What is the purpose of web analytics?

There are two schools of thought when it comes to web analytics.
The first is that it is there to improve the performance of your
web site: by analysing drop out rates, the number of people
following cross-selling links, the amount of time spent on the site
and so on then, at least in principle, you can optimise the
performance of your site.

The second school of thought does not disagree with this thesis.
However, it goes further: it views the capture of this data as
providing an important secondary benefit in that it can be used to
collect information about your customer’s preferences. Indeed, if
you think about it for a minute, a company like Tesco can learn
more about its customers that use its web site than those who visit
its supermarkets. For example, on-line you can monitor what people
look at but don’t buy: you can never do that in store.

Rather surprisingly, most web analytics companies have
traditionally belonged to the first school of thought outlined
(perhaps because the latter is arguably more like traditional BI),
though there are signs that some of them are moving towards having
the database and analytics capabilities that are necessary to
support the more customer-centric viewpoint described. In part,
this is not least because the use of Google Analytics is hurting
that end of the market. So, for example, both Omniture and
WebTrends are now positioned within this second school. However,
one company that has always been there is Site Intelligence, which
is a relatively small UK-based company that has, nevertheless,
managed to gain the likes of Tesco, Royal Bank of Scotland,
MyTravel, Expedia (in the UK), SAP, Dyson, MacDonalds, Argos,
B&Q, Boots and a variety of other household names as
customers.

Site Intelligence’s differentiator is that it is database-based:
all clicks are stored as part of a visit, and all visits have
associated visitors that can be analysed from the database. VBIS
(visitor behaviour information system) runs on multiple CPUs and
multiple servers and has been designed (the company has done some
clever things with its database schema) so that you should never
have to aggregate data (so you never lose any information, which
would be the effect of aggregation) and nor do you ever have to
sample it. Site Intelligence claims that it is the only vendor on
the market to support this level of throughput and that all of the
other vendors either have to sample or aggregate when it comes to
very busy sites. Site Intelligence, on the other hand, states that
even its largest customers are able to load all of its day’s data
within four hours, so that relevant reports will be available first
thing in the morning.

Another interesting facet of Site Intelligence is that it is
working with both Business Objects (it licenses its dashboard from
Business Objects) and Netezza. In the latter case, Site
Intelligence has a joint customer with Netezza at Carphone
Warehouse and the company is investigating a partnership with
Netezza in order to maintain its performance advantage over its
competitors.

Site Intelligence is small compared to the likes of Omniture,
SAS or WebTrends but it punches above its weight and is certainly
worth looking at if you are a customer-centric company.