Showing posts with label sentiment analysis. Show all posts
Showing posts with label sentiment analysis. Show all posts

Tuesday, March 04, 2014

Case Study: How Dell Converts Social Media Analytics Benefits into Strategic Business Advantages

Transcript of a BriefingsDirect podcast on how social media creates a gold mine of information for businesses of all sizes and how proper analytics and responses build a competitive advantage.

Listen to the podcast. Find it on iTunes. Download the transcript. Sponsor: Dell Software.

Dana Gardner: Hi, this is Dana Gardner, Principal Analyst at Interarbor Solutions, and you’re listening to BriefingsDirect. Today, we present a sponsored podcast discussion on the expanding role and heightening importance of social media data as an essential ingredient for companies to gain strategic business advantage.

Gardner

We’ll examine specifically how Dell has recognized the value of social media for more than improved interactions and brand awareness. Dell has successfully learned from social media how to meaningfully increase business sales and revenue.

According to Dell, the data, digital relationships, and resulting analysis inherent in social media and social networks interactions provide a lasting resource for businesses and their customers. And this resource has a growing and lasting impact on many aspects of business -- from research, to product management, to CRM, to helpdesk, and, yes, to sales.

To learn more about how Dell has been making the most of social media for the long haul, please join me now in welcoming Shree Dandekar, Senior Director of Business Intelligence and Analytics at Dell Software. Welcome, Shree.

Shree Dandekar: Good Morning, Dana, thanks for having me on.

Gardner: We’re glad you’re here. Businesses seem to recognize that social media and social-media marketing are important, but they haven’t very easily connected the dots in how to use social media for actual business results. What do you think has been holding them back from not yet realizing the payoffs?

Dandekar: There is an interesting dynamic happening between enterprises and some of the small businesses. But before I go there, let’s talk about the landscape in terms of the adoption of social media by businesses overall.

Dandekar
It’s not that businesses don’t realize the value of social media. In fact, many businesses are looking at simple, social media listening and monitoring tools to start their journey into social media.

The challenge is that when you make these investments into any kind of a listening or monitoring capability, people tend to stop there. It takes them a while to start collecting all the data on LinkedIn, Facebook, or Twitter. It takes some time for them to make some meaningful sense out of that. That’s where the dynamic comes in when you talk to an enterprise business. They’ve really moved on.

So, there are several stages within a social media journey, the very first one being listening and monitoring, where you start capturing and aggregating data.

From there, you start doing some kind of sentiment analysis. You go into some kind of a social-media engagement, which leads to customer care. Then, you go into questions like social return on investment (ROI) and then, try to bring in business data and mash up that together. This is what’s known as Social CRM.

So, if you say that these are the six stages of social-media maturity model or a social-media lifecycle, some of the enterprise businesses have really matured in the first three or four phases, where they have taken social media all the way to customer care. Where they are struggling now is in implementing technologies where you can derive an actual ROI or business value from this data.

Listening and monitoring

Whereas, if you look at some of the small businesses or even mid-sized companies, they have just started getting into listening and monitoring, and the reason is that there are not many tools out there that appeal to them.

I won’t name any specifically, but you know all the big players in the social media listening space. They tend to be expensive and require a lot of reconfiguration and hands-on training. The adoption of social media in the small-sized business or even mid-sized businesses has been slow because these guys don't want to invest in these types of tools.

By the way, here is another big differentiator. If you look at enterprises, they don't shy away from investing in multiple tools, and Dell is a great example. We have a Radian6 deployment, social-media engagement tools, and our own analytic tools that we build on top of that. We tried each and every tool that's out there because we truly believe that we have to gain meaningful insights from social media, and we won't shy away from experimenting with different tools.

Mid-sized companies don't have the budget or resources to try out different tools. They want a single platform that can do multiple things for them – essentially a self-service-enabled social-media intelligence platform.

If I start with listening, I just want to understand who is talking about me, who my influences are, who are my detractors, what are my competitors talking about, and whether or not I can do a quick sentiment analysis. That's where I want to start.

Gardner:  I was very impressed to learn that Dell has been doing social media since 2006, so going quite a ways back. How important is this to Dell as a company and how important do you think other companies should view this? Is this sort of a one-trick pony, or is there a lasting and expanding value to doing social media interactions analysis?
Dell was built on the value of going direct to consumers and the blog had to communicate and live by those same values.

