by G3

The concept of behavioural targeting in search is to observe a user’s online search behaviour anonymously and then serve the most relevant ads based on their behaviour. Theoretically, this helps advertisers deliver their online ads to the users who are most likely to be receptive to their messages.

Implications

There are growing forms of internet advertising vying for advertising budgets increasing accountability in terms of return on investment (ROI) and cost per acquisition (CPA). To determine where the budget is spent, marketers are looking for more ways to target their advertising to customers who have a high propensity to convert. In theory, behavioural targeting fits this criteria and can increase the effectiveness of campaigns by reaching the right target audience online.

It is possible to build consumer profiles based on the users’ demographics, geographic location, lifestyle and affinities – gathered in search engines by cookies (a small text file on the user’s PC that identifies the user’s browser so a user can be ‘recognised’ when they re-visit a site). By using this data advertisers can ensure adverts in search engines are more relevant to what the user has previously shown an interest in and the interests of other users with similar profiles. Another method is done live during a search session. For example, in Google it is possible to search for your town, and then carry out an entirely new search for a restaurant without the location, and the search engine remembers the previous search to refine results specifically to your town.

A further step

Another step in behavioural targeting, is using data gained from customer relationship management (CRM) systems within websites to present visitors with the products they’re most likely to purchase. This is done while a customer is searching for products; the time when they have a high propensity to purchase and using ad copy that is in tune with their needs. Once customers have selected products and have them in their basket, behavioural trends can be used to present optimised cross-sell (similar) and up-sell (higher price) products. Travel and retail sites can use this to a particularly beneficial end by monitoring the behaviour of anonymous customers and establishing behavioural patterns. This often precedes bookings of different types of products, before monitoring these trends among future customers to display appropriate products which are likely to be attractive and fulfil their needs.

How to use behavioural targeting in search

Behavioural targeting is available to a certain extent in all search products, through geotargeting options and the choice of sites where adverts should appear on content sites. MSN goes one step further for their search and display advertising; advertisers can incrementally bid for location, day, time, age and gender, based on historic trends of types of visitors who have shown an interest in your site and products.

Google’s personalised homepages gather increasing amounts of data on user behaviour and the recent implementation of an ad quality initiative shows how this search trend data can be used. If you frequently search on certain keywords in Google, you may start seeing ads in different positions on the page. This is done in an attempt to match the ads with your previous behaviour. The theory behind it is that if you have done lots of searches for the same keywords, without displaying normal user behaviour (i.e. you haven’t clicked on any ads), the ads displayed can’t have been relevant to you and therefore Google will show you different ads to improve your user experience.

Case study

This approach has had proven success for Eurostar, when they used MSN AdCenter’s geographic, demographic and day-parting tools to focus on their key target audience during a traditionally low sales period. This saw an increase in the conversion rate on the previous month’s figures (MSN case study, 2007).

Caveats

As with all these tools, behavioural targeting is only as effective as the data which is fed into it. There are limitations, including the dangers of relying on past trends to predict future behaviour and also how users’ characteristics are determined and linked back to the individual consumer.