Market Research 3.0 Is Here: Attitudes Meet Algorithms in Sentiment Analysis

This is the marketer's and researcher's dream.

Reconciling the natural tensions that challenge and befuddle brand planning:

  • Feelings & Facts
  • Sentiments & Statistics
  • Qualitative & Quantitative
  • Focus Groups & Surveys
  • Subjective & Objective
  • Why & What
  • Art & Science

algorithm facesI'll admit, when I first heard about Google, Facebook, and Nielsen studying, decoding and monitoring language and chatter on the Web and "listening to conversations," the consumer part of me got a little bit of the creeps (Big Brother idea).

On the other hand, the market researcher part of me was excited about all of the possibilities. Market research has been stale for a while. Everyone knows about the limitations of the traditional focus group and survey. Do group respondents even tell the truth in an artificial setting where they are served finger sandwiches and paid $100? How can the group think be weeded out to get a real picture of the market? Are the right people answering online surveys? Are panelists professional survey respondents or representative customers?

The explosion of social media channels has the potential to revolutionize market research. New social media-based studies can be conducted more cheaply and efficiently, in real-time and may more accurately capture individual and group opinions. Companies are already mining the words, tones, streams and demographics of social media consumers for their own purposes. User data is deconstructed and then constructed into a picture on brand attitudes, market needs and social habits.

This field is in the early stages of development and there appears to be an overlap and competing terminology among: Sentiment Analysis, Social Media Analysis & Monitoring, Listening, Opinion Mining and Brand Monitoring, to name some.

Sentiment Analysis uses natural language processing (see Google Wave), computational linguistics and text mining to identify the attitudes among writer(s)/speaker(s) on a particular topic.

Companies and brands are interested in using the data to understand how they and their products are perceived and to help predict future developments and market trends.

Bo Pang, a Yahoo researcher and Sentiment Analysis pioneer, identifies three areas for measurement: polarity (positive/negative feelings); intensity; and subjectivity.

More sophisticated service providers include Newssift (Financial Times), Scout Labs, Nielsen Buzz Metrics and Jodange. Twitter and Twitter-based offerings also play in the field.

Sentiment Analysis has the potential to transform not only marketing research but also areas from search to public relations to product development.

But the practice is far from a perfect science.

Computer deciphering of word meaning is not always accurate and tone can be completely missed. Even the leading vendors acknowledge that the data is 70-80% reliable. For example, we may know that the phrase "quite interesting" means one thing in America, another in Britain, but the computer would see the same meaning. Note some of the long-standing issues with voice recognition technology.

There are questions about how robust or representative the data is. Are a brand's tweeters the key WOM influencers or are they just a small vocal segment?

Some brands and products may be under the radar for this technology. Yes we love to chat about Apple but are we prolific, public and passionate, blogging and tweeting away about Charmin or business insurance?

There are conflicting approaches, metrics and offerings; over time a common Microsoft, Google, Nielsen type platform may emerge.

We also need to look at social media as just a channel in the consumer experience mix, and Sentiment Analysis as just a market research tool that is part of a bigger research mix; in-person research and surveys for all their shortcomings offer benefits that social media research cannot duplicate. (even aside from the M&Ms and two-way mirrors!)

Read more of Kevin Randall's blog

Kevin RandallKevin Randall is Director of Brand Strategy & Research at Movéo Integrated Branding (krandall@moveo.com). Kevin consults on brand matters to Fortune 500 companies and specialized health care and business-to-business organizations. His expertise is understanding, integrating and applying research and brand strategy to create business and customer value. Kevin has been invited to speak on brand topics by Google, Harvard Business School, Wharton and Kellogg, and his articles have been published on six continents. His clients at Movéo include Siemens, CareerBuilder, Cardinal Health, Molex, GOJO/Purell and Cancer Treatment Centers of America. Prior to joining Movéo, Kevin worked for Interbrand where he developed brand strategies for companies such as Abbott, Alcatel, Cricket Communications, Ford, GE, McDonald's, Motorola, Nationwide, Roche, Smucker's, and 3M.