Dandekar: In addition to leadership from the top, it took a perfect storm to propel us fully into social. In July 2006 when pictures and a report surfaced online out of Osaka, Japan of a Dell laptop spontaneously combusting due to a battery defect (which happened to impact not just Dell, but nearly every laptop manufacturer), it was a viral event of the sort you don’t want. But we posted a blog titled “Flaming Notebook” and included a link to a photo showing our product in flames – which caused some to raise an eyebrow.

I will pause there for a second. How many of you would do that if something similar happened to your business? But Michael Dell made it crystal clear: Dell was built on the value of going direct to consumers and the blog had to communicate and live by those same values.

This is 2006, when the internet and the true value of blogging and everything was just becoming more relevant. That was a turning point in the way we did customer care and the way we engaged with our customers. We realized that people are not only going to call an 800 support number, but are going to be much more vocal about it through sources like social media blogging on Twitter and Facebook.

That's how our journey in social media began and it’s been a multi-year, multi-investment journey. We started looking at simple listening and monitoring. We built a Social Media Command Center. And even before that, we built communities for both our employees and our customers to start interacting with Dell.

Idea Storm

One of the most popular communities that we built was called Idea Storm. This was a community in which we invited our customers to come in and share ideas product improvements they want. This community was formed around 2007. To date, there have been close to 550 different ideas that we got from this community that have been implemented in Dell products.

Similarly, we launched Employee Storm, which was for all the employees at Dell, and the idea was similar. If there are some things in terms of processes or products that can be changed, that was a community for people to come in and share those ideas.

Beyond that, as I said, we built a Social Media Command Center back in 2010. And we also stood up the Social Media and Communities University program. We started training our internal users, our employees, to take on social media.

Dell firmly believes that you need to train employees to make them advocates for your brand instead of shying away and saying, “You know what, I'm scared, because I don't know what this guy is going to be saying about me in the social media sphere.”

Instead, we’re trying to educate them on what is the right channel and how to engage with customers. That's something that Dell has developed over the last six years.
Social media has become a core part of our DNA, and it fits well because of the fact that our DNA has always been built on directly interacting with our customers.

Gardner: Well, that's very interesting. You’ve taken a one-way interaction, made it two-way, and then expanded well beyond that. How far and wide do the benefits of social media go? Are you applying this to help desk, research, new products, service and support, or all the above? Is there any part of Dell that doesn't take advantage from social media?

Dandekar: No, social media has become a core part of our DNA, and it fits well because of the fact that our DNA has always been built on directly interacting with our customers. If a customer is going to use social media as one of their primary communication channels, we really need to embrace that channel and make sure we can communicate and talk to our customers that way.

Gardner: I also was impressed to learn that you are directly making sales through Twitter to the tune of millions of dollars. How does that work? How do other companies learn from your example of not only having many business benefits that are so called soft benefits, but also direct sales? How do you make direct sales through Twitter?

Dandekar: We've invested a lot in this space. We have a big channel through Salesforce.com where we interact with all the leads that come in through Salesforce.

Taking that relationship to the next level, is there a way I can smartly link the Salesforce leads or opportunities to someone's social profile? Is there a way I can make those connections, and how smartly can I develop some sales analytics around that? That way, I can target the right people for the right opportunities.

Creating linkage

That's one step that Dell has taken compared to some of our industry competitors, to be very proactive in making that linkage. It’s not easy. It requires some investment on your part to take that next step. That's also very close to the sixth stage that I talked about, which is social CRM.

You’ve done a good job at making sure you’re taking all the social media data, massaging it, and deriving insight just from that. Now, how can you bring in business data, mash it up with social data, and then create even powerful insights where you can track leads properly or generate opportunities through Twitter, Facebook or any other social media sources?

Gardner: Shree, it seems to me that what you’re doing is not only providing value to Dell, but there is a value to the buyer as well. I think that as a personal consumer and a business consumer I’d like for the people that I am working with in the supply chain or in a procurement activity to know enough about me that they can tailor the services, gain insight into what my needs are, and therefore better serve me. Is there an added-value to the consumer in doing all this well, too?

