A first look at IBM’s Watson Analytics

An analytical tool for the business user. It isn’t a new idea, more a nut that many of us have been trying to crack for years. Today, of course, this problem is more pressing than ever. There are more questions that can be answered with analytics, more data to analyse and fewer trained analysts to go round. And, as IBM rightly point out, analysts (and other data specialists) can be a bottleneck in the process.

Where I’ve seen it work best in the past is generally when a software app, rather than a tool (like IBM/SPSS or SAS), has been designed to a sufficient level of specificity for a particular use. Examples of such apps include:

  • Online GIS mapping for local planners that run sophisticated sustainability models under the hood.
  • Statistical testing software for biochemists where the statisticians wrote the embedded statistical decisions such as should this be a parametric or non-parametric test.
  • Push-button apps that run daily sales forecasts and stock plans for store managers.

And so on. With a very specific application you can constrain the problem space and pin down the functionality including the data structure. Watson Analytics is much more ambitious than that. It sets out to be more things to more business users. It does look to address various domains: financial services, marketing and so on. But it is planning to attack all of them at the same time.

We know that data is usually the main issue and Watson Analytics promises to make a broad range of data sourcing management with embedded information services. This is the first value proposition with its tag line “Get better data”.

The other 3 key functional areas – and propositions – are

  • Predictive Analytics – “Think ahead”
  • Automated Intelligence – “Understand your business”. More of the “Watson” bit (we assume)
  • Visualisation – “Tell a story”

Can IBM really deliver all that to the line of business user? It does feel like this is an important product development for them and that means a lot of brainpower and cash are probably being invested in it. Adding the prestigious IBM Watson brand to it also means they won’t want it to fail.

I am also curious about how closely linked it is to the broader IBM Watson project and in particular to the cognitive learning methodology that famously won a special episode of Jeopardy in the US. Our sense is that Watson is becoming an umbrella term for a set of tools and technologies that embrace the idea of expert learning in software. Watson Analytics patently has that but it’s not yet clear how much of it there is versus more prescribed (but doubtless sensible and helpful) hard-coded rules.

Did I mention that it is in the cloud? Well of course it is! All in all I can’t wait to work with the beta.

This post was first published on the Analytical People blog.

Scroll to Top