Q&A: The Untapped Potential Inside Social Media, Analytics (Part 2 of 2)
Using predictive analytics on a combination of social media and data from in-house systems can be a powerful combination. We offer insights into the potential and how to make the business case for using social media analytics.
- By Linda L. Briggs
- October 21, 2014
Analyzing a blend of social media and data stored in-house can yield rich insights, explains Information Builders' Dan Grady in this interview, the second of two parts. [Editor's note: Read part 1 here.]
Grady, the social media analytics and enterprise search sales manager at Information Builders, has worked on social media analytics, search-based business intelligence, mobile applications, predictive analytics, and dashboard design in his 15 years at the company. He blogs about social media, business intelligence and more at http://www.informationbuilders.com/blogger/Dan%20Grady#sthash.UesjdylA.dpuf.
Grady recently spoke, along with Fern Halper, TDWI's research director of advanced analytics, at a TDWI Webinar on "Social Media Analytics – Getting Beyond Tracking the Buzz."
In this interview with BI This Week, we continue our discussion of the untapped potential of predictive analytics and social media.
BI This Week: When you talk with clients, how do you make the business case for social media analytics? Where's the payoff?
Dan Grady: That question comes up quite a bit. The answer I find most effective is this. ... When we talk about social media now, we're talking about one of the voices of your customers. It's a feedback channel. When customers are out there shopping, they go online. They investigate, not only on your company's website but with their friends -- what to buy, what not to buy, and experiences with the products. They're listening at your Facebook page to see what people are saying about your brand and products. It's a channel that customers use to communicate with their suppliers of goods and services.
If you're trying to make the business case, you might say, "Hey, would you eliminate feedback and analysis from your call center? Would you eliminate feedback you're getting through customer service? Would you eliminate feedback through channels we're used to? "
Social media now is just another channel for capturing feedback, and it's growing as we speak. It's growing to become one of the primary channels you're capturing feedback from and communicating with your customer base from. It's just another voice -- and a rapidly growing voice.
And it's a very public voice.
Yes. That's one of the reasons I tell people that you can't ignore social media. When someone calls your call center or sends you an e-mail, it's between you and them. When someone posts something negative on Facebook, it stays there forever.
You mentioned earlier that some companies are collecting and analyzing social media, but they tend to be very fragmented in their dealings. They collect feedback but it's not often melded together effectively before analysis. How can that approach be remedied?
Yes, [we see larger companies in particular] using social media, but as you said, there's a very siloed approach to it. They're usually not blending the channels of feedback together.
We're in a good position at Information Builders because we sit on top of all data, not just social media data. Regarding that, one of the things we're focused on here is the concept of cross-channel feedback analysis. ... Because we sit on top of all of the data that organizations are collecting -- whether it be social data, call center data, or whatever -- we can help them bring all that data together across all of the different channels and start analyzing it very quickly. ...
Forrester put out a report recently on agile BI [Forrester Wave for Agile Business Intelligence Platforms] that I blogged about. They point out that data warehouses are somewhat restricted in trying to be agile. ... Cross-channel feedback is one of those areas in which the data warehouse potentially could become restrictive because the feedback across the different channels is stored in different data repositories, and if you tried to come up with a single model to support the analysis, it would be very difficult.
In fact, one of the biggest challenges many organizations are facing today is the increasing numbers of channels with valuable customer feedback, whether it be social media, online surveys, calls to the call center, e-mail to a help desk, online chat sessions, or something else. Along with more channels, they're also dealing with both structured and unstructured data.
To create and manage a data model to handle this constantly evolving diversity of data would be quite a challenge, and I would imagine it is one of the reasons Forrester calls out "restrictive data models" as a barrier to agility. If you use a search index as the backbone of your application, you will eliminate that restriction. With WebFOCUS Magnify, we're able to leverage all of the data access and data management capabilities of the platform, to streamline the process of loading both structured and unstructured data into a search-based application. What many would think would be a daunting task, even for a seasoned IT professional, is now something a business professional can accomplish.
What is the value of extracting data from "outliers," because the average consumer typically neither complains about nor praises a product? After all, it's the average consumer that companies want to recruit or retain.
We've talked about that reason earlier in this interview. Social media is there forever. ... When people are evaluating your company and your brand, they're going to social media. They're going to your Facebook page to see what other people are saying. ... As a customer, when you're going to buy something, honestly, you're mostly interested in the bad stuff: "Hey, what's wrong with this product?" More and more, if I go to a site about a product, I expect some negativity, but also some good outweighing the bad -- that's helpful from a consumer standpoint.
As a company, you need to understand the outliers. How you handle the negativity is also important from a brand perception standpoint. When someone posts something negative, how you handle that and how you respond to it affects your brand. You can't ignore the outliers. You need to handle them as appropriately as possible.
What about issues with someone planting false data or posting negative information? After all, as you've said, everyone including your competitors has access to social media.
I haven't seen that happen, but that's part of the reason that you analyze the data. If you analyze over time and consistently see the same name posting negative comments, that gives you a heads-up. How you respond, again, is important. I haven't seen it, but you analyze the data so that you're aware if that is happening.
How accurate can sentiment analysis really be?
It's a work in progress -- not only in the overall industry but specific to your business. It's never going to be 100 percent accurate. Even if you and I were having a conversation in the same room, how might you interpret what I said -- positively or negatively? To expect a computer to do so 100 percent of the time just isn't realistic. That's why they have taxonomies that you train to understand the things you want to weigh as negative in your industry and your world. The models mature over time, but they will never be 100 percent accurate -- between 60 and 80 percent accuracy would be considered acceptable.
The example I use quite often is if you have 10 posts on your Facebook page, or 10 tweets, then you don't need analysis software -- you can just read them. What if you have 10,000? You need some direction. Even if it's just 60 or 70 percent accurate, you need to be able to say how many of these are negative. Is that a problem? Social media analytics is there to help you realize when there are problems and when there are opportunities.
What are some suggestions for a company wanting to get started using social media and analytics more effectively?
Fern made a very good point in her presentation -- start with the proof of value. Here at Information Builders it's one of the things we do to help customers get started. We have a program called a challenge -- we say, "Take the Social Media Analytics Challenge." They let us know that they're interested in looking at social media analytics, and we'll go out and pull down their data, score it with sentiment analysis, do some of other things we've talked about today, and we'll come back and present the dashboard of their actual data.
The proof of value piece is a key. You want to see if you have enough volume to make it worthwhile. Depending on how much traffic and how much volume you have from a social media perspective, that will dictate the amount of time and effort and money you invest in a solution.