The power of big data technology is being successfully applied to understanding such complex unknowns as consumer sentiment and even intent. And that understanding then vastly improves how retailers and myriad service providers manage their users' experiences—increasingly in real time.
Fortunately, today's consumers are quite willing to share their intents and sentiments via social media, if you can gather and process the information. Hence the rapidly developing field of social customer relationship management, or Social CRM.
Part of the equation for making Social CRM effective comes from properly capturing the natural language knowledge delivered through the many social channels available to users. But even that is but a first step to being able to gain ever-deeper analysis, and rapidly and securely making those insights available where they pay off best.
And so the next BriefingsDirect thought leader discussion brings together customer analytics services provider Attensity, with its natural-language processing (NLP) technology, and HP Vertica, with big data analytics capabilities, to explain how to effectively listen to the social web and rapidly gain valuable insights and actionable intelligence.
Our guests are Howard Lau, Chairman and CEO of Attensity, and Chris Selland, Vice President of Marketing and Business Development at HP Vertica. The discussion is moderated by me, Dana Gardner, Principal Analyst at Interarbor Solutions.
Here are some excerpts:
Gardner: Sellers and marketers worldwide have always wanted to know what their customers are anticipating or what they want next. I guess we could go back hundreds of years with these questions.
But as someone said recently, it seems that the ability to know what customers want and how to respond to them rapidly has changed more in the last 5 years than in the past 500. Do you agree with that? And why is that the case? What’s so new and different?
Lau: What has happened and emerged in the past 10 years or so, especially in the world of Twitter—Twitter has been around since 2006—is that consumers are finding a voice to express their opinions about companies, products and brands. They can express their voice immediately through social channels.
That’s one of the new emerging things where, not only are they finding their voice online, but they’re also realizing that they’re able to amplify that voice by connecting with their friends and their followers.
Gardner: Why is that making such a big difference in how we know what customers want? I understand that the social part is new and innovative, but how is this changing marketing?
Lau: The way things have happened before is that companies, as they engage with consumers, controlled the conversation. Whether you fill in an online form or you call an 800 number for customer service or purchase, you’re greeted initially with an automated prompt, and the whole prompt system navigates your engagement.
What makes Social CRM so unique and empowering for consumers is that, for the first time, it’s transferring the control and ownership of the conversation to the consumer, the customer. What that means is that the customer now controls what they want to talk about, where they want to talk about it, and what channel they want to use to communicate their needs or issues.
They don’t want to do it in a predefined form, where you check off boxes or answer specific prompts. They want to express their interests more organically and use the company’s branded channels on Facebook and Twitter and non-branded channels on industry forums and communities. That’s what’s key about Social CRM and that’s what’s so unique about this new generation of products to analyze the social web.
Gardner: Let’s go to Chris Selland. Chris, HP Vertica is dealing with a lot of organizations that are trying to do new and innovative things with marketing. Do you also agree that marketing and what we can do have shifted just dramatically in the last five years? Has it really changed the game?.
Selland: There’s been a very dramatic shift in the last five years in marketing. That’s driven, not exclusively, but certainly heavily, by what’s been going on in the social-media world—Twitter and other channels, Facebook, LinkedIn, and so forth.
It has had two impacts. First, it has amplified the voice of the customer. I always remember that commercial about I will tell two friends and she will tell two friends, and so on. Customer voice has always had an impact, but the impact of customer voice these days is dramatically amplified by social media.
The other thing that’s really changed the game entirely is that now organizations that are seeking to understand their customers can no longer exclusively rely on internal data, and by internal data I mean things like customer relationship management (CRM).
In the past, when I, as a marketer, or any customer-facing exec running support or something else, wanted to understand my customer relationships, as long as we have had computers and applications had been able to look at something like my CRM system to see when my customer called the call center or when they bought something. Or I can view my transaction logs with them.
But what I haven't been able to look at and analyze is what they are doing when they’re not interacting with me, when they are interacting with the world, or when my customer is tweeting or on Facebook. Obviously, there is a very rich vein of data there. There is also a lot of noise to screen through, but if you do it right, there is potentially a very rich vein of data to help enhance relationships.
As I said, companies can choose to ignore that, but generally that would be strategically disadvantageous to do. Most companies recognize that there's a tremendous amount of data out there that doesn’t belong to me and that’s not necessarily all about me, but I can certainly use it to understand my present and future customers better.