Add New Comment

8 Comments

  • Eric Brandenburg

    Great stuff Kevin.
    True that social media is crazy and it is amazing how it is progressing today. Hard to follow lol ;o

    --
    Starting a business? Launching a new product or service?
    Test the market first!
    Ask online for a free quote

    Marketest
    http://www.marketest.co.uk

  • Sid Banerjee

    Great survey of the space. It's important to note that the field of social media monitoring is not a hypothetical field - as you mention there are many brand monitoring services out there now being used and providing real business value to product and service brands today.

    Natural Language Processing is a key part of the solution, and not just in the way it's used by Google to suggest proper spelling of words. NLP technologies help linguistically associate the relationship of words to each other in a sentence - ie "my coffee tasted bitter but the muffin was fabulous" can be processed by NLP-savvy tools to properly assign the right negative sentiment to the coffee and positive sentiment to the muffin.

    Importantly, NLP, sentiment analysis, and customer experience analysis is not simply confined to the world of twitter, blogger, and forum sites on the internet. Many companies leading companies are using sentiment analysis/customer experience analytic solutions from companies like Clarabridge to mine call center conversation notes, customer emails, and open ended survey feedback. Content from these sources can provide more meaningful customer perception and experience feedback because you generally know what product or experience triggered the feedback (captured in the call center or survey ID), some details about the customer (making social/demographic analysis more granular), and you can also "fuse" additional details about the store, customer, product from existing applications to allow more interesting analysis, slicing and dicing of the data, and interesting statistical/driver analyses.

    Our blog contains a range insights derived from real world implementations of sentiment analysis and customer experience analytics solutions deployed over the past few years.

    Many people haven't heard of this interesting solution space, but it is real, and it is growing fast, and it will be deployed by many many more Fortune 1000 customers over the next couple of years.

    Sid Banerjee
    CEO, Clarabridge

  • Joy Levin

    Hi Kevin - I think your comment about it is just one tool is so important. I have used tools such as this and while they can be of some help, they really need to be used within an overall marketing research approach. These tools only show a part of the overall picture, and it is concerning that some companies will use them as stand-along complete market research programs, since the data can be very incomplete and misleading. Thanks for taking the time to blog on this topic.

  • Frank Smith

    A great informational article Kevin. My company, http://www.datascension.com is just now starting a social media and SEO campaign and your article really showed some of the many possibilities that the Google Wave can offer in terms of marketing data to get more traffic and clients to our site.

  • Joel Rubinson

    The ARF will host the definitive industry event on this subject on November 3rd in NY. It's all about listening and the power of social media. Hear Chris Brogan, Google, Umbria, IBM, Motivequest, Lynne d Johnson, Diva blogger for Fast Company, me (also blog for Fast Company) and many more. http://www.thearf.org/assets/f...

  • Aly-Khan Satchu

    The stream of conciousness that is Twitter represents a step change in this kind of analysis. Interrogating that stream will be a major growth area and for brands its plain dynamic and disjunctive.

    Aly-Khan Satchu
    www.rich.co.ke
    Twitter alykhansatchu

  • Kathryn Korostoff

    Another company to watch in this space is Crimson Hexagon--they have a very innovative approach. http://www.crimsonhexagon.com/...

    As a long-time market researcher, I think the challenge is a very practical one. We now have so many methodologies to choose from--whether time-tested traditional ones or bleeding edge, SocMed-based ones, it can be hard to design an optimum research strategy for a given need while balancing a client's timeline and budget needs. As you say, mixing traditional and newer methods can often be ideal--but it can also get very expensive and very time-consuming. Still, striking the right balance is possible especially when working with open-minded clients. For one recent case, I worked with a client to use SocMed-based methods to help inform hypothesis development that they then used in a broader, more traditional study. Due to time and budget, we were not able to use an elaborate approach, but even using simple tools to monitor relevant brand discussions (for the client and 3 of its top competitors) we discovered some brand perceptions that were very helpful.

    Overall, I am thrilled that we researchers now have this powerful class of tools to add to our mix!