Chapter 2
Evaluate and Select Listening Solutions
Conversational listening methods are rooted in the tradition of media-content analysis, a process that enables researchers to learn what people are saying, writing, or commenting on. Used during and since World War II, content analysis has provided insights for military, academic, and business purposes. For many years, the process relied on trained individuals (“coders”) to manually extract terms, brand names, people, organizations, issues, actions, relationships, and sentiments from all types of published materials: magazines, newspapers, journals, transcripts, images, and proprietary comments from open-ended questionnaire items. The ability to process the coded information for statistical analysis has advanced from manual methods to mainframes to smaller, more powerful computers and, more recently, Web-based services.
Free or low-cost blogging tools from such service providers as Blogger, Twitter, and WordPress, along with the creation and deployment of customer feedback and ratings systems for products (Bazaarvoice), video and photo sharing (YouTube, Flickr), social bookmarks (Digg), and social networks like Facebook, Orkut, and CyWorld have broadened content analysis from the study of mass media sources to everyday conversations, especially those held online. With a few mouse clicks or touchpad taps, anyone can become a writer, replier, publisher, producer, commentator, reviewer, pundit, advocate, or even jerk, on a new and unprecedented scale.
People's thoughts, feelings and conversations, previously hard to get to, are now shared, visible, directly accessible, collectible, and analyzable. No longer are people's expressions restricted to “letters to the editor” that may or not be published, correspondence with the “complaint department” that may or may not be acted upon, or practically irretrievable remarks in customer relationship management (CRM) notes.
The result: Marketers can listen to great numbers of the people who are interested in their categories, who may buy or currently own their products; learn from them and develop insights; engage in conversations with these consumers if they choose; and act on what they hear.
These remarkable developments have stimulated today's widespread interest in listening, and contributed to the rapid growth of listening-related services and products, including search-term analysis, text-analysis software, full-service platform vendors, community operators, and offline word-of-mouth researchers. Innovations coming forward are providing the technology, services, and expertise that help brands listen to people and develop insights capable of creating business advantage.
The variety and growing number of solutions available to brands and their partners is mind-boggling. A Google search for “social media monitoring” garners results with titles like: “A Wiki of Social Media Monitoring Solutions,” listing 149 of them, or “12 Social Media Monitoring Tools Reviewed.” Go to LinkedIn and you'll see a thread in the Social Media Marketing group of 294 responses to the question: “Can anyone recommend a good social media monitoring tool?” (LinkedIn 2010). The first answer was: “My boss just showed me these two. They look pretty good.” I'll spare you more examples; the point is that while many people create lists or comment, meaningful guidance is hard to come by. This chapter aims to correct that trend and provide you with the ability to navigate confidently through an active, yet often confounding, marketplace to the solutions suitable to your listening needs.
Five Types of Listening Solutions
Today's solutions market leverages 30 years of product experience and addresses the most up-to-the-minute customer demands. Countless companies are backed by venture capital and private equity firms that see the enormous potential of listening. As it is in any dynamic market, it can be a challenge to sort out which solutions—or combination thereof—meet brand listening requirements and budgets. We developed a classification system based on the various properties that software solution classes share in common. Five groups that we discuss here comprise our scheme:
- Search engines
- Media monitoring
- Text analytics
- Private communities
- Full-service listening platform vendors
The following pages describe the key features and analytic capabilities of each class, and considerations for their use. Leading vendors in these categories are compiled and summarized in the Appendix, “Listening Vendor Profiles: A Resource Guide.”
Search Engines
This group includes two types of solutions: general search engines and real-time search engines. Search engines are typically used to get quick reads on topics of interest in pop culture, and also for competitive analysis (see the Tassimo case in Chapter 13), or to gauge consumer interest in an issue, such as iPhone4 antenna-gate. Without being hyperbolic, the range and variety of topics that can be explored through search is virtually unlimited. The key to using them well resides in the ability to search on keywords that capture the topic of interest.