Dandekar: Absolutely. The power of social media is real-time. Every time you get a product from Dell and tweet about it or say you like it on Facebook, there is a way that I can, in real-time, get back to that customer and say I heard you and thanks for giving us positive or a negative feedback on this. For me to take that and quickly change a product decision or change a process within Dell is the key.
The power of social media is real-time.

There are several examples. One example that comes to mind is the XPS 13 platform that we launched. The project was called “Project Sputnik.” This was an open-source notebook that we deployed on one of our consumer platforms XPS 13.

We heard a lot of developers saying they like Dell, but really wanted a cool, sexy notebook PC with all the right developer tools deployed on that platform. So, we started this project where we identified all the tools that would resonate with developers, packaged them together, and deployed it on the XPS 13 platform.

From the day when we announced the platform launch, we were tracking the social media channels to see if there was any excitement around this product.

The day we launched the product, within the first three or four hours, we started receiving negative feedback about the product. We were shocked and we didn’t know what was going on.

But then, through the analytics that we have developed on top of our social media infrastructure, we were able to pinpoint that one of the product managers had mistakenly priced the notebook higher than that of a Windows notebook. The price should not have been higher than that of a Windows notebook, and that’s why a lot of developers were angry. They thought that we were trying to price it higher than traditional notebooks.

We were able to pinpoint what the issue was and within 24 hours, we were able to go back to our product and branding managers and talk to them about the pricing issue. They changed the pricing on dell.com and we were able to post a blog on Engadget.

Brand metrics

Then, in real time, we were able to monitor the brand metrics around the product. After that, we saw an immediate uptick in product sentiment. So, the ability to monitor product launches in real time and fix issues in real time, related with product launches, is pretty powerful.

One traditional way you would have done that is something called Net Promoter Score (NPS). We use NPS a lot within Dell. The issue with it is that it is survey-based. You have to send out the survey. You collect all the data. You mine through it and then you generate a score.

That entire process takes 90 to 120 days and, by the time you get it, you might have missed out on a lot of sales. If there was a simple tweak, like pricing, that I could have done overnight, I would have missed out on it by two months.

That’s just an example, where if I had waited for NPS to tell me that pricing was wrong, I would have never reacted in real-time and I would have lost my reputation on that particular product.

Gardner: It’s pretty important nowadays to have that short latency between reacting to a market and satisfying a market. We don’t have the luxury of going a year or two. That’s super important these days.
The ability to monitor product launches in real time and fix issues in real time, related with product launches, is pretty powerful.

Shree, as you mentioned, Dell has doing this now for quite a few years. What are they getting for their effort? How extensive is your listening and analysis from social media?

Dandekar:  Just to cite some quick stats, Dell has more than 21 million social connections through fans on Facebook, followers on Twitter, Dell community members, and more across the social web.

We talked about customer care and the engagement centers, and I talked about those six stages of the social media journey. Based on the Social Media Command Center that we have deployed within Dell, we also have a social outreach services team that responds to an average of 3,500 posts a week in 14 languages and we have an over 97 percent resolution rate.

We talked about Idea Storm and I had talked about the number of ideas that have been generated out of that. Again, that’s close to 550 plus ideas to date.

Then, we talked about the Social Media and Communities University. That’s an education program that we have put in place, and to date, we have close to 17,000 plus team members who have completed the social media training certification through that program.

Social-media education

By the way, that’s the same module that we have started deploying through our social media professional services offering, where we’ve gone in and instituted the Social Media and Communities University program for our customers as well.

We have had a high success rate just finding some of the customers that have benefited through our social media professional services team and also deploying Social Media Command Center.

Red Cross is a great example where we have gone and deployed the Social Media Command Center for them to be much more proactive in responding to people during the times of calamities.

Clemson University is another example, where we've gone and deployed a Social Media Command Center for them that’s used for alternate academic research methods and innovative learning environments.

Gardner: Tell me a little bit about what SNAP is. It seems like we are taking some of this ability to peer into markets and go even deeper than just input and data -- we are really getting into what's motivating people and how that works in a complex ecosystem.

Dandekar: SNAP stands for Social Net Advocacy Pulse. This was a product that we developed in-house. As I said, we have been early users of listening and monitoring platforms and we have deployed Social Media Command Centers within Dell.
It takes a long time to get to that ease of use ability for anybody to go in and look at all these social conversations and quickly pinpoint to an issue.

