Heating up, cooling down
Editor’s note: Andrew Jeavons is a Cincinnati-based consultant on technology and software issues related to the market research industry.
There is certainly no shortage of market research software. The software directories at www.quirks.com and Tim Macer’s Web site (www.macer.co.uk) list hundreds of packages. A quick search on Google for the words “market research software” nets a lot of hits. Even eBay has listings for market research software!
It’s obviously partially the result of the Web. The nature of the Web lends itself to the development of interviewing systems, thus the number of Web interviewing systems has skyrocketed. When the Web as we know it was invented it was based on a document-linking technology called hypertext. It was read-only, and interaction was limited to jumping from document to document. However in a short space of time features were added to enable information to be collected by the Web browser software and passed back to the Web server for processing. The Web then gained a structure that made it a question-and-answer-based environment. CGI scripts (common gateway interface) that process information from Web forms on the Web server served as the starting point for Web interviewing systems. A fundamental feature of the Web is to ask questions and provide an answer (search engines for instance).
CATI software hasn’t had such a growth, and this is almost certainly because is far, far harder to develop than Web interviewing software and a lot harder to sell. The belief, which was largely erroneous, that there was a huge amount of money in Web interviewing software led to a rash of Internet start-ups, very few of which have survived. I’m willing to bet that those who have survived now realize that survey software is not a gold mine.
Business models have mutated drastically. The sheer volume of software available for Web interviewing now means that Web interviewing software is getting close to being a commodity, which is a mixed blessing for the buyers. It means that you can buy on price, but it also means that quality software, which you may need, faces a tougher time surviving. Cheaper is not always better in the long term.
The survey software world is, at the moment, an interesting place. In this article I’m going to emulate (with all due acknowledgment) Wired magazine’s Wired/Tired/Expired feature for technology and culture and present a view of what’s hot and what’s not in the survey software world.
Not: Web interviewing software. Enough already. There are more systems than you can shake several sticks at and it’s getting hard, very hard, to tell exactly what is unique about each one. Do we need any more? Well, unless it has some really unique features - and “drag and drop” questions, customized JavaScript pages, thin client scripting systems with tree-based questionnaire views and embedded Java applets aren’t unique - then I don’t think we do. We’re still stuck with CATI-on-the-Web and there is little sign of any real innovation from anyone, be they dot-coms or established vendors. What about quality control metrics and procedures for Web interviewing? Because Web interviewing is completely software based we have a unique opportunity to track what is going on in an interview and use this to improve the quality of the data and the quality of the experience for the respondent. E-mail sample management systems? How about making the interviews more FUN (and providing some technology to help do this)? When we lost the interviewer in the transition from CATI to the Web we also lost the ability to drag a respondent through 45 minutes of mind-numbingly tedious questions. People just quit now; there’s no bonding with the interviewer. We can’t rely on the “just a few more minutes...” gambit anymore. Web interviewing software is facing a real challenge now as surveys carried out on CATI are moved to the Web. A lot of these CATI surveys are very complex and the newer Web interviewing systems simply can’t cope with them.
Hot: More online qualitative software. This is one of the new frontiers for survey software. The Web is very good at connecting people, we just need some new ways of gathering information from groups. It’s well documented that open-end responses on traditional Web questionnaires tend to be longer and more detailed. On the back end of this there is a place for automated text analysis programs. I would never propose that software can replace the insights of an experienced researcher, but a lot of work has been done on text processing and now we have the computer power to make it work for market research.
Not: Data file formats that smell of punched cards. The Jacquard Loom was first developed in 1801. There doesn’t seem to be a good reason to be commemorating it in 2004 with data file formats that trace themselves back to punched cards, which in turn were inspired by the Jacquard Loom. Multi-punch was a good idea at the time, but this isn’t the time. Relational databases make good data repositories. There are still all sorts of issues to be resolved over how survey data should be organized for databases and how to do efficient processing of the data, but for data management and process reporting you can’t beat a database.
Hot: Panel management software. There has been a huge growth in the use of panels in research, mainly driven by the Web. This hasn’t been reflected by a growth in the availability of panel management software. Only two companies come to mind for this type of software, Nebu and GMI. What is needed is community management software, systems that track incentives, sampling, usage, panelist responsiveness and communication with the panelists. Instead of new Web interviewing systems, how about more panel management systems? I suspect the issue is the same as with CATI software: it’s hard to develop something that is appealing to enough companies.
Not: CAPI software. CAPI took off in Europe years ago. In the U.S. it has a firm place in social research circles, but for market research in the U.S. it seems stuck. It’s more likely that in-person interviews will now take place via the Web in Wi-Fi hot spots or using GPRS cell phone connections. The problem of CAPI was sample management (hard) and machine costs (high). These days with a Web-enabled cell phone you may well be able to do quite a lot in person for very little cost. CAPI should be hot. The regulations governing cell phones mean that it is illegal to call them at random if the call recipient is charged for the call. Add to this number portability, so that subscribers can switch their land line number to their cell phone, and suddenly in-person interviewing has a whole new appeal. In theory you can obtain a list of numbers that have switched from land lines to cell phones, but the provision of this list is way behind (see www.neustar.com ). And that is without the issues raised for CATI with the new DNC (do not call) list.