General Search Engines
Search engines index a range of online sources to make them searchable and the results accessible. Google, Yahoo!, Bing, Ask, and AOL are the top 5 engines, with Google the most popular.
Google and Yahoo!, in particular, offer useful analytic capabilities for listening. Google's robust research tools permit studying search term trends, related or contrasting concepts, and competitors (Google Trends, Google Insights for Search), whereas Yahoo! is especially strong in the areas of popular culture and buzz, providing a snapshot of a searcher's current and recent interests (Y!buzz and Yahoo Buzz Index).
For tracking, Google provides an alert system and the capability to download results from Insights for Search and import them into spreadsheets or databases for further analysis. The major engines innovate and occasionally add features that are inspired by Twitter and real-time search.
Real-Time Search Engines
This newer class of search engines specializes in social media, and chose not to duplicate the big engines’ coverage. Real-time search vendors concentrate on harvesting social media sources where people are actively commenting, writing, Tweeting, uploading, or updating. Sources include blogs (e.g., Technorati), Twitter, social bookmark services (Digg, Delicious), photo-sharing sites (Flickr), video-sharing sites (YouTube), and full-on social networks like Facebook. Real-time search engines monitor and report activities as close as possible to the time they happen.
Real-time search engines often provide information about items’ “sociability” by detailing the popularity of posts or tweets, identifying trending topics, assigning sentiment, linking to authors and recipients, and listing influencers. Some even provide tools for people to act on the information by e-mailing, commenting, or re-tweeting. For companies who listen, real-time search can be a no-cost, bare-bones monitoring option and entry level engagement tool for customer service and sales, for example. But keep in mind that while they are helpful, they do not offer much in the way of administration and reporting capabilities.
Search Engine Considerations
Match engines to listening requirements: It's crucial to keep in mind when using search engines or real-time search for listening that they differ from one another in various areas, including popularity, the sources they cover, their ranking routines, the audiences they attract, and their value propositions. Google sets out to “organize the world's information,” whereas competitor Bing specializes in helping searchers use the information (especially in travel, shopping, and finance). Since the engines differ from one another, the results do so as well. The implication for listening is clear: Every listening initiative needs to vet the engines in light of research requirements to make sure that those selected match their needs. Additionally, real-time search engines can present spam in their results. Ensure quality in data by applying filters to remove as much noise as possible before the conversations are processed.
Social Media Monitoring
Social media monitoring provides a major step up from real-time searching. Tools and capabilities allow end users to observe, measure, analyze, and report on social media activity. Monitoring tools help organizations track mentions of their brands, products, competitors, and industry; topics and issues; words or phrases of interest; key executives; as well as marketing messages, the sentiment around them, and key influencers (which can be people, blogs, or sites). As we mentioned in Chapter 1, the primary goals of monitoring are public relations, including reputation management; company and brand protection; and customer service, outreach, and engagement. Chapter 12 on managing reputation and Chapter 14 on customer care and satisfaction present a variety of cases that use monitoring.
Most solutions offer dashboard displays that report and visualize the monitoring data as stats or trends over time, and as word clouds that reflect the popularity of keywords. Filters to narrow down results by keyword, source, or geography, for example—coupled with drill-down capabilities—enable users to access and read the individual posts, tweets, or comments underlying the numbers that help to explain the “whys.” (See the M.D. Anderson case study in Chapter 13, which used social media monitoring to uncover and meet men's health needs.) Several companies offer social CRM; workflow and collaboration features that integrate monitoring with managing accounts on services like Twitter or Facebook; and alerting, escalation, routing, response, and resolution services that help with outreach, engagement, and customer care.
Most vendors provide their solutions as a service that is accessible through an Internet connection and Web browser. Though pricing and licensing schemes vary, charges may include a flat rate per user per month and/or tiered fees based on quantities or usage, such as the number of keywords tracked or the number of results returned.
