One of my favorite perks about this whole blogger thing is meeting some amazing people, and every once in while even getting to know people I consider to be my “heroes”. Seth Grimes is one of those folks. Since I began exploring the possibilities of social media analysis, text analytics, “Big Data”, etc.. over and over again I would run into some piece pf genius written by Seth. In many ways our current careers parallel in terms of overall positioning and strategies, but Seth has achieved a level of reach, influence, credibility, and thought leadership than I can only aspire to. He’s just that good.
Seth is an analytics strategist with Washington DC based Alta Plana Corporation . He is contributing editor at TechWeb’s InformationWeek, founding chair of the Sentiment Analysis Symposium, and Text Analytics Summit, and text analytics channel expert for TechTarget’s BeyeNETWORK.com. He is the leading industry analyst covering text analytics. Seth consults, writes, and speaks on business intelligence, data management and analysis systems, text mining, visualization, and related topics.
Seth invited me to attend his upcoming Sentiment Symposium as a guest blogger, but I just couldn’t fit into my schedule. Instead, I suggested that we do an email interview to talk a bit about the symposium, but also about his views on where market research fits (or doesn’t fit) in the new Business Intelligence paradigm taking shape right now. I think his take is important for market research to hear and I think you’ll find a lot of value in what he has to say.
We conducted this interview via email over the past week. Enjoy!
LM: Thanks for agreeing to chat with me Seth! As a long time fan, it’s a real honor. First off, you’re deeply involved in text analytics and sentiment analysis. How and, more important, why? And what’s in it for market researchers?
SG: My thanks to you Lenny for inviting me. I’m always honored when people value what I have to say!
How is easy: I’m a consultant and industry analyst. I help user organizations and solution providers with analytics strategy. This work involves business intelligence and text analytics and their application to meet business challenges.
Why? Personally I’m fascinated with language and use of automated technologies — natural language processing (NLP) and computational linguistics — to help machines get at meaning and discover patterns.
What’s in it for market researchers? First off, the technology will help you automate analysis of free-text survey responses, verbatims. There’s huge potential ROI just in that step. I know of one organization that, via use of text analytics software, was able to reduce processing of periodic surveys from one person-week to half a day’s work. But beyond surveys, the technologies allow you to turn the Net — online and social sources — into one huge focus group and to draw insights in near real-time.
LM: You’re involved in a number of conferences including the upcoming Sentiment Analysis Symposium. Can you tell me a bit about these events and what your goal is in having them?
SG: The conferences are a natural for me, an outgrowth of the writing, speaking, and consulting I’ve been doing for years. So, we have —
The Sentiment Analysis Symposium, coming up November 8-9 in San Francisco, and the Text Analytics Summit, — where folks in market research, marketing, customer experience, financial service should be in order to best exploit attitudes and emotions in online and enterprise source, and where they should be heading. And I’m founding chair of the summit, which started in 2005 with similar goals, covering a broader area however.
The conferences are at the intersection of technology and business, about discovering insights in content that contribute to better decision making. They’re about learning and making connections.
LM: Over the past year my career has followed a similar path (consulting, writing, speaking leading into event organizing), so I can relate to the trials and joys of putting these things together! I’ve found an intense interest in topics related to emerging business intelligence technology from a fairly small segment of the marketing and insights communities, and a lot of resistance to embracing these new approaches from the rest. Has your experience been similar, or are you finding growing interest from a broader audience? If so, what is fueling the change?
SG: The audience is growing, as folks understand the technologies’ potential and as they learn how leading-edge organizations are benefiting from it. I try to cross-pollinate where I can, to evangelize analytical technologies in business domains that could profit by adopting them, and to bring business concerns to technology companies. There’s significant need.
Text and content analytics appear first on my agenda, as means of discovering and exploiting the business value of content. I’m also a booster of integrated analytics. The aim is to link content-sourced information with data from transactional and operational systems and — given new, renewed interest in location intelligence and in the “when” of clickstreams, transactions, and behavior tracking — geospatial and temporal analyses, joined via semantics.