Hot: CATI software. It’s certainly true that CATI is no longer in quite the dominant position it was in for data collection a few years ago. Joe Rafael, chairman of Opinion Access Corp., commented at a recent CASRO Data Collection workshop that “CATI, like Rodney Dangerfield, gets no respect.” He’s right. The Web has certainly changed the landscape for interviewing, but CATI isn’t going away anytime soon in the U.S. or the rest of world. The economics of CATI are in flux because of offshore collection (Canada, India, etc.) taking off. CATI is now becoming integrated with Web interviewing, so-called “mixed mode” systems. The logistics of running CATI in remote locations is leading to new software (such as that from CfMC, Sawtooth and Pulse Train) to cope with distributed and mixed-mode CATI. There is even a chance that voice over IP (VOIP) technology will finally, really blend the Web with telephone interviewing. The commercial pressures in CATI are leading to changes in the form of CATI software packages. Sample management systems are now coping with different sorts of modalities. It’s only a matter of time before paper interviewing joins the mix of mixed mode to give respondents a choice of modalities in which to complete questionnaires.
The DNC list is another challenge for CATI (and sample providers). It may seem like it restricts calls even more (whatever the regulations, a lot of DNC list members feel they should be exempt from calls for surveys even if this isn’t the case). There is a chance that the respite from telemarketing calls may make DNC list members more receptive to calls from research companies. Maybe CATI will get its respect after all.
Not: Microsoft. I don’t subscribe to or agree with attacking any specific company, as is the fashion with the detractors of Microsoft. But there are some numbers that indicate that Microsoft is failing to dominate the Web server market. Netcraft (www.netcraft.com ) publishes a monthly study on the software used by Web servers. In December 2003 it tested 45,980,112 Web sites and found that 79 percent of them were not running Microsoft’s Web server products. In 2002 I did a small survey (via Netcraft) of the Web server products being used on the Web sites of the Fortune 100 and the world’s top 25 market research companies. I repeated this survey at the end of 2003. The chart shows the results.
Of course there are corrections applied for availability of information, and mergers, etc. I used the same list for 2003 as for 2002 for consistency. It’s clear that MR companies tend to like Microsoft products for Web servers, whereas the Fortune 100 don’t. It’s also interesting that there has been a small shift to non-MS-based servers in the MR companies (but with a sample of 25 sites, I don’t think too much can be made of that). Extrapolating these figures further is risky, but it must say something about the computing infrastructure of Fortune 100 companies as compared to research companies. Does it matter? Maybe not. But with research companies becoming more involved with software as a way of adding value to their research findings, having some infrastructure in common with large potential clients may not be a bad idea. Talking the same language as your clients on all levels never hurts.
Hot: Using software to make money. Software is a way for research companies to make money. It isn’t just an infrastructure overhead. Using software to develop unique products for the research marketplace is a challenge, but it isn’t impossible. Research companies will always do market research and add value to information, but the mode of delivery to the client of these insights is becoming more important. Excel and PowerPoint were great in the ’80s and early ’90s, but it’s 2004 now. Technologies such as XML (extensible markup language) allow all sorts of information to be packaged in a form that allows clients to integrate it with their business processes far more easily than before. Research companies all want to be seen as an integral part of companies’ management process (repeat business!). Software can help this to happen. Become a part of your client’s desktop!
Post-Web era
The influence of the Web on survey software, now that we are in the post-Web/Internet era, seems to have concentrated on front-end, data acquisition tools. Web-based data/client portals are growing in use, but nowhere near as fast as the front-end Web tools which have gained acceptance. The Web has distorted the survey software marketplace; it’s time for some balance to return. It’s a fact that the back-end tools have been neglected by the software developers to a large extent. Every day unglamorous software like Quantum, Merlin, Mentor and Uncle run thousands and thousands of crosstabulations which still form the core of many research companies’ analysis process. Like the venerable computer language COBOL, these programs are not going to die anytime soon, and there doesn’t seem to be much new around to take their place. An industrial-strength crosstabulation system that can use a relational database intelligently would be really useful.
We’ve got a lot of survey software available compared to 10 or 20 years ago. The question is, is it focused on what the research industry needs or what is easier to develop? Now that some of the hype has subsided from the Web we may see some progress in the neglected areas of need. Market research was an early adopter of computer technology because parts of market research really need software. If you make widgets then software can help you do that, but you probably can get by without software. The challenge for the software vendors focused on market research is to move their products from an overhead and infrastructure role to a revenue generation role. Return on investment is the issue. The challenge for the research companies is to take the opportunity that software provides and use it to make money.