Social Media Monitoring Considerations
The primary role of social media monitoring solutions is to equip companies with tactical tools and capabilities, often for customer service, such as JetBlue or Comcast show (Chapter 14) or reputation management, as seen in the Hasbro example (Chapter 12). Since these are essentially self-serve solutions, it's important to ensure that they are configured for each company's listening needs, and that companies understand the strengths and limitations of their solution. Several considerations influence the business value derived from monitoring:
Retrieving brand mentions: Organizations that conduct brand-name tracking must make certain that they're able to retrieve brand names in meaningful quantities. That is not always easy to do because—except for ratings and reviews and brand backyard sources—brand names may appear only occasionally in consumer backyard conversations. For example, only 5 percent and 40 percent of conversations in food and automobiles, respectively, specify brand names. Retrieving these mentions that are about the brand and not something else is another point to consider. O'Brien and Rabjohns (2010) found that during the Beijing Olympics it took 250 search arguments to separate out the brand Visa from the travel document. They add that car rental companies with generic names, such as Enterprise, Dollar, and Thrifty, face this problem. Research for which you are unable to separate out generic terms from the brand terms does not yield good results.
Understanding context: Because the people talking are not always mentioning brands, analysts need to search for and retrieve conversations relating to brand discussions in order to understand the different contexts in which they occur. For example, conversations about hot dogs hardly mention brand names; but they take place in the context of family gatherings, sporting events, and feeding children. Unearthing those discussions requires subject expertise and artful querying of the social media sources.
Reporting: Companies need to make sure that dashboards present data that is appropriate and relevant to their listening requirements. They must require that reporting is flexible so that they can generate reports that support their needs and apply to their decisions.
Text Analytics
Text analytics software companies are another rung up the ladder from social media monitoring. These vendors supply software that provides tools and capabilities for collecting both brand and consumer backyard conversations; processing, analyzing, and reporting them; providing workflow and collaboration tools for groups or teams; and integrating data with CRM solutions, third-party, and in-house systems to provide a comprehensive view. Although they perform the brand monitoring functions just described in the social media monitoring section, their strength resides in their advanced analytic capabilities, which include uncovering and evaluating the importance of topics and themes; discovering relationships and facts; and revealing emotions and motivations.
Several of the case examples that we cover in Part II utilize text analytics, and illustrate its powers quite well. Three that are worth looking into are: developing a new women's food product (Chapter 11); understanding the different meanings of terms and contexts for “broadband,” to create messaging and inform media planning (Chapter 25); and detecting the competitive positioning of two brands, Tylenol and Advil, in the public's mind (Chapter 18). One quick example: Gaylord Hotels text-analyzed post-stay comments and discovered that although noisy rooms were not a frequently experienced problem, they were one of the most serious because they were highly correlated with “wouldn't return” or “wouldn't recommend” responses (Stodder 2010). Text analysis is also used just like quantitative data: as inputs into models that predict business outcomes. Examples in Chapter 18 show how text-analyzed listening data was used in models that predict product sales.
Text analysis solutions aim to assist in all aspects of listening initiatives. They are available as enterprise solutions installed on your company's hardware and as Web-based self-service tools made accessible as a service.
Text analysis has a long history; it's over 45 years old, and is a very technical field that has emerged from computer science and computational linguistics. Describing it technically is beyond our scope. If you are interested in an excellent and accessible primer, read Seth Grimes's two-part series on text analytics basics (Grimes 2008).
Text Analytics Considerations
Complexity: Text analytics is a complex undertaking that requires specialized expertise. Many listening initiatives that use it do so through vendor and consultant services that offer the right mix of people, process, and technology. Companies that purchase this kind of assistance should carefully evaluate suppliers for their knowledge, practical experience, and capacity. Text analytic services are not commodities; the related people, skills sets, and domain expertise vary from one company to another, as does their capacity. In fact, one large company buyer we interviewed declared that while some companies did great work, many of them couldn't handle several projects simultaneously, which left them with the option of waiting or selecting a less-preferred supplier.
