Main | Is market research breathing its last? Well, the smart phone isn't helping. »

February 17, 2012

Data - structured or unstructured - means business

Posted by Akash Bhatia (View Profile | View All Posts) at 11:33 AM

This post is co-authored with Chantrelle Nielsen.


It's estimated that last year, the world created four exabytes of data - that's four quintillion bytes, or more than the combined knowledge base of the last 5,000 years.  I don't know how much an exabyte is. But it seems like a lot of information to me.  So what do we make of it?  Here's one story:

Frank Eliason, a lone customer service manager for Comcast, was sitting in his living room fielding customer service calls on his headset when he saw an article about a new "status messaging" service. He thought it might be another way to reach the customers who called him daily with service problems. So Frank logged into the service, registered under the handle @comcastcares, and began to make history. Frank turned around Comcast's image almost overnight by monitoring Twitter behind the handle "@comcastcares," and responding to complaints on a timely and personal basis. Although it was labor-intensive for Frank, Twitter users were vocal about his quick response to their issues and they sang praises of Comcast. But even more than being a new vehicle to reach and serve customers, Twitter data taught Comcast what kinds of problems its customers were having, so it could tweak its products and eliminate the problems. It also made Frank into something of a social media star. Frank was a pioneer in a new form of business - social business - wherein customers expect to talk to brands, and expect to be answered, often in near real-time. He got information about Comcast customers that was not available before the arrival of social media.

Now, imagine if you could get that kind of insights into your prospects, your customers, your competitors, your products and what customers like and dislike about your product without all that manual monitoring. People today are leaving a vast "digital data trail" on social media sites, in customer call center issues logs, and in emails. But most companies don't know how to tap this data and use it to improve their business. Although it's impossible to know for sure, some estimates say 80-85 percent of the data handled by modern businesses is unstructured.  We call it "unstructured" because we gathered it, sort of, by accident with no real idea of what we are going to use it for.  Even as the volume of structured data explodes, unstructured data grows even more quickly because of the shift to social business. Customers never talked back to the brand before. Now they can, and they do.

The "structure" in structured data comes from an abstract model that is created to describe the data. The model gives names to each field in a database and defines the relationships between the fields. Take, for example, an automobile manufacturer. From the structured data, we can learn that "2010 Model X" is a car model, "Navigation System" is a component, and "defective" is a condition: we can define the relationships between those three fields so we know that the navigation system in a particular 2010 Model X is defective.  Unstructured data, on the other hand, is not stored in a relational database, or at least not one in which the data model is relevant to the meaning of the data. For example, someone posts to an automotive forum on the Internet: The GPS in my 2010 Model X is constantly getting me lost! It stops receiving any signal for minutes at a time, and when it finally decides to tell me where I am, half the time it's wrong. The same basic information is contained in this paragraph. But the database that stores the text cannot tell us so. This database is organized around different pieces of text, dates, times and the users who post to the forum. Examples of unstructured data also include e-mails and notes from call centers. The best way to get a good view of your business is to look at a mixture of unstructured and structured data.

Let's return to the example of the defective GPS to demonstrate how both kinds of data are important. Assuming that the defect is present across multiple vehicles, the automaker will find out eventually from service center records (structured data), that there is a flaw in the component. It will then launch an investigation into the root cause, craft an appropriate, cost-effective process for handling the defective part, and then communicate and train its service centers on this process. The sooner this process can be developed, the lower the overall cost of the defect will be to the automaker. Hence the earliest warning benefits the automaker the most - especially if this warning comes before customers even start showing up to have the defect repaired (unstructured data). For certain vehicle components, it may be more cost-effective to simply replace a defective component than try to diagnose the problem and fix it. Hence, the service center records (structured data) may contain very little data about the symptoms of breakages, while the online forum postings (unstructured data) may contain a lot of detail - and this information would be necessary to determine the root cause of the problem.

Comcast has Frank Eliason to thank for understanding the value of unstructured data.  Do you have someone like him in your team?


the social business collaboration certainly has lot of value and benefits to the business. the challenge is to analyze and utilize the value of the big data that is getting generated from various sources and then injecting that in improving the business process and customer satisfaction etc. Enterprise has to adopt the ever growing need of utilizing the big data and related analytic out put to feed in it to their existing SOA and BPM models to make it more cost effective and add value to their business...

Post a comment

(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)

Please key in the two words you see in the box to validate your identity as an authentic user and reduce spam.

Search InfyTalk

+1 and Like InfyTalk

Subscribe to InfyTalk feed

InfyTalk VBlogs: Watch Now

Infosys on Twitter