These notes are from a September 2nd webinar presented by Alexander Linden, Research Director at Gartner. The event is available on demand and can be viewed here. You don’t have to be a hard core analyst to benefit from this event – the take aways were interesting and applicable to procurement even though it wasn’t a procurement-specific event.
Among my favorite ideas were:
The concept that real time analytics doesn’t mean faster analytics, it means providing a constant stream of fresh data to the people and systems responsible for analytics so that there is no time lag between the event, the data being handed over, and the actions or decisions that result. This gets more difficult as data is less and less likely to be in a tabular (or ‘rectangular’) format and more likely to include unstructured data from video, social media, and audio.
The use of ‘B2B propensity to buy’ as a use case for advanced analytics. In other words, sales and marketing departments are using data to figure out if their business development activities at a company – yes, this includes companies with advanced, strategic procurement teams – is likely to result in a deal and what (if anything) they can do to turn the tide in their favor. If you don’t think you’re being analyzed by your suppliers – think again. With the cost of sales representing a significant expense for suppliers, they would be crazy not to take advantage of any improvements analytics can offer.
Machine learning as a ‘hot trend.’ I’ve covered robotic process automation (RPA) a number of times, and that addresses the growing role of technology in outsourcing. This serves as a timely reminder that the same kind of ability to learn and advance in artificial intelligence can be applied to analytical programs as well. And just as there are certain processes that lend themselves to RPA, some analytics are better suited to machines than humans – when human logic, attention, or judgment is likely to fail or be unreliable.
The best use case story by far from the event was the back story on your FICO (or credit) score. Fair, Isaac, and Company (the source of the F-I-Co acronym) successfully developed a model that allowed lenders to determine a person’s likelihood to repay a loan with a few easily available pieces of information: whether or not they had a job, whether the job was white collar or blue collar, and where they lived. There were other variables to be sure, and more have been added since their original models, but those were the big ones. And that was all based on analytics they were doing in the 1950’s!