When a firm has finally decided that gathering and analyzing data is a crucial part of business decision-making, once they’ve exited “Excel Hell” and have an enterprise data warehouse, once a culture of data use is in place and everyone has access to the right information at the right time, an age-old question usually arises:
You see, it’s one thing to know where the business is and how it got there, and quite a few organizations have reached that point. It’s quite another thing to know what is coming (or could be if the right actions are taken).
The Diminishing Competitive Advantage
A few months back I wrote that some level of analytics will eventually become the table stakes for being in business, much like IT has over the past three decades. The rate at which this will happen could surprise even those of us who claim to see it coming, since the barrier to entry continues to plummet; there are now more Business Intelligence solutions than I care to count, some of them Open Source, and as a result the cost of implementation has become a fraction of what it used to be.
Gaining visibility into the inner workings of an organization, whether in sales, operations or another area, is usually an enlightening experience. Long-held misconceptions evaporate (but not before a “Well, the data must be wrong” phase), and problems that have festered below the surface for years can finally be brought to light and dealt with. Efficiency skyrockets, costs plummet, and the bottom line benefits. Early adopters view analytics as a competitive advantage but as their competitors catch on that advantage diminishes. This should all sound familiar, as product development follows a similar cycle.
I’d Like a Little Secret Sauce on Mine, Please
Whether a firm is using Tableau, Pentaho, Qlik, or some other product, Business Intelligence solutions tend to follow certain well-established best practices. “Garbage in, garbage out” is as true as it’s always been, and those who don’t adopt some level of Data Governance risk reducing the return on their BI investment. The effectiveness of dashboards is decided by the metrics the organization chooses to measure, and how management responds to that data. These truisms are well-known by now, so deployments follow a similar pattern of best practices.
What is not standardized, at least not yet, is the analysis of data through Machine Learning. While there are certain use cases of Machine Learning that are well-documented and somewhat consistent across practitioners (fraud analysis in insurance comes to mind, as do recommender systems), firms are still finding unique uses for the technology that undeniably provide a competitive advantage. UPS was the first to figure out the benefits of minimizing left turns, Cisco figured out that customers on the verge of a major ‘refresh” purchase tend to hit cisco.com many times more often than they usually do, and so on. Most of these stories are usually accompanied by an impressive dollar amount in savings or additional revenue. Secret sauce, indeed.
Special Orders Don’t Upset Us
The trick to extracting value from Machine Learning is to look beyond the obvious. Many companies dipping their toes in the ML waters want to forecast sales, and there is an obvious benefit to improving the accuracy of those forecasts (it’s a hell of a lot better than “What did she do last year? OK, add ten percent.”). It also helps to start with a small win like this to show value and build competency–boiling the ocean is nearly always a bad idea. To experience a radical return, however, it makes sense to choose an area that the firm considers within its core competency and figure out a way to revolutionize that process, which is exactly what UPS did when they revamped their routes to use the minimum number of left turns. It’s also what a number of disruptive startups are doing to long-established industries like hospitality and transportation.
So, to recap: while some level of analytics will become common to nearly every firm, developing a competency in Machine Learning (or partnering with an organization that already has it) will still provide a competitive advantage for the foreseeable future. In the end, it will be those who find creative ways to revolutionize core competencies using ML who will benefit the most.