Survey Says
Customer surveys have long been cornerstones of good market research. Surveys allow companies to measure attitudes about products, concepts and services. They also help them build customer profiles for marketing purposes. Instead of modeling predictions about customer behavior based solely on the "what" of customers, demographic profiles or buying histories for example, surveys help explain "why" customers do what they do. With better modeling, companies and organizations can respond better to customer needs and increase loyalty.
Today, it has never been easier to conduct surveys. New technologies are creating a wider role for surveys and new opportunities for incorporating such customer feedback into CRM analytics. Not only does e-mail present a ready vehicle for contacting and soliciting participation, there are by some estimates, hundreds of companies that have sprung up offering Web-based survey services as well.
These vendors, many of them startups, provide easy, browser-based tools for constructing surveys, selecting sample groups and launching surveys. Some, like Zoomerang from MarketTools and QuickTake from San Francisco-based Greenfield Online, offer as many as 100 ready-to-use or customize templates for such common survey varieties as customer satisfaction, brand awareness and market potential. Virtually all it takes to get started is a credit card.
The unprecedented advantage of online surveys is the potential to contact and get responses from a sizable percentage of the customer base very quickly and cost effectively. "Everything we do happens in real time," says Mark Hardy, Greenfield Online's vice president and general manager of QuickTake. "The minute you fire off a survey and regardless of whether it is going to an online environment, an offline environment, voice recognition or a mobile environment with wireless, what you are doing is capturing the information as it comes in real time."
Speed Reading Customer Attitudes
Survey usage is on the rise enterprise-wide. Survey tools are finding application in various functional areas: among product development teams to get customer feedback on product concepts, advertising departments for evaluating ads and HR departments for measuring employee satisfaction or doing exit interviews.
But the market research department is perhaps the key beneficiary of the ability to conduct quick, inexpensive and wide-reaching online surveys. "The group that's responsible for market research, particularly in the area of marketing and product development, is going to have more complex needs with respect to data collection and the capabilities of the tool, whereas other departments are likely to have less complicated needs," says Nigel Hopkins, vice president of MarketTools in Sausalito, Calif.
The attraction for market researchers lies in unprecedented opportunities to conduct cost-effective research and do things like present visual stimuli and streaming video. "Interactive media is going to enable collection of data in ways not previously thought of," says Hopkins, "You might be able to visually move things around on a page on the Web by dragging and dropping things, whereas with paper and pencil you only have one chance of doing that."
Just as important as the great reach and flexibility of the Internet is the speed with which surveys can be completed on it. Launch a survey on your way to a meeting and have results by the time you get there. That's the kind of promise some of the easy-to-use online survey tools offer. "There used to be a set term, which was a week. That became two days. It's now minutes. In fact, although it may be statistically questionable, it is minute by minute as the survey is going on," says Richard Kottler, director of product direction with the market research arm of SPSS, a marketing software and services company in the UK. The sooner responses come in, the sooner they can be reported. "So we are putting a lot of effort into analytical reporting and distribution of results instantaneously in a browser, whether that is simple tabulations, simple reports or graphics," he adds.
Online survey tools typically provide basic reporting and data mining functionality for exploring data and summarizing it both statistically and graphically. "If you want to do some quick analysis, we allow you to do some cross-tabbing on the fly. QuickTake quickly allows you to assess what are some of the opportunities or red flags you may have with your customers," says Hardy.
There are even some tools for data mining on open-ended questions, which are likely to include a variety of unique responses. For example, with QuickTake, responses to open-ended questions are captured in a comma-delimited file that can be included in cross-tabulations. You could view the actual responses of different subsets of customers, perhaps comparing the reasons why people who have been customers for more than three years like your product versus why new customers do.
"We are improving text searching and matching, and we have a product that will classify responses so that you can analyze them in categories," says Kottler. "But you are at the mercy of respondents' spelling, spacing and things like that, which is never good enough." He notes that what many corporations particularly like to do for CRM is to actually read through all of the open-ended responses people have submitted. "Even long, boring text reports can be of interest because you get better flavor." A simple reporting of the text is not enough, however, he says. Instead, it's preferable to have them broken into certain classifications, favorable customers and unfavorable customers, for example.
Developments in voice recognition and voice over the Internet promise to create even more high-tech reporting options in the future. Much in the same way that telephone surveys are recorded allowing researchers to hear customer voices, interactive voice surveys could be conducted and recorded over the Web. "What we do in telephone reporting today is that we report the answers verbatim on the computer so that you can listen to them," explains Kottler. "I think that is coming to the Web pretty soon. As people are using more and more voice communication on the Web, we will be recording actual statements rather than requiring people to type in."
Combining the "What" and the "Why"
What matters most is that survey data can provide new insight into why customers buy what they buy and do what they do, why they become loyal and why they defect. Survey information must provide a new dimension of attitude and opinion that goes with the behavioral information that's captured through a CRM system.
"A big point of differentiation here is, in the past, when companies mined their databases for CRM information, it was based on behavior patterns--what a person bought, how often they came to purchase something--whatever the attribute may be that they were looking at. What QuickTake allows you to do is perform that same type of analysis, but based on what people are thinking and how they are responding to your questions," says Hardy.
