27 February, 2007

Web analytics in the 21st century


I went to a discussion on Web Analytics on Thursday evening at SDSIC. (Definition: Web analytics provides usage & segmentation strategies for web services to serve up more relevant and timely information eg cross selling or provide advertising.)

The issue of profiling customers and customer behavior was front and central, then two minutes later it was followed by privacy and identity.

When an end user / consumer manages their identity in the 21st century meaning of the world, will internet profiling be easier? Read this article on why it's important: Win the keys to the customer kingdom - Retailers will fight for ownership of online user identity.

The knee jerk reaction would be YES, of course - if a user provides a richer set of identity information, the analysis of web traffic and usage would be much easier.

But in reality, the answer yes, to a small extent, but mainly no. Here's why.

Some sites we demand unique personalization: eg logging into MY internet banking account. Other sites become more relevant if the web service knows a little more about us - whether we've visited the site already or that we have an interest in news stories about London for example.

Most of our internet usage is anonymous... which is great for the most part. Sure, we have to surrender our internet (IP) address which can (usually) be tracked down to a pretty precise location (a sub-section of a city for example) using IP lookup - a feature that most web service don't use. In general though, that's it.

Having to create and manage username & password pairs & profiles on every site that we visit would be painful and needless. If you've signed up for a Yahoo! account, then the profiling is ridiculous - websites that provide valud usernames+passwords such as bugmenot become very useful.

21st century ID solutions such as Microsoft's CardSpace (check out an overview of Cardspace in use - technical, but good) will make identification easier - users will create several identity cards for different purposes. Those identities with lots of information accessible by the service provider will be used with highly personalised sites (eg financial institutions). Users will also create a range of weak identities that allows them to glide easily onto websites.

These sites will be unwise to needlessly prompt users for their 'Identity Cards' - imagine being prompted for your household salary or whether you have children if you surfed onto a news site - it would be even more irritating than pop-ups.

So users will remain largely anonymous, unless the site can coax an identity out of a surfer in return for a better experience (hmm, marginal) or other benefits (eg 5% off price).

As a result, web analytics will divide into analysis on those with identity and those without - the latter requiring all the skills that are required to match usage to segmentation.

No comments: