Using offline data to target and track online ads

By April 15, 2010 One Comment

Back in February I wrote about Yahoo’s deal with Nectar that takes offline loyalty card data and uses it to target ads on the Yahoo network, and then looks back again at the Nectar card data to see if the targeting has any impact.  Yesterday there was more detail on this scheme on the Econsultancy blog, coming from an interview with David Buckingham of Nectar.

The most important new information for me was that only 30,000 people have opted in so far, BUT that they have looked at the surfing behaviour of these 30,000 and found 1-2m others on the Yahoo network who are similar.  If this extrapolation from a small data set to a larger one works and is repeatable with other data sets then we have the beginnings of a scalable model that could dramatically improve the performance of online advertising and hence publisher profitability.

These are big ‘ifs’ though.  As Iain Henderson pointed out in the comments it is easy to see scenarios in which people buy stuff with their Nectar card which isn’t indicative of their normal buying behaviour (e.g. someone comes for their one a year visit and you buy something for them that you wouldn’t otherwise get).  These sorts of problems can be ironed out by the statisticians if the data set is large enough, but if it isn’t then the targeting won’t be much good.

The acid test of course is how the results turn out, and David Buckingham informs us that it is still too early to say on those.  He does promise to release the results as soon as they are available, so I look forward to those.

Thinking about it some more, there is a ton of data out there now which isn’t being used effectively, and even if this initiative doesn’t work I expect that before too long someone will find a way to link views of online ads to offline behaviour and make a lot of money from it.

Reblog this post [with Zemanta]