±±¾©Èü³µpk10

Research is a key weapon in the marketer's armoury, but interactive and digital media offer so many more opportunities to collect data and information.

Accountability has always been held up as a key virtue of the internet and, certainly, online marketers have often suffered from being bombarded with too much information, rather than too little. The question is: which data is useful to your company and its activities, and what isn't?

At the most basic level, the internet appears to be growing up as a research medium. Bob Ivins, research director at Yahoo! Europe and a board member of the European Interactive Advertising Association (EIAA), points out that researchers have had to make some compromises with traditional media. Of the three basic parameters of research - accuracy, speed and expense - it is difficult to hit two of them on any given project, let alone all three. But this scenario has changed thanks to the web.

Fast and accurate

"If you want to be accurate using traditional methods, it is hard to be quick and cheap. The internet changes this, however. And we can now be accurate, fast and very cheap using online research methods," Ivins points out.

Another difference, notes Chris Russell, partner at eDigitalResearch, is that with traditional research, particularly questionnaires, the costs are incurred in collection. As online surveys have no collection costs, resources can be concentrated on making the questionnaire better and more efficient.

One big change in the last few years, which backs up Ivins' claim, is that the number of users in major markets who are using the internet has increased, making the online population more representative of the population as a whole.

"The brush with which the internet was painted always used to be that it was not representative of the total population," Ivins recounts. "However, 50 per cent or more of the population is now online in markets such as the US, the UK and northern Europe. If the online population is representative of the overall population then online research becomes faster, cheaper and more accurate."

Ivins sees this message coming through as the advertisers and businesses not only conduct more research on the internet but become more serious about it as well.

"We are now at a tipping point. More and more people are getting serious about online research. A couple of years ago it was about experimentation. Now they are seeing the results borne out," adds Ivins.

This sense of scale and the make-up of the online population as we approach 2004 is key.

With panel-based research, we are seeing higher numbers of people using the internet. This gives the medium more possibilities and credibility as a research tool, points out Arno Hummerston, head of interactive solutions worldwide at research giant TNS. "Initially, the challenge was to get a panel that was representative of a 'normal' population or specific target segment. But that has changed," he continues. "As more and more people have come online, more possibilities have opened up - in technology terms, we could have done this years ago."

This scale means that a company looking for a panel of doctors, for example, would find it easier to generate a group of them because, when you have hundreds of thousands of members of the public using the internet, a sizeable proportion will also be members of a particular profession.

Track awareness

Brands and businesses carrying out research online come from a range of sectors, such as FMCG, automobile and financial services, to name a few. Ideas such as brand-tracking studies, which are popular areas for research offline, are now moving online, observes Ivins. "It is about moving this online to track awareness of sales, and it can be done both faster and cheaper online," he adds.

Online research can be behavioural and attitudinal, which refers to, firstly, information that can be glean-ed from observing visitor behaviour and activity and, secondly, feedback from simply asking those users what they want, feel and intend to do.

Basic forms of behavioural research can come from server-log analysis. This research, and the results it can give, has become more specific and detailed as technology has progressed.

A boon to Yahoo! advertisers, Ivins believes, is the detail of reach and frequency that the portal can provide through its inventory management system.

Other media, from television to radio and print, have industry standardised or 'accepted' measurement systems. This can cause some controversy, particularly in the case of measuring radio audiences for different stations but, ironically for a medium that is hailed as the most accountable, there are many different ways of measuring traffic on the internet.

"I don't think that there is just one data source that can give us the one answer online," says Ivins. "Unfortunately, the internet industry painted itself into a corner by saying 'the medium is the message' and how much information could be found out. We've got better, but there is no silver bullet or one solution. We should use all the data that is available to us."

One progression with web analytics software is that it is no longer solely the preserve of IT. According to companies like IT services firm Detica, the sector is now entering the real world of business and marketing. This means that the information gathered is useful and, increasingly, digestible.

The most basic form of information is weblog analysis, points out Imam Hoque, head of information interaction at Detica. But research tools and solutions are moving beyond this to take data a step further. Hoque cites Detica's client, dialaphone.

For the mobile phone retailer, its web analytics are all about saving money on advertising and making it more cost-effective by working out what is effective. Dialaphone can use its online research to change or cancel ads that are not working in a day's time. "You can report on the user's journey with weblog analysis and from the point of arrival to the actual sale," says Hoque. "You can see where they are coming from."

Hoque says the lessons being learnt are important as they are effectively channel-neutral. "Not only can you affect advertising on the web side, but, generally, there isn't too much difference between its cust-omers on different channels," he adds. Detica uses client-side data-collection and analysis tool Prophet, effectively a service run from a separate server.

