Article #15: Web Bureau, ITE/Web and Media/Web

Tim Macer reviews new products for surveys via the Internet and hears from their first users

Keeping control of

Open a newspaper or a business magazine, and you are certain to find an article about the Internet. It has become the copywriter's faithful standby: always something new to cover, and hardly anyone to contradict you. It was enough to make me avoid the topic, until now. For, during February, three new Internet products for quantitative market research community were launched: one at the "front end" and two at the "back".While each of these products seems well crafted, intuitive and thoroughly effective software solutions, you cannot help wondering if they are solutions looking for a problem. On the one hand, with usage penetration of the web barely touching the 5% level in the UK, surveying on the web is like trying to imagine doing a telephone poll in the year 1904. On the other, developing a tool to allow you to publish commercially sensitive data for friends and enemies alike to view on the Internet sounds like a strange way to make a living. Is this the niche carved out for ISPC's new ite/web tables? What if you then give it away for free, as do Canadian developers Speedware, with their public domain on-line analytical processing tool (or OLAP) called Media/Web. What kind of madness is this?

Quantime's new Web Bureau is using a special Internet version of their highly successful Quancept CATI product to put surveys on the woldwide web . Richard Kottler at Quantime identified two different sectors where research on the net seems viable now: hi-tech, where there is already much higher penetration of Internet usage, and surveys about products and services available on the Internet itself. He told Research "You can either publicise the URL [the web site's unique name] and have a free-for all (then you will have some interesting sampling issues) or you can recruit people by phone and ask them to go to the web site to complete the survey, so that you control the representative nature of the sample."

Richard Kottler was keen to impress me with the economics of doing a survey on the web. They recently performed a test during which they achieved 1,950 simultaneous interviews. It took just one person to perform this survey; costs come down to just hardware and telecoms. "It is all set up in Quancept, then converted automatically to HTML [the language used to create the pages that appear on the Internet]. Clients can write the scripts themselves if they wish in Quancept then we will convert it and put it on the web for them." Unlike previous attempts to do surveys on the web, where respondents essentially fill in a form on screen, using the CATI/CAPI approach means that questions are presented one by one, in a controlled sequence.

Fiona Warren, Associate Director at BMRB International was just putting together their first Quancept Internet survey when I spoke to her. It was about, not surprisingly, the Internet. "We have been asking people in our normal surveys if they have an e-mail address and Internet access. We will then contact them by e-mail and invite them to visit the site so that we are effectively controlling the sample. We don't get a high percentage of e-mail addresses, so we are not expecting many responses. But once you get it going it is very cheap to run, and quick as well. Comparing it to paper, one week with a postal survey is like a day with the web. For example if you expect 85% of your target after two weeks [by post] you would expect this after two days on the web." She felt that points to watch, when designing a survey for the web were to keep it short and interesting for the respondent, to remember that Internet users tend to be more sophisticated consumers and should not be patronised, and to anticipate the complete range of responses to each question. Unlike with a printed questionnaire, but just like CATI or CAPI, if you do not leave space for all the answers, including "None of these" or "Don't know", respondents will get stuck and will not be able to complete the questionnaire.

Nicky Perrot is head of Market Research and Analysis for Reuters European, African and Middle Eastern operations. They have followed the second route when successfully conducting research using both the Internet and their Intranet (their own private web) by providing links so that people using the web will just drop into their survey. They have recently tried Quancept, and were pleased with the results. "It's very cheap, definitely user friendly, and respondents are impressed. It is quite flattering when the survey has obviously taken your answers into account, such as with row masking on a list of answers". Nicky felt that open-ended questions were a bit too laborious at present-something that Quantime are addressing. But that the real problem is in ensuring enough people find the surveys to complete them. Other people must publicise your survey and if necessary, you must pay them to do this. In Reuters' case, representitiveness is not an issue. Because they are interviewing people about the products they are using on the web, it makes sense to use the web to interview them. When we spoke, Nicky was putting the finishing touches to the paper she was presenting at the MRS Annual Conference on the subject of Internet surveys. She had nothing but praise for Quantime and the support they had provided to her when using their Web Bureau service.

The ITE/web is the logical extension to ISPC's ITE electronic fiche. It essentially allows you to call up and sift through large volumes of pre-produced tables in almost any format. Once again, it was at BMRB International that I found someone experimenting with this tool. DP head, Phil Torode explained how they had tried using it on a survey where the topline report alone ran to 500 pages. They took the normal output file, and directed it towards ITE/web. This then formats it automatically, creating an index, table of contents and all the various links on screen that web users expect. "It is very, very good", Phil told me. "Speed is the essence. When people talk of speed they often think in terms of delivering the results as fast as possible. But it is the speed of access the information once you have got it that is more important. Here, something that is key for the end user is that they do not need to have any specialist software. If you give them Quanvert or Pulsar, they need a license and they need to learn how to use it. Give them this and they can do what they want". This is because the software they are using is a standard web browser such as Netscape or Microsoft's Internet Explorer. This also gives instant multi-platform support: UNIX, Macintosh and OS/2 users are no longer left out in the cold. If these tools succeed, there niche will not be on the Internet, but on private web-based servers called Intranets.

There are several particularly appealing aspects to ITE/web. Set-up time is negligible, unlike most web publishing software. An independent set of style templates allows you to add font, colour or style changes to different headings and even particular words and phrases, add logos, pictures and graphics to pages. You can do all of this without being learning the black art of HTML programming. By using Netscape or Explorer, it unlocks some very powerful features. For instance, while browsing Internet or Intranet, you can click on a button to view the file as spreadsheet. This will then fire up your copy of Excel and present it to you as a spreadsheet for you to move into charts, or reformat for inclusion in a report.

Speedware's Media/Web is available free to web site developers, and also promises simple set-up without using HTML, and the advantages using web browsers in place of specialist software to view the results. Sue Lingard at Speedware told me that it takes about 15 minutes to set up a file and "publish" it using Media/Web. "It is completely dynamic. You can pre-define the level that people view the data, or cut off at any level". Unlike ITE, Speedware sits between the end user and a database full of raw data. As you request pages of information, it crunches the data, creates the HTML and presents it on screen in a wide variety of pretty formats, from plain old tables to very snazzy charts. It is not principally a cross-tab engine, but more a "drill-down" tool, and is likely to be of much more appeal to those processing large files of simple records such as media or retail data, rather than survey data. And all this is free of charge. The catch? The version that runs on an Intranet, and presumably the one that will make money for people, is the version you must pay for.

For once, you do not have to take my word for it. If you have access to the Internet, you can enter the web site addresses listed below, and look for the link to "live demo" at each site. For the resolutely unconnected, telephone numbers are provided too.

Published in Research, the magazine of the Market Research Society, April 1997, Issue number 371.

© Copyright Market Research Society/Tim Macer 1997. All rights reserved. Reproduced with permission.

Top | Index