If you interview a typical consumer, when are you more truthful, when you are interacting directly with the company or when you are actually tweeting or making recommendations to your friends or liking something on Facebook, a lot of the real information is outside of the walls of traditional IT. That’s what’s really changed things dramatically as well.
Gardner: Of course, that’s also provided quite a challenge when the information is in the form of sentiment or intent that we see through social interactions. It's more difficult to attain that and assess it.
Let’s go back to Howard. What are some of the challenges when it comes to getting information, maybe through NLP in order to extend it into this analysis capability?
Lau: When people go online in a social realm, they don’t think about their intent. They just express themselves. So the challenge is letting people communicate the way they choose to communicate and then try to figure out and infer what is their intent and their sentiment.
Trying to determine that is what we do using NLP in an effort to understand what the chatter is about and what the sentiment is about that chatter.
Gardner: In doing so, have you developed limits in terms of what you can do with the technology? It seems like this is a fairly a vast amount of information?
Lau: It's vast, and it's also very domain specific. There’s different terminology based on the domain. For example, in the hospitality and travel industry, when you use the word “service,” service means the service you are getting from the hotel or from the airline.
But when you use word “service” in the telecommunications space, that means something totally different. It means, your service plan, how many minutes you have, do you have text, and so forth.
So when you get down to what people are talking about, you have to understand from which domain they’re talking, infer their meaning and understand their sentiments.
Gardner: So there is a difficult issue in terms of language issues and then there are also technology issues around scale and depth, but let’s stick to the ones about NLP. What is it that Attensity does in order to solve that problem?
Lau: First thing is that we ingest a tremendous amount of data. Most of it is social, but we also ingest company’s internal emails, customer notes, employee notes, and online surveys.
Then, we analyze it and annotate it. Part of the annotation is trying to explain the meaning of a sentence or a sentence fragment. The way we do annotations is driven by our proprietary NLP technology.
One of the first things we do is figure out who is this person and what he’s talking about. We’re trying to find the right industry domain that they are talking about and then distill that into the actual meaning—the intent, as well as the sentiment.
Gardner: Howard, tell me a little bit more about how your relationship with HP has evolved. You have been working with Vertica for a while. Tell us a little bit about why Vertica was of interest to you as you’re trying to accomplish your goals with NLP.
Lau: With the annotations, we generate a lot of intelligence, a lot of metadata. Prior to our relationship with HP, we basically serviced the online surveys and certain internal notes and customer notes for corporations. As we embraced social, we had an explosion of content and annotations.
For us, our relationship with HP was indispensable. HAVEn is not just a product; it's a platform. And it's a platform that scales well, not just handling the process of injecting large amounts of data, but also creating stores, a large store for us, as well as customer stores for each of our clients.
There’s absolutely no way we could have scaled our solution to address the continuing growth of the social realm without this relationship and partnership we have with HP and on the HAVEn platform.
Gardner: Just to be clear, HAVEn, of course, includes quite a few things. Maybe you could just help us understand which elements of HAVEn you’re using and which ones are the most beneficial to you?
Lau: First, it's Vertica. We use Vertica for every customer we have for analytical tools. Vertica sits behind that. Then, for managing the whole ingestion and the storage of the documents that we get from the social space, we use Hadoop and HBase from Hadoop. That’s how we embraced the HAVEn platform.
Gardner: Chris Selland, what is it about the Attensity use case that you think demonstrates some unique characteristics of Vertica and perhaps even more elements of HAVEn?
Selland: First of all, it demonstrates the complementary nature of Vertica and Hadoop. The Vertica platform has been built to do very high-performance analytics on very large volumes of data. That’s really what we’re all about.
Obviously, Hadoop is also built to scale for very large volumes of data, and so we have bidirectional integration, actually huge integration and increasing convergence with Hadoop. Attensity is doing a great job of showing that.
Then, as we were talking about, it’s just the massive volumes of data that they’re managing. When you’re in the realm of the social world, again, it's not just the volume. I always say that big data is not just big, but it's the velocity, the variety, the ability to ingest very fast, and interpret, analyze, and produce results very fast. That’s really what the Vertica engine is all about, and it’s doing that with very high performance.
