthwhile exercise. Look for companies and suppliers that have been working in the space for an extended period of time. Who has a track record of using or implementing this type of technology successfully?
Think about the types of problems that would be useful to solve.
If you can develop “organizational curiousness” and a general open-mindedness about the potential that AI/ML can have inside of your company, the second step is to think about where the opportunities might exist. Remember, the business is not there to create work for the technology, the technology is there to serve the needs of the business. Don’t try and figure out how to jam a tech solution into every area. Start with the problems your organization has been traditionally trying to solve since time immemorial. Every company has a functional area that they intuitively know is managed by brute force. Whether it’s through sheer manpower, a heavy-handed process, or moving in a strategic direction based on instinct and a roll of the dice. I think if most organizations stepped back and asked, “Is there a more nuanced, insightful way to approach this problem? If we could glean a fraction of the potential out of a complex, seemingly impenetrable mound of data?” The return would be enormous.
Some of these problems are going to be very specific to your company and will require the expertise of your team. Others will be challenging but not necessarily unique to your business. These are more common issues related to the oft-cited “V’s” in data science circles: volume, velocity, variety and veracity of data. Once you have thought about the types of problems worth solving and where they might fall on the proprietary versus common continuum, the next step is determining who is best suited to solve it.
Build vs. Buy?
This is an area that has been extensively written about and dissected many times over, particularly when it comes to AI. The Boston Consulting Group builds a case around four quadrants, arranged as a function of Value Potential versus Differentiated Data Access. I tend to think in simpler terms: are there vendors out there with specific expertise in the problems I’m trying to solve? Is their experience in a functional area that’s core to their business, or is the problem I’m solving for core to my business?
“Companies can work with AI vendors in many ways, ranging from outsourcing an entire process to buying selected services, seeking help in building in-house solutions or training internal staff. Executives should view these options in light of two questions: 1) How valuable is the process or offering to your future success?; and 2) How strong is your ownership, control, or access to high-quality, unique data, relative to the AI vendor?”
-Boston Consulting Group
As you consider this issue, I think one of the best ways to help ascertain where on the spectrum your problem exists is to open yourself up to conversations with technology vendors. Now I know there is nothing worse than a software salesperson. They are persistent, they ask a lot of questions, they are always looking for access to decision-makers. They do however usually have a considerable amount of insight.
Pick a spot, get a win. Rinse, repeat.
Jump in! But don’t feel like you have to boil the ocean. Tackle a specific problem area – a customer segment, a troubled sales region, a particularly broken process – to prove to yourself and your management the power of data science. In my experience as a customer of AI-based software and now on the vendor side, I’ve often seen small data science projects catch on and spread like wildfire throughout an organization.
Good news travels fast, especially when it pertains to bottom-line profit growth, increased sales and/or removal of unnecessary process overhead.
About the AuthorMore Content by Mick Naughton