January 5, 2017 by Paul Hausser, Envisn, Inc.
With Watson having become the point of the IBM marketing spear an obvious question being raised by many Cognos administrators is how one relates to the other. Our focus here is on that audience primarily because there is a perception of an overlap between the two, and if there is, what is the extent of the overlap? Here we’ll try to address both questions.
Watson Analytics became a real product for IBM following its success with the game show Jeopardy. IBM discovered it had something real on its hands and the question was how to take advantage of it. Watson was really good at taking an amorphous blob of data and finding jewels in it. And with the explosion of data being collected over the past five plus years many companies were asking the question of what to do with it all. Big data became a reality.
Preparation Meets Opportunity
Watson’s commercial challenge was in creating the minimal structure needed to get it in broad use so companies and institutions could at least try it. Plus, they created some dialog tools to make it interactive along with the ability for machine learning so it can develop expertise in the users own domain. They did this while also finding a number of successful use cases that cross a broad range of major industries. Finally, they created a pricing structure that makes it virtually risk free to try it out.
Where Watson works best:
- Exploration of data focused on discovery and insight - Most often Watson Analytics is looking at a large amount of data and trying to discover how the data are related and what can be gleaned from it. Does it provide any new insight into what we already know about our business or this new area of knowledge?
- Trying to answer “What’s in here that’s useful?” – It has the ability to look at data multiple ways in order to assess its relevancy and it does this with great efficiency, because if it’s not useful you want to know that right away.
- Good for looking at data with no boundaries – One thing that Watson is really good at is working with data that has no edges. Its algorithms and data tools are designed to look for relevancy and associations and then let the user assess value. Not every use case will identify value but where it does can often be very powerful. Just read some of the use cases.
- Primary focus is on using systems of record data – This is what companies and organizations use to manage the enterprise on a daily basis. It represents the work the organization and its people do and most of the data comes from ERP or related systems.
- Both the data and its use tend to be structured – Most of the data used by Cognos comes from a data warehouse where transaction data has already been processed and organized for analysis and reporting purposes. From there it’s typically segmented or put into packages defined by a particular area of focus.
- Strong focus on trends, exceptions, deviations from plan or the past – Analysis and exception reporting are the predominant areas of use and this also drives data based decision making.
Cognos analytics needs to be able to proof the numbers which means validating data sources and the quality of the data when necessary. Why? Because much of the output is used for financial reporting and must meet certain standards. Watson analytics for the most part has a much lower threshold for quality since its focus is on discovery, insight and usefulness.
Some things you need to know:
- While we make some distinctions above between Watson Analytics and Cognos Analytics these are not hard definitions. For example, the structured data that drives Cognos can be easily used by Watson for discovering new insights into existing operations. In fact, many companies have done this with success and learned how to improve their operations and better focus on the customer.
- Cognos administrators are often concerned with the metadata that drives Cognos. This is a key part of validating the output from Cognos. An example of this is data lineage or the provenance of the data being used. While this is easily done with a tool like NetVisn or something similar, no such capability exists currently with Watson Analytics.
- Companies using Watson often expand the structured data that they use with Cognos to cover areas first identified using Watson. While it’s not always clear where this will occur it can be an inevitable result.
- Learnings from Watson will continue to have important spillover effects on Cognos Analytics. Just as intent-driven modeling has enabled new users to pull together data without knowing details of data sources, there will new areas where Cognos continues to expand its value as a result of Watson.
- IBM has made Watson analytics available with a pricing model that enables nearly everyone to test in out at whatever level is a fit for their organization. Users are able to do this without making a major financial commitment to something they might never be able to justify. If they perceive a clear return after tying it out, they can expand their usage to whatever best fits their needs.
- IBM sees both Watson Analytics and Cognos Analytics as having legitimate market spaces. And while there is some overlap it’s likely that they will both continue to have complimentary market spaces.
© Envisn, Inc. – 2017 - All rights reserved.