OK, that’s a load of techno-speak, so I’ll restate by saying this stuff is going to be — getting to be — huge. It will surface in augmented reality and other consumer-facing applications, with smart content and advertising delivery, sensitive to context and situation, critical tools for business competitiveness. Yet the mainstream BI and market-research worlds are only starting to clue in. Resistance? More, I’d say, a lack of vision in some cases and of time to consider the possibilities in others.
LM: There is certainly a lot of energy being applied to developing new tools in this space; what is your take on the current “state of the industry”? How close are we to fulfilling the potential of these technologies?
SG: I’ve been using a photo of Alan Turing in recent presentations. Turing’s 1930s work defined computability, and he was also a marathoner who almost qualified for the 1948 British Olympic Team. I use a photo of Turing running, and only once has someone in my audience recognized him. I show Turing as a runner because we’re engaged, with text analytics and sentiment analysis, on a course toward machine comprehension of human language and, complementing understanding, machine language generation, toward machines that can pass a modernized Turing Test. The race is on, but we’re still a long way from the finish line.
LM: What recent developments in the field are you most excited about, and which company do you think is closest to “getting it right” in terms of the practical application of these technologies?
SG: What’s cool? Beyond-polarity sentiment technologies, which detect mood and emotion, not just in text but also in speech. Image and video analytics: Information extraction from even more sources. Identify resolution: What’s someone’s demographic and psychographic profile? Question answering — that is, semantically-infused information access that goes way beyond search — the kind of stuff we’re seeing in IBM Watson, Wolfram Alpha, and Apple’s Siri.
Cool stuff, but in terms of meeting basic, right-now business needs, there are actually a fair number of companies getting it right. I won’t answer in print, but folks should get in touch or attend the conference to learn about them!
LM: LOL, fair enough! On that note, who are you most excited about hearing at the upcoming Symposium and why?
SG: I curated the program to appeal to a business and technology common ground. It’s designed for people working in customer experience and CRM, marketing and market research, competitive intelligence, financial services, and so on, and not just myself. You should check out the agenda, which is online ; it’s all great (although I admit to bias)!
But actually, what I’m really, really looking forward to is just chatting with people — speakers and attendees — during the breaks and the pre- and post-conference receptions. Frankly, I learn the most in those informal, unscripted conversations.
LM: A lot of media coverage has been given to the idea of “Big Data”, and I certainly see what appears to be a fairly rapid wave of consolidation, new entrants, and repositioning from the big tech firms taking place. It seems as if the focus of all that activity is to make a play for data ynthesis/convergence to support the “Big Data” idea. What role is text analytics and sentiment analysis going to play in bringing this brave new world to life?
SG: Yeah, Big Data, this season’s buzzword. It’s marketing speak, and we’re already seeing backlash, that the challenge most often isn’t volume, it’s complexity and data integration.
Much of that complexity is created by the desire to bring text-sourced information — facts and opinions — into the analytical mix. You need “natural language” to explain what you’re seeing in the numbers… hence our conversation now.
LM: OK, last question Seth. What changes do you expect to see in the next 5 years in the market research space as a result of the advances in text analytics, sentiment analysis, and “big data” integration/analysis? How does the traditional survey/focus group paradigm fit into that future?
SG: heard a speaker say, earlier this year, that with a “culture of listening,“ there is “no need for surveys.” I posted a photo of his slide to Twitter — the gentleman is director, consumer services at large CPG company — which ignited a Twitter exchange with consensus: No, you need surveys. For customer-experience initiatives, for market research, you can’t learn everything you need to know without systematically asking a set of directed questions to a known set of respondents. Text analytics, sentiment analysis: These technologies will help you do better surveys, with larger numbers of respondents, even flash surveys (let‘s call them) that can be turned around really quickly.
Focus groups, on the other hand, are slow, expensive and subjective. As I see it, they are very replaceable by online/social-media monitoring. Bye-bye.
We‘ll see even further linking of survey- and social-sourced insights with behavioral and psychographic profiles inferred from “big data” clickstream, location, service utilization, transactional, and other tracking data and mined from content. This triangulation — ensemble methods that coordinate and combine multiple models and approaches — is the way to go.
LM: You’re preaching to the choir my friend; I couldn’t agree more that the future is about the synthesis of multiple data streams. Thanks so much for the great conversation Seth and good luck with all of your efforts!