Other CRM components like call centers and sales force automation systems may answer the who, what, where and when of a customer purchase, but they don't explain the why. The impulse to purchase, say, a red long-sleeved silk blouse at $49.99 could have been because it was red and the customer wears only red. Or it could be that she prefers silk. Or long sleeves. Or thinks cheaper blouses are no good. "You now have the ability to interact with customers and ask them questions that are relevant, rather than just making assumptions on sales figures," says Hardy.
When attitudinal information is integrated with all of the demographic, historical and transactional information a company may have collected in its databases, it can be analyzed to give a more complete view of customers and improve the quality of predictive models. "We provide fairly straightforward univariate statistics," says Hopkins. "It's a calculation of means, frequency and summary statistics at a variable level so that you can get a quick overview of the basic trends, patterns and so on in the database collected. If somebody wants to go further into the data, we support different export types and allow them to download data out of the relational database in a format that they can import into other applications like SAP and SPSS for further analysis."
Other vendors also enable such links. SPSS provides a solution for this comprehensive analytical CRM approach with its CustomerCentric application. CustomerCentric is designed with predictive data-mining models that draw from a company's many customer inputs. "The idea behind CustomerCentric is that it extracts, transforms and loads data from a variety of sources and puts it in a data repository that is set up specifically for the purpose of analy-tics. Once the data is in that database, we have a set of pre-built, industry-specific measurements and predictive models that give you profiles of customer affinity."
Profiling and collaborative filtering can be used to create scores for customers regarding their propensity to respond a certain way. That kind of data mining helps a company be proactive about offering customers what they might want. "We have pre-built models by industry based on what we have seen are best practices there. Then we also have some cross-industry models that we bring to the table," says Mark Battaglia, executive vice president of corporate marketing at SPSS headquarters in Chicago, Ill.
"That is a very effective model since it is derived from the data that we have, the facts and the opinions," says Battaglia. CustomerCentric also links with operational CRM systems, completing the loop of understanding customers more fully though surveys and then responding to them with solutions that are likely to fit specific needs. It also allows for an organic process of data mining and modeling. "Every time you have a new piece of information come in, it updates itself and asks, 'Is the original set of models you built still effective?'" explains Battaglia. If those models become less effective, an analyst can be notified to set up a new model that works more effectively.
The Old Guard Grumbles
Despite their promise, online survey solutions do catch some flack for lacking the complexity of scientifically designed and custom market research surveys. E-mail surveys and some of the simpler technologies out there don't offer the sophistication of their more robust survey cousins. Questionnaires in flat text format, for instance, cannot employ such desirable survey techniques as skip pattern logic, randomization in order to remove any top click bias or error checking. Nor are the basic systems much good at creating an interview "experience." That is, instead of just tabbing from field to field, respondents are given a more interactive experience by such techniques as branching them around a questionnaire and asking questions based on previous responses.
The loudest complaint from the traditional market research industry is about the quality of online samples. In an industry that takes great pains to eliminate biases and get at objective results, they warn that sampling an Internet population is inherently biased. Only certain customer types use the Internet, they argue.
"I remember a quote from one of my very first clients in the Web survey field, way back in the dark ages of 1997, a lady at Reuters who said it may not be representative and scientific, but if all you want is just a gut feel of what's happening out there, you can get a quick answer-as long as you don't make major strategic decisions based on it. So, while the Puritan in me says it isn't any good, there's a hell of a lot of use for it," says Kottler.
Part of the reason that criticism of this sort is starting to fade is that the Internet now reaches an audience that is representative of most segments in society and business. "In the early days, it was difficult to find people other than maybe technology buffs online," says Hopkins. "As the Web has become more prolific, the probability of finding a suitable sample on the Web has increased significantly. Now, it's fairly easy to find anyone and anybody."
Targeting techniques are also getting more sophisticated. Companies can utilize clickstream data, on-site intercepts, newsgroup postings and sophisticated banner ad networks among other online and offline approaches to reach target groups and invite their participation in a survey. There are also more interactive technologies like chat, as well as hosted solutions and downloadable survey applications that can deliver more complicated survey experiences, high-end graphics and sound.
The number of people on the Internet has many of the traditional market research companies keenly watching the opportunities to increase the breadth of survey reach. Not only is the total number of potential respondents large, but the medium also offers the potential to connect with many people who might otherwise refuse to participate in traditional market research, like a telephone survey.
There are a growing number and variety of ready and willing panels on the Internet also. Minneapolis-based Custom Research (CRI), part of GSK Group and the fourth largest market research firm in the world, for example, is in a partnership with Digital Marketing Services that provides technology that can reach over 23 million America Online members via Opinion Place.
"I think for the research industry, we are at a really evolutionary stage here," says Beth Rounds, senior vice president of CRI. "I think the Internet is challenging some of the old companies, including CRI, to change the way we do business and take the old consulting part of our business and combine it with this new technology to make an even better company."