The challenge, as we kept using some of these tools, was that we realized that the sentiment accuracy was really bad. Most of the times when you take a quote and you run it through one of the sentiment analyzers, it pretty much comes back saying it's neutral, when there’s actually a lot of rich context that’s hidden in the quote that was never even looked at.

The other thing was that we were tracking a lot of metrics around graphs and charts and reports, which was important, but we kind of lost the ability to derive actual meaningful insights from that data. We were just getting bogged down by generating these dashboards for senior execs without making a linkage on why something happened and what were some of the key insights that could have been derived from this particular event.

None of these tools are easy to use. Every time I have to generate a report or do something from one of these listening platforms, it requires some amount of training. There is an expectation that the person who is going to do that has been using this tool for some time. It takes a long time to get to that ease of use ability for anybody to go in and look at all these social conversations and quickly pinpoint an issue.

Those are some of the pain points that we realized. We asked, “Is there a way we can change this so we can start deriving meaningful insights? We don’t have to look at each and every quote and say, it's a neutral sentiment. We can actually start deriving some meaningful contact out of these quotes.”

Here is an example. A customer purchased a drive to upgrade a dead drive from a Dell Mini 9 system, which originally came with an 8 GB PCI solid state drive. He took the 16 GB drive and replaced the 8 GB drive that was dead. The BIOS on the system instantly recognized it and booted it just fine. That’s the quote that we got from one of the customer’s feedback.

Distinct clauses

If I had run that quote through one of the regular sentiment analyzing solutions, it would have pretty much said it's neutral, because there was really nothing much that it could get from that it. But if you stop for a second and read through that quote, you realize that, there are a couple of important distinct clauses that can be separated out.

One thing is that he’s talking about a hard drive in the first line. Then, he’s talking about the Dell Mini 9 platform, and then he’s talking about a good experience he had with swapping the hard drive and that the BIOS was able to quickly recognize the drive. That’s a positive sentiment.

Instead of looking at the entire statement and assigning a neutral rating to it, if I can chop it down into meaningful clauses, then I can go back to customer care or my product manager and say, “Out of this, I was able to assign an intensity to the sentiment analysis score.” That makes it even more meaningful to understand what the quote was.

It's not going to be just a neutral or it's not going to be a positive or negative every time you run it through a sentiment analysis engine. That’s just one flavor.

You asked about sentiment gravity. That’s just one step in the right direction, where you take sentiment and assign a degree to it. Is it -2, -5, +5, or +10? The ability to add that extra color is something that we wanted to do on top of our sentiment analysis.
I can really mine that data to understand how I can take that and derive meaningful insights out of that.

Beyond that, what if I could add where the conversation took place. Did it take place on Wall Street Journal or Forbes, versus someone’s personal blog, and then assign it an intensity based on where the conversation happened?

The fourth area that we wanted to add to that was author credibility. Who talked about it? Was it a person who is a named reputed person in that area, or was it an angry off customer who just had a bad experience. Based on that, I can rate and rank it based on author credibility.

The fifth one we added was relevance. When did this event actually happen? If this event happened a year or two back, or even six months back, and someone just wants to cite it as an example, then, I really don’t want to give it that high rating. I might change the sentiment to reflect that it's not that relevant based on today’s conversations.

If I take some of these attributes, sentiment, degree of sentiment, where the conversation happened, who talked about it and when and why did that conversation happen and then convert that into a sentiment score, that’s now a very powerful mechanism for me to calculate sentiment on all these conversations that are happening.

That gives me meaningful insights in terms of context. I can really mine that data to understand how I can take that and derive meaningful insights out of that. That’s what SNAP does, not just score a particular quote by pure sentiment, but add these other flavors on top of that to make it much more meaningful.

Make it usable

Gardner: So the information is out there, people are telling you what they want, they’re interacting in such a way that you can gain very valuable insights, if you take the proper steps to get that information and make it usable in your own organization.

Dandekar: That’s right.

Gardner: I’m going to put you on this spot here, Shree, because you mentioned earlier that small to medium-sized businesses are looking for a one-stop shop to do this.

You’ve already demonstrated what Dell is doing internally. Have you considered productizing this and perhaps creating a service for the smaller companies that want to do this sort of social analysis and help them along the way?
We also want to make sure we’re bringing tools to market to service those mid-market companies as well.