It's a very important market segment for us, and it's great to have partners. Vertica is a platform. We rely on our partners to provide solutions to run our platforms. It's social CRM and social analytics and all the kinds of solutions we’re looking to highlight. We love it when we have great partners like Attensity bringing those to market, being successful, and making our joint customers successful.
Gardner: Of course, Howard, your customers are probably not so much concerned about what’s going on underneath the hood, whether it's Vertica, HAVEn, or Hadoop. They’re interested in getting results. I’d like to go back to that Social CRM aspect of our discussion and help people understand why that can be so beneficial, which then of course makes it clear why the technology that supports it is so important.
Can you give us any examples, Howard, of where people have used Social CRM, where they have leveraged NLP and Attensity and what that’s done for them in real business terms?
Lau: Absolutely. Some of the industries we service include industries such as telecommunications, hospitality, travel, consumer electronics, financial services, and eCommerce. We provide the services, the tools for our customers and they implement them for very different use cases based on their priorities.
One of the leading prepaid mobile phone providers use Attensity’s deep semantic approach to analyze sentiment about their service and alert the brand management teams to their unique voice of the customer (VoC).
Attensity effectively measures the overall experience for each brand taking into account their different products and services to determine the accurate wants and needs of the customer. Their whole return-on-investment (ROI) story is how can they use what’s going on in the social realm to manage their install base and minimize customer churn.
Focusing on that, they were able to achieve a 25 percent reduction in customer churn. Now, in the mobile telco space, that directly translates into a 25 percent increase in revenue. Keep in mind that this company is somewhere between half a billion to one billion dollars in revenue. That’s a very sizable return on investment.
We also have other cases where we have an insurance company in the financial services space, and they focus on fraud detection. They use our technology, not only in social space, but also reviewing claims. They were able to reduce workers’ compensation pretty dramatically, to a tune of over $25 million annually, just using our technology, and using our NLP to analyze the data and then figure out which ones they could go after to manage their fraud cost.
Gardner: Where do we go next with this, Howard? We have a capability to deal with large data and the variety of data. We certainly have a great treasure trove of information available from the social media and social web. Combining that with the traditional datasets in CRM, where do you go next? Are you looking for even more datasets and what do you have your eye on?
Lau: Getting more datasets is always helpful. The more you get, the more complete your analysis is, but the view right now is just analyzing big data. We are finding that, within that big data, there are tremendous amounts of individual voices. So the goal is to figure out where these individual voices are and how to build relationships with ones that are important to you.
I’m going to go back to a book that Malcolm Gladwell wrote way back called The Tipping Point. He talks about mavens and the influence of mavens. In the social chatter, there are all these people that have outside influence on other people. The next step in applying our NLP technology in the social realm is uncovering these mavens, so that companies can build relationships with these outside influencers. So that’s one of the next things that we’re really excited about.
Gardner: Tell us also where you are going in terms of services for business. Obviously we have talked about marketing, but are their other aspects—maybe product development? How deeply does this extend into how it can influence a business, not just on the selling and marketing, but perhaps even knowing where their business should be going, a strategy level?
Lau: When people hear about social, the first thing they do is listen, but there is a whole model for how people adopt business solutions in the social realm. We have a model we call LARA, and it stands for Listen, Analyze, Relate, and Act.
The first thing that a lot of companies do is become aware that they need to pay attention to what’s being discussed socially. So they put out these listening posts and they use us to ingest all this information and analyze it for them. The benefit of that is sentiment analysis on companies, on brands, and products. They want this type of sentiment in real time, and we’re able to deliver it in real time.
The next thing companies want to do is analyze the data they have accumulated, and it's for variety of different use cases. I mentioned fraud detection and customer churn. They also want to surface emerging trends. Having an analytical store where you can do what-if scenarios after the fact is incredibly useful for them.
Once they have the store of customer data and they’ve analyzed and segmented their customers, they want to define how they want to relate to the customers, in aggregate or in smaller segments.
The last and final thing they want to do as part of the whole consumer experience is figure out how to engage with the ones that are important to them.
As an example, if someone tweets that they like this phone, that’s great sentiment. But if somebody else tweets that they don’t like the service they’re getting from this mobile phone provider, if that mobile phone provider is an Attensity customer, we actually take that tweet, route it into their customer-care organization, route it to the proper person, and respond to someone in the social realm.
