Quick Thoughts on Watson Analytics


IBM has had a ton of announcements around Watson, their cognitive computing platform. A couple of weeks ago it was Watson Discovery and earlier this week it was Watson Analytics. Still a Beta program, Watson Analytics is a cloud based cognitive computing analytics package with a user friendly interface. It is an important step in the evolution of cognitive computing platforms for the following reasons:

  1. It makes it easy for anyone to use cognitive computing. Truth be told, a lot of cognitive computing have been science experiments. They relied on teams of Ph.D-type data scientists to build models, extract and import data, and build a presentation layer for mere mortals. Watson Analytics allows people with expertise in something other than statistics and data models (in other words, real domain experts) to take advantage of the power of a cognitive computing platform. This is step one in the democratization of these platforms which might actually lead to mainstream adoption.
  2. Democratization also means lower cost. Let’s face it, the cost of all those Ph.Ds is not trivial. If a business user can do much of the work then it will cost less to leverage cognitive computing for business purposes. If you look at the use cases most vendors cite (and I have no reason to disbelieve them) they are extremely science heavy such as curing cancer (no joke!) or exploring for natural gas.
  3. Which brings up the use case question. Can cognitive computing have use cases beyond esoteric, science heavy applications? Watson Analytics looks to bring cognitive computing to the masses of business users, greatly expanding the use cases. The featured demo was, in fact, a marketing application. With Watson Analytics, IBM is attempting to make cognitive computing relevant to more business users.
  4. Danger Will Robinson! Ultimately, Watson Analytics represents a competitive danger for other vendors and, hence, an opportunity for businesses. It’s not big data analytics companies that should be worried though. The majority of analytics needs is just crunching big volumes of data in known models. The bigger problems will be for data integration vendors, especially those that integrate structured/transactional and unstructured data. Tools that help walk you through the data to build a static model will become obsolete or have limited uses cases when Watson Analytics can build a dynamic model for you. This is not a “today” problem for data integration vendors but one looming on the horizon.

The one thing I dislike about Watson Analytics is that it’s not at general availability yet. IBM has a tendency to take it’s time getting products to market. Markets are like dogs – once you show them the treat you can’t hold it back too long. Otherwise, they lose interest, find a “treat” of their own, or simply bite your hand. I’ll admit, though, that Watson has moved at something nearing light speed for a company like IBM. Here’s to hoping that the agility of the Watson Group can get Watson Analytics out of Beta soon.