<
Event Summary
SAS
Institute, Inc., the largest privately held software company, introduced
a family of "E-intelligence" products that provide analysis of the data
collected at a website. Although the announcement covers three products,
the SAS strategy is to present a solution that encompasses all of the
data mining needs in a large enterprise.
The
WebHound product is a high-end clickstream analysis product. It
analyzes the paths taken by customers viewing a website and can both produce
standard traffic reports and answer ad hoc questions about the behavior
of individuals or groups.
E-Discovery
is a data-mining product that can integrate information about Web behavior
with other information from across the enterprise. This product uses a
number of data mining techniques such as collaborative filtering and neural
networks, so that in addition to answering specific questions it can point
out patterns that might not have been recognized or asked about.
The
third member of the suite is IT Service Vision, a previously existing
product that can report on server performance and problems.
The
three products are based on existing SAS products to which the ability
to parse and analyze web logs has been added. The core of the company's
strategy to offer what they call a "closed-loop E-infrastructure" is E-Discovery
that has the data mining capabilities of existing SAS products such as
Entreprise Miner. E-Discovery can be used withWebHound or other log analysis
programs to find patterns in the data that are not identifiable from logs
alone. Such patterns might be that users who visit pages A and B also
tend to visit pages C and D, or that users who purchase baby oil and are
shown ads for diapers are likely to purchase both diapers and intimate
apparel. E-Discovery can take into account the historical traffic patterns
and any demographic information a site has; it can return real time characterizations
of surfers that can be used to personalize pages or target ads.
According
to the company its three key differentiators - integration capability,
the ability to deal with multi-channel data, and scalability to arbitrarily
large data collections - are fully reflected in these E-intelligence products.
Market
Impact
The
company has little interest in selling WebHound competitively against
WebTrends or similar tools. The smallest sale that would be of interest
to them is one that bundled WebHound with E-Discovery. A product like
this would compete with other web-base data mining products such as those
from Accrue and NetGenesis. We do not believe that there will be a strong
marketing push aimed at smaller companies for which a product that analyzes
only web data is sufficient. In particular, we don't expect that the typical
dot.com will be visited by a SAS salesperson.
The
company is instead looking to sell the product into clicks-and-mortar
companies that have a variety of data channels that must be combined for
analysis. The target market is an enterprise-wide sale. E-Discovery will
be an easy sell into existing SAS customers who are already familiar with
the company's abilities with managing and analyzing large quantities of
disparate data. Success there will fuel success in a wider sales effort.
Of
course, SAS's competitors - SPSS, MathSoft, and especially IBM - have
or will soon have their own competitive entries.
User
Recommendations
We believe that identifying user trends and behaviors is one of the absolutely
most critical things that a website must do to survive. If you don't know
exactly what kinds of data you want to collect and how to use it, then
any of a family of similar products will almost certainly be appropriate
- and much more cost effective than the hand-crafted system that your
programming staff has offered to whip up.
As
we noted, there is no compelling reason to look at WebHound if your only
interest is daily log analysis. If you intend to mine current and historical
web data then the combination of WebHound and E-Discovery may be worth
a look. Note that there are significant feature differences between products
that are intended to mine data from web logs, starting with source data.
SAS
and WebTrends take data from web logs, whereas products from Accrue and
NetGenesis can capture data at a deeper network level. Some sites, more
typically the technically savvy dot-coms will appreciate the advantages
of this, such as showing when people cancel the loading of a page, something
that cannot be determined from web logs. Some products have add-on features;
NetGenesis, for example, has one that is optimized for sites that use
shopping carts.
We therefore strongly recommend a very careful selection procedure to
identify all the ways that clickstream data can be used, either by itself
or in conjunction with other enterprise data. Unfortunately, the richness
of the data that are available means that the kind of traditional thinking
that marketing and IT people bring from their past experiences is not
a good guide for a forward-looking set of requirements. Unless someone
on the selection team has experience with web traffic analysis it would
be a good investment to bring in some consulting assistance for requirements
definition before purchasing any analysis product.
The
companies that should definitely take notice of this are the ones that
need to integrate many other streams with their web data. For these companies
SAS's track record makes putting the E-intelligence suite on their short
list an absolute necessity.