This ability to kind of close that loop, from a person just tweeting generally to his friends about an experience, and then actually getting the customer to hear them and respond to them is incredibly powerful for organizations.
Gardner: For companies that see the value here pretty readily, what steps should they take in order to be in the position to follow that path, that LARA path? Do they need to gather this data themselves? Should they try to ramp up how social media interactions are focused on their products or services? Are there any steps that companies should take in order to better leverage something like Attensity, that’s built on something like Vertica, to get these really powerful insights? Howard?
Lau: That’s part of the value that we bring. All the customer needs to do is recognize that social is important for them. We’re not just talking about corporations that are in the B2C space, but also in the B2B. Once they have that recognition, we’ll handle it for them afterwards.
Part of our products and services offering is that we ingest all this data for them, whether from the social sphere or in the companies emails or customer service notes. We ingest all that information, and they're all defined by one common trait, which is that they are unstructured data. We apply our NLP technology to provide an understanding of the big stream of data and then we create the analytical store for them.
All companies need to do is recognize the importance of wanting to hear their customers, listen to the customers, and ultimately, engage with them socially. They just have to have that motivation, and we will work with them as a partner to realize that solution for them.
Gardner: Chris Selland, I’m thinking that organizations that are sophisticated about this will go to a company like Attensity and get some great value, but eventually they’re going to want to try to get that holistic view of analysis. That means that, not only would they leverage what services and insights that Attensity could provide to them, but they’re going to want to share and correlate and integrate that with what they have going on internally and across many other systems.
Is there something about HAVEn that we should bring out for them in terms of open standards and integration capabilities that allows, over time, for more and more of these different data activities to relate to one another, so that we do get a whole greater than the sum of the parts?
Selland: HAVEn certainly provides a very broad platform of which, as we mentioned, Vertica is obviously a key part, the V in the middle. Yes is the short answer. The solutions ultimately need to be part of a much broader data architecture and strategy around how to leverage all sorts of different types of data, that’s not even necessarily customer data.
Just to give you an example and to make that tangible, there was an airline that I was engaged with not too long ago, probably about a year-and-a-half ago at this point. I can’t name them, but it's a well-known airline, and it was one that didn’t have a particularly good reputation for customer service.
They were working on their social-media strategy and trying to figure out how to make customers who were tweeting unhappily that they hated the airline say nicer things—so how to analyze and respond more quickly.
What they quickly discovered was the reason so many of these customers were angry and saying they hated the airline was that their flight wasn’t on time. What they also realized was they had an awful lot of data on their maintenance operation, and sensor data from the planes, and so on from their fleet.
They saw that by maybe doing a better job of predictive maintenance, keeping their flights on time, and keeping their fleets better maintained, they would actually have much more impact on customer satisfaction than responding to the tweet from the customer who was stranded, which kind of makes sense, if you think about it.
I just bring that example out because that’s an example of data that has nothing to do with the customer. It might be a sensor on an engine, or it might be a performance data of some sort, but it's related obviously to customer satisfaction.
So ultimately, yes, there needs to be a data infrastructure and a data strategy that spans the different solutions. It's not to say you don’t absolutely still need Social CRM solutions and all sorts of different solutions, predictive maintenance solutions and operational, financial analytic solutions, but ultimately the data infrastructure needs to be unified.
That’s really where this is going next. In many leading organizations that’s where it's going already, which is, these solutions absolutely play a key role, but they can’t be 24/7. So there needs to be an infrastructure and a strategy behind them that is very, very holistic.
We're talking about the competitive bar moving here, and that’s the direction that the competitive bar is going to continue to move in.
Gardner: Howard, do you have any reaction to what Chris has said in terms of seeing a value of a holistic data architecture, not only from what Attensity can do, but extending it across many aspects of business?
Lau: I totally agree with what Chris just said. What he’s driving towards is a world where it's really the Internet of Things, where everything is wired to the Internet and they broadcast messages or communicate messages related to their purpose and their focus.
Where we provide our value is that before we get to the world of Internet of Things, there is the Internet of People. People need to express themselves the way they normally do. Where we add value is trying to understand, distill the customers in a person’s voice, and have that complement the future of the Internet of Things.
I totally agree that having an integrated architecture, integrated approach to data management, big data management is crucial going forward.