Web analytics

Gone are the days when the number of hits was the be all and end all of web analytics. Publishers are now equally interested in who their visitors are, what they do on their site, how long they do it for, and how often they visit. Andrea Kirkby looks at the current trends.

By Andrea Kirkby

Print media have grown their web presences over the past five years. There is more content online, and there are more users accessing that content. But what do we know about those users?

Five years ago, even the number of users was debatable. Measurement was not an exact science and there were few agreed standards. Alexandra White at the Association of Online Publishers says this just wasn’t tenable - "we’ve sold ourselves so much as a measurable medium" that auditable figures were vital for credibility.

And the industry has now got its act together, as Richard Foan at ABC Electronic points out - "we’ve agreed some basic definitions and we have them agreed globally." Many of the web analytics packages now build these definitions into their tools to deliver ABC compliant data automatically. However, just knowing the number of users is not enough. Richard Withey says that "people get hung up on pure traffic numbers far too much. It’s what they do on your site and why they’re there that matters. For a long time advertisers were seduced by the idea that you buy the hits and you’ve done your job" - which doesn’t reward the editor for creating good content and a higher quality target audience.

Panel or bespoke

So generating better information on exactly who the web users are, and what they’re doing, is vital both for content creation and for advertising. Fortunately a number of different methods are now available to the publisher. Alex White divides these into two main areas - panel based research, which can be generic (eg the Nielsen/Netratings statistics) or bespoke (like the AOP’s latest research into the ‘online elite’), and secondly, monitoring software such as Webtrends, Nedstat, or Sagemetrics which tracks actual user behaviour on the site. Alex says both are complementary. "People do both hand in hand - they want to get the level of detail measuring their own site, but also to see how they’re doing against their competitors, which they can do with a panel."

Much of the data that comes from monitoring software is anonymous. Bill Murray, at Haymarket, points out that an identifying return IP address doesn’t tell you who is behind it; so site registration can deepen the publisher’s knowledge of the user. But, he says, you have to offer something worthwhile to encourage registration - the magazine’s relationship of trust with the user must be maintained. Tying together the web stats with offline information can take this even further - Julien van Dommelen at Sagemetrics says that postcode information can now be integrated with the web data for further demographic analysis of users.

Compared with a few years ago, the software is now well developed and reasonably affordable. Nedstat’s Sitestat product, for instance, costs £2,750 a year, with most customers using it as an ASP (application service provider) solution. Richard Foan sees more and more publishers using third party software rather than building their own systems, owing to "the continuous pace of change in how the websites are delivered, for instance rich media, and the investment going in from third parties in developing analytic tools." Only the largest media groups can afford the R&D needed to run their own tools.

The amount of information these tools can give is extensive. Where did users come from? Where are they located? What are they doing while on the site, and how long do they stay? Are they coming back within a day, a week, a month? Publishers can evaluate their web audience’s loyalty, not just its size - "who’s a browser, who’s a grazer?" as Richard Withey says.

Using the data

What use can be made of all this data? Obviously it is useful ammunition for selling advertisement space, but it is also of use to the editor both in the design of the site, and in evaluating content. Linda McDougal at Nedstat points out that the software can now follow users all the way through the site and show the paths they take to information. "Through web analytics," she says, "publishers can understand stumbling blocks on the site, and take action to increase the number of visitors and the number of quality visitors," for instance by removing dead ends. And, she says, if publishers want users to take particular action, for instance registering or paying for content, web analysis can help design the site navigation to make that target easier to achieve. Paid for content will make analytics even more important, she says. "Web analytics gives you an understanding of what kind of content is quality - what do they come back and look at again?"

And just from a cost point of view, Bill Murray says, web analysis can help to "declutter" a site - de-emphasising content that is not finding an audience. Although most content is free, he believes the industry will move towards paid for content, and, he says, it’s analysis that will lead them towards it. Focus groups are now the way of the past as far as he’s concerned - "they’re bloody expensive and hard work, and web analysis is cheaper and better."

Linda points out that seeing where users have come from is highly useful in marketing the site to users. Publishers can see not only where users are coming from, but also how many users referred from Google compared to users coming from Yahoo - how long do they spend on the site, and how likely are they to register?

What advertisers want

Julien van Dommelen , sales director, Europe for Sagemetrics, though, makes an interesting point when he says that news and sport drive people on to a newspaper site, but isn't relevant to advertisers. "Sport, news and entertainment is low value content. What advertisers are making money out of is you going to the jobs section, money section, property section, travel." Web analytics can help generate a strategy to push people into these sections, but, he says, publishers can then use the analysis of user behaviour to serve ads on a personalised basis to users who have displayed an interest in a certain section, even if they have navigated back to the home page or to another section of the site. "It’s a question of monetising your website."

There have been a few surprises for his clients. One middle eastern news site majored on insightful political analysis, but the web stats showed that everyone was logging on to pictures of ladies in bikinis in the entertainment section. Julien says that "the editor kind of laughed, but I don’t think he really thought it was funny."

Shaping content

In Julien’s world view, content is only worthwhile if you can sell it to advertisers, and the web analytics will tell you how much it’s worth. That’s not a view that goes down well with editors. While Richard Withey finds the detailed analysis of web usage interesting, he has real reservations about using it to dictate content. "We wouldn't try to drive editorial policy through the website," he says - the editor’s job is to create a tone and direction which isn't susceptible to analysis at that level.

Bill Murray, on the other hand, admits that traditional editors may have to learn a little humility in the face of web stats. He says the web information is "hugely valuable, because it lets us build better content, and better targeted content". For instance, What Car used web analysis to decide which video test drives should be done first - and surprisingly, the list of most accessed reviews wasn’t the same as the top ten selling cars. "VW Golf and Polo don’t sell anything like as much as the Ford Focus does, but the road tests are looked at just as much. I’m sure VW’s marketing department would be interested in that!"

Bill also believes that web analytics are a potent tool for the print side of the business. "The information starts to give you a really good steer for your editorial content - not just for the web but also for print." Richard Withey bears out this point that the two constituencies are similar - "we have found very strong correlations between what we know about Independent readers in print and what we know about Independent readers online." That’s borne out by the AOP’s research, too; the web does attract some new users, but the characteristics of offline and online readers are highly similar.

The future

Where is the industry headed? According to Linda McDougal, it’s not the technology that is holding publishers back - in fact, publishers still don’t use half the data they get, and use only 50 percent of the functionality of the software. She says the real development will be in the use that publishers make of the data - "web analytics has become much more part of the business process, not a bit of technology, and that development will continue."

One functionality that may find more innovative uses is behavioural clustering. It already exists, but it’s mainly used for ad serving. Julien van Dommelen believes it has the potential to create true personalisation - putting the right ads on the page. But Tim Brown at 24/7 Real Media takes this further - "at the end of the day, whether it’s advertising or editorial, what we’re doing is enabling people to put relevant content in front of the user." While he believes web pages that require the user to ‘build’ their own news preferences have been poorly used, with this technology users don’t have to do anything - news content will automatically be served based on the user’s previous behaviour. That’s particularly good news for specialist sections of publications, often hidden deep in a site - but potentially, now, on the front page for that percentage of users who want them.

What’s certain is that the web analytics area will continue to grow and develop extremely fast. As Bill Murray says, "this is very definitely an evolving area, and my feeling is we’re closer to the beginning of it than we are to the end - we’re just beginning to scratch the surface".