Project focus: As with any major initiative, establishing and sticking to clear goals up front improve one's chances of getting a completed—and, hopefully, fruitful—research outcome. Clear leadership, excellent project management, and communication are vital, as is the willpower to resist the temptation to expand the initiative's scope as it unfolds, and thereby risk losing focus. It is occasionally necessary to change scope; if this is the case, then operate in a change-and-control fashion, whereby you first approve changes and then establish their impacts on project budget and timeline.
Staff resources and new skills: Every solution discussed requires hiring people who are trained in the use of the various tools and have an aptitude for listening. Initiatives that incorporate text analytics require employing staff with a higher level of skills in working with unstructured data, such as content analyzing, coding, classifying, clustering, and taxonomy building. These individuals must also understand the business, industry, and customers. Software, of course, is not perfect. For example, though it will suggest topics, categories, and themes, it usually takes a human being or team of them to make sure that these parameters are relevant and meaningful to the research objectives; and if not, to tweak and modify them so they are, and iterate until they're right. Companies that purchase text analytic services need at least one knowledgeable staffer who can interact with and guide the vendors, and serve as the bridge between the company and the research team.
Private Communities
Private online communities invite qualified customers and prospects to engage with brands in prearranged ways through projects, assignments, and conversations, and with other community members in free, natural, unstructured methods by using social media. In these respects, communities are best regarded as “brand-backyard” sources. By referring again to Part II, you can see how NASCAR listened to its community and took their suggestions to change racing to make it more exciting (Chapter 13), or how publisher Meredith used its community to reposition Ladies’ Home Journal (Chapter 16), or the way CDW listened to its community and developed an effective sales strategy (Chapter 14).
Companies that run communities manage sizable “living labs” for marketing research. By having members participate for varying periods—from at least one month to several months or longer—these communities make it possible for brands to run multiple projects simultaneously or progressively, and to listen to the evolution of their core customers’ and prospects’ discussions. Communities aid marketers in developing customer insights, brand strategy, new products, and innovation strategies. Community vendors typically provide benefits such as member recruitment, qualification, and activation; member accounts and social networking; market research capabilities; deployment; analysis and reporting; and professional services for community operation and ongoing engagement.
Community Considerations
Be open to community guidance: Use communities for more than just “quick answers” to marketer-initiated questions and routine evaluations. Connect with members to explore concepts—like wellness—or behaviors—such as stocking up on products—more deeply and develop richer insights. Companies that run communities, like Communispace, strongly advise marketers to remain open and receptive to the guidance the community gives. That requires humility and the capacity for marketers or advertisers to see themselves in their customers' lives, rather than as consumers of their products.
Full-Service Listening Platform Vendors
Listening platform vendors provide technology, services, and consulting that offer end-to-end solutions for analyzing both offline and online word of mouth.
Full-service vendors furnish many research capabilities, including those of the text analysis vendors, but surround them with value-added services. Consulting and service offerings focus on listening project management; analysis; reporting; and evaluation; and, sometimes, creation, implementation, and evaluation of marketing, media, advertising, and public relations programs.
Considerations
Match vendor strengths to listening requirements: Although vendors may have the right mix of offerings and business strategy, they differ from one another on various dimensions, like the availability of consulting, the strength of their services and analytic capabilities, the robustness of their technology, data source coverage, or ease of use (Forrester 2010). Additionally, strong project management and clear communication are essential for initiatives to keep momentum, stay on track, and meet milestones. Taking on a solution is always more than the technical capabilities of software; it's an engagement and relationship in which both parties should contribute and benefit.
Listening solutions—the software and methods for doing social media listening research—fit into five groups: search, media monitoring, text analytics, private communities, and full-service listening platform vendors. Each category has its own capabilities and value, ranging from tracking search terms to counting and trending mentions to slicing and dicing to predictive modeling of business outcomes. Solutions can be used alone or in combination, depending on how they fit with goals for the listening initiative.