Segment visitors

Other users of Prophet include PC World (www.pcwb.com). According to e-business supplier speed-trap, which provides the tool, Prophet implements web measurement and user-journey analysis to track and segment visitors to its site. Simon Burton, sales director of speed-trap, explains: "Unlike other analytics tools, such as those based on web-server logs, Prophet works by observing events in real time in the user's browser, rather than on the server. This enables us to target what really matters - monitoring the visitors and customers, not the technology."

He claims: "We can provide more accurate data, such as click-maps showing where users are most likely to click through on web pages, mouse trails charting users' scrolling and reading habits, plus hotspots indicating the most viewed areas, the end result being a visitor-friendly, more profitable site."

According to Detica, by using such tools as part of an objective measurement exercise, managers can quickly identify areas in need of improvement that will directly effect return on their web investment.

Even as technology becomes more sophisticated and allows marketers to find out more information by monitoring or tracking users, eDigital Research's Chris Russell is a firm believer in simply asking users what they want to do online and where they are actually encountering any problems. "We haven't actually found anything that they won't want to comment on yet," he remarks. "Even with unsolicited research, people are usually very happy to give their opinion."

Naturally, this system works well with e-tailers and shopping sites because they involve the site visitor in a linear process and are prepared to be active. However, you must be careful not to interrupt the flow. Simply asking the user to opt-in for a questionnaire at a later date can also be effective.

But users can be just, if not more, responsive on non-retail, especially if the site occupies a special-ised niche and trades on the quality of its content. For instance, online newspapers have particularly responsive users, notes Russell. "They are very tactile and give interesting replies, but it is different with more general sites, like portals, where users are effectively going through doors to get somewhere else. They are not interested in stopping and you have to target them on the right parts of the site."

To help improve a business's online ROI, companies can find out more from their users' view of the homepage and their first impressions, and functions such as search and browsing. "It is all about ensuring that the user has enough of a 'good feeling' to move on to the next stage," points out Russell.

One of the big problems facing e-tailers, and something they are continually seeking an answer to, is the number of online shopping baskets which are being abandoned by potential customers before they have completed a transaction.

"Retailers should understand the fact that the digital consumer is very promiscuous," insists Russell. "Consumers look for a lot of information on the internet and may go to, say, five or so different web sites, fill up their baskets with the same products and then print them out. So, the shopping baskets are often used as wish-lists," he adds.

If you don't know what the users are doing and where they are going, as long as you do it in the right way, you can always ask them, adds Russell.

User reactions can also be very useful when testing out other ideas such as advertising and creative. TNS has developed an advertising evaluation tool, called AdEval, which highlights how good digital channels are in evaluating visual information and questions. Users look at ads on their PC and AdEval asks them for their reactions (indicated by mouse movement) and how these change when they are looking at longer pieces of creative, such as TV advertising.

Responses can be viewed in four or five days, says Hummerston, and this provides a good pre-test bed. "Anyone from electronics or car firms to the armed forces has used this method," says Russell. "It can be comparative, so you can compare the respective strengths of three different ads." Although TV advertising is the most common creative that is tested, it can also be used for radio and press advertising.

Careful management

Asking the user, of course, is the basis of panel-based measurement. This can be a powerful tool, particularly with the speed of online panels and their responses. But Hummerston preaches caution to both marketers and researchers.

He believes the context of a survey needs to be appreciated and understood.

"In political polling, in particular, very, very careful context management has to be applied, and that should be applied to any polling for politics," adds Hummerston.

Context also applies to other channels. It is not confined to internet research. After all, points out Hummerston, if you embark on telephone polling during the daytime, you need to consider the demographic of the person who will be answering the telephone at home at that time. "However, people are used to dealing with telephone data now. And, possibly, for that reason, online data has come under more scrutiny," he continues.

One possible drawback with internet research is the fact that it is perceived to be very quick, but this is not always the case, warns Hummerston. "This can be partly because you have less control over the sampling and response rate than with other methods," he says. So, if you send out a certain number of questionnaires, you might not get the numbers you want and may have to embark on the project again.

"There is also the operational side of this - it is no quicker to create an online questionnaire than an offline one," he says.

Hummerston, Ivins and Russell all stress the differences between various geographical markets. For instance, the online population of places in southern Europe is not as representative of the overall population in those countries as it is in the UK and US. Also, due to the 'clutter' of messages and ideas online in the US, the response rates to marketing questionnaires may not be as high.