Dandekar: We’re still working through those details and figuring out as we always do the best ways to bring solutions to market, but for us, mid-market is our forte. That’s an area where Dell has really excelled. For us to be in the forefront of enterprise social media is great, but we also want to make sure we’re bringing tools to market to service those mid-market companies as well.

By the way, we have stood up several solutions for our customers. One of them is the Social Media Command Center. We’ve also stood up social media professional services and we offer consulting services even to small- and mid-sized companies on how to mature in a social media maturity cycle. We are also looking at bringing SNAP to market. But if you’re talking about specific software solutions, that’s an area that we’re certainly looking into, and I would just say, “Stay tuned.”

Gardner: We’ll certainly look for more information along those lines. It's something that makes a lot of sense to me. Looking to the future, how will social become even more impactful?

People are increasing the types of activities they do on their mobile devices and that includes work and home or personal use and a combination of them, simultaneous perhaps. They look to more cloud models for how they access services, even hybrid clouds. It’s stretching across your company’s on-premises activities and more public cloud or managed service provider hosted services.

We expect more machine-to-machine data and activities to become relevant. Social becomes really more of a fire hose of data from devices, location, cloud, and an ever-broadening variety of devices. Maybe the word social is outdated. Maybe we’re just talking about data in general?

How do you see the future shaping up, and how do we consider managing the scale of what we should expect as this fire hose grows in size and in importance?

Embarking on the journey

Dandekar: This is a great question and I like the way you went on to say that we shouldn’t worry about the word social. We should worry about the plethora of sources that are generating data. It can be Facebook, LinkedIn, or a machine sensor, and this fits into the bigger picture of what's going to be your business analytics strategy going forward.

Since we’re talking about this in the context of social, a lot of companies that we talk to -- it can be an enterprise-size company or a mid-market-size company -- most of the time, what we end up seeing is that people want to do social media analytics or they want to invest in the social media space. Some of their competitors are doing that, and they really don’t know what to expect when they embark on this journey.

A lot of companies have already gone through that transformation, but many companies are still stuck in asking, “Why do I need to adopt social media data as part of my enterprise data management architecture?”

Once you cross that chasm, that’s where you actually start getting into some meaningful data analytics. It's going to take a couple of years for most of the businesses to realize that and start making their investments in the right direction.
It's going to take a couple of years for most of the businesses to realize that and start making their investments in the right direction.

But coming back to your question on what's the bigger picture, I think it’s business analytics. The moment you bring in social media data, device data, the logs, sources like Salesforce, NetSuite -- all this data together now presents the unified picture using all the datasets that were out there.

And these datasets can also be datasets like something from Dun and Bradstreet, which has a bunch of data on leads or sales, mixing that data with something like Salesforce data and then bringing in social media data. If I can take those three datasets and convert that into a powerful sales analytics dashboard, I think that’s the nirvana of business analytics. We’re not there yet, but I do feel a lot of industry momentum going in that direction.

Gardner: I agree that that’s the end game and we’re only a few innings in, but it's very impressive and an exciting time to be looking out to that.

I'm afraid we’ll have to leave it there. You've been listening to a sponsored BriefingsDirect Podcast discussion on how social media data has emerged as an essential ingredient and how companies gain strategic business advantage.

And we've seen how Dell has been making the most of social media for the long haul by positively impacting many aspects of it's business. We've also heard that getting a handle on managing the flow of social media puts organizations -- whether they are small, medium or large enterprises -- in a position to better manage all kinds of data as that data becomes available.

So a big thank you to our guest, Shree Dandekar, Senior Director of Business Intelligence and Analytics at Dell Software.

Dandekar: Thanks, Dana.

Gardner: And of course a big thank you, too, to our audience for joining this insightful discussion. This is Dana Gardner, Principal Analyst at Interarbor Solutions. Thanks again, for listening and don’t forget to come back next time.

Listen to the podcast. Find it on iTunes. Download the transcript. Sponsor: Dell Software.

Transcript of a BriefingsDirect podcast on how social media creates a gold mine of information for businesses of all sizes and how proper analytics and responses build a competitive advantage. Copyright Interarbor Solutions, LLC, 2005-2014. All rights reserved.

You may also be interested in: