Behavioural targeting it could be the holy grail of online advertising

Publishers and internet services providers, including Guardian Unlimited, Amazon.co.uk, Yahoo!, Tiscali, Channel 4, are using it to generate more revenue from their user base, or to justify charging a premium for advertising in certain sections. And major brand owners such as General Motors UK have used it to track who watches particular ads and then better target future communications to them.

Even Stannah Group, best known for its chairlifts for the elderly, has used behavioural targeting to work out where to place online ads, as part of a push to update its image.

But experts say that many marketers are failing to understand that the term covers two quite different approaches to behavioural targeting currently being used online.

One is very closely linked to website optimisation and tracking the user journey, and, arguably, is a reactive technique usually used within a website, targeting site visitors. It is perhaps best exemplified by its use with so-called abandoned shopping trolleys – those occasions when a visitor to a particular website gets a certain distance through a process (usually online shopping, sometimes filling in forms, for example for an online insurance quote), and then stops.

Tempt them back
In such cases, trailblazing online retailers and service-based sites have found that sending an e-mail to people who have dropped out, perhaps with the offer of an extra discount or some other incentive, can frequently bring them back and encourage them to finish the process – although usually only if systems are in place to remember the information they had already input, so they can effectively take up where they left off.

Similarly, when someone arrives at a site, that site’s systems can start analysing data about what previous site they came from, which ISP they are using, what information they are accessing on the current site and other data they may provide as a result of registering on that site.

As Catriona Campbell, a director of usability and Web analytics firm Foviance, says/ “Sites can tailor ads – or the latest development is they can tailor content. Different people will see different things, depending on what the site knows about them.

A second form of behavioural targeting is being used by ISPs, publishers and by networks of sites: they are pulling together much more data on Web users – mainly through the use of tracking “cookies” – and using it to add value to their advertising sales. So while certain parts of a publisher’s site may not get particularly heavy traffic, if the publisher can show that those people who do visit those parts are more likely to buy particular things or respond to certain stimuli, they should be able to charge more for it.

As Campbell says: “That’s real advertising targeting, based on all the things you have ever looked at.”

This form of behavioural targeting goes beyond tracking what people are doing on one particular site. Instead, it collects as much data as possible about web users, and uses that data to make predictions about likely behaviour.

In a sense, though, the basic concepts behind this version of behavioural targeting are actually nothing new: 20 years ago, lifestyle database companies such as NDL or Claritas held files on the personal habits and purchasing patterns of millions of UK households, and were able to tell marketers with considerable accuracy how likely someone was to buy a particular product, based on comparing data about them with others in the computer banks.

Removing time delays
The big difference, though, is that with today’s behavioural targeting, marketers can analyse what visitors to their sites are doing, possibly even in real time, and react accordingly – sending them a tailored selling message, or delivering ads relevant to their immediate interests, or showing them content related to whatever they are actually researching at that moment. With the lifestyle database companies of 20 years ago, everything was done in periods of weeks or even months – glacial speeds compared with today.

Tiscali, for example, appointed behavioural targeting company Wunderloop in January. Wunderloop has now advised the ISP on how to segment its subscriber base into 150 distinct segments, and advertisers are now being offered the opportunity to target their ads at these segments.

In the US, behavioural targeting network Tacoda claims to have recruited 4,500 sites, reaching 89% of the US population, 50 times a month. In the UK, where Tacoda launched only this month (March 2007), it has only 12 publishers signed up so far. But even so, Paul Goad, Tacoda UK’s managing director, says that it can already reach 59% of the UK population (mainly because so many British web surfers visit US sites which are members of Tacoda’s network on a regular basis).

UK expansion continues
Goad adds that marketers – in particular, big brand marketers – have been reluctant to use online advertising because they have (wrongly) felt it was somehow less accountable than other media such as TV. Goad says that the classic question online’s promoters have been asked is “how can I know I’m getting the audience I want in the scale that I want?” He continues: “These marketers have always had questionmarks about accountability, reach and frequency” – but behavioural targeting networks such as Tacoda’s can deliver advertisers “a pre-qualified audience”. But not everyone who is talking up their ability to deliver behavioural targeting necessarily has the clout to deliver, industry experts warn.

Andy Mitchell, managing director of AdLINK, observes: “Doubleclick developed a form of behavioural targeting ten years ago in its Boomerang product, and ten years later it’s oddly now a ‘new buzz word’. The success of behavioural targeting is not about the technical ability of the product. It is simply about the product’s ‘opportunity to see’, and herein lies the problem. To deliver the full effect of a behavioural targeting product it needs large reach and a large ‘opportunity to see’ across many thousands or millions of unique users. Ad networks are therefore well-placed. Any site or network of sites is well-placed. So take heed of who is promising what and, more importantly, the types of organisation those promises are coming from.”