Even in the UK, the experts preach caution when asking for user opinions. Russell warns: "In the UK, we have to be very careful about what we do. You don't want to approach the user at the wrong time in the wrong place."

Thanks to our panel of experts, Imam Hoque, head of information interaction at Detica; and Chris Handford, director of research, and George Davidson, qualitative research director, at RedSheriff.

MASTERCLASS PANEL

Bob Ivins is research director at Yahoo! Europe and a board member of the European Interactive Advertising Association, which researches the effect of online advertising. He has more than 20 years' experience in the information and consulting sectors.

Arno Hummerston is head of interactive solutions worldwide at research giant TNS, previously Taylor Nelson Sofres. TNS has offices in Europe, the Americas and Asia Pacific, and provides custom research and analysis in more than 110 countries.

Chris Russell is a partner at eDigitalResearch, which offers online market research, specialising in web-based tools and data. EDR clients include Comet, B&Q and AOL. It measures customer opinions, satisfaction and loyalty, with the aim of improving ROI.

DTI USES FINDINGS OF USABILITY RESEARCH FOR SITE REDESIGN

The Department of Trade and Industry (DTI) promotes business at home and abroad, so its site (www.dti. gov.uk) reaches a diverse audience.

This year, it gave the green light to a site redesign, forming a Web Transformation Team (WTT) with several agencies.

The WTT needed to understand current usage of, and attitudes to, the DTI site, says George Davidson, qualitative research director at RedSheriff, part of the WTT.

RedSheriff was asked to find out how the current audience uses it. It took a 'multi-pronged' approach, involving browser-based measurement of site traffic; group discussions with current and potential users, including SMEs, large businesses, consumers, and other opinion-leading stakeholders; a pop-up survey on the DTI site to identify user segments and understand their use of and satisfaction with the site.

It also used RedSheriff Insight, a spider that crawls the site to enable the DTI to audit for pages available to users plus broken links, consistency of branding and the depth of clicks.

This combined approach delivered recommendations on delivering a site to such a diverse range of audiences, says Davidson. "The DTI is redesigning its site from a strategic perspective, with an impressively holistic approach," he says.

"Given the extraordinary depth and breadth of its activities, the DTI has a difficult task in redesigning its site to meet these diverse needs, but we are confident the information and insights we gathered will help in the process of launching the new web site in 2004," he adds.

Rupert Marsh, head of internet and new media at the DTI, says: "These days, it is imperative that government web sites are customer orientated, but often their needs are lost in the design process.

"We wanted thorough knowledge of our customers and their requirements before thinking about the design. We referred back to the results of the research throughout the process to keep us on track." TOP TIPS ON ONLINE RESEARCH

1. Make the invitation to take part in the survey prominent (for example, use pop-up windows and email invitations) to get a higher response.

2. By tracking user behaviour from entry to exit, and surveying them when they leave your site, you'll find out who they are, why they visited, their perceptions (from the survey) and the ability to analyse results by visitor behaviour (web metrics).

3. A pop-up survey should take no longer than 10 minutes to complete, or 15 minutes if it is via an email invitation or online panel.

4. To avoid respondents dropping out of the survey mid-way or towards the end, let them know how many more questions they have to complete via a progress indicator.

5. Launch any surveys from Tuesday to Thursday. On Mondays users may be too busy. At the weekend they might not check their email.

6. To get a representative sample (for example, weekday and weekend visitors) ensure the fieldwork period covers at least one full week.

7. Brand the survey clearly with the host site's logo, look and feel. Any third-party branded pop-up survey is unlikely to be as successful.

8. Don't think of surveys as just text-based. You use all the media available on the internet.

9. Cookie the people you invite to participate. This ensures they don't complete the survey more than once and they won't be annoyed by receiving your invite more than once.

10. It's not just about percentages. Setting up a small customer research panel to talk about related topics via a bulletin board is a cost-effective means of gaining fast, in-depth feedback.

- Chris Handford, director of research, RedSheriff.

CHECKLIST

Questions that should be asked before running online research

- Is what you find out appropriate for a survey?

- Where can you find appropriate users?

- Can you simply ask users to answer the question you want them to?

- Are you approaching users at the right time in the right place?

- Have you positioned your survey on the correct part of the web site?

- Are you trying to interrupt users when they are trying to complete a task or a search online?

- Is your survey short and to the point?

- Can you ask users to opt-in for receiving the survey and present it to them at a later date?

- Have you adopted a long-term 'little and often' approach, so not to bombard consumers?

- Have you forgotten your customers simply because a transaction has been completed?

- Do you have offline stores or outlets that you want to ask your consumers about?