User Tracking Discussion
Kevin Vaughan
analyze the statistics of their site usage and apply that information in
changes and updates to site hierarchy and the way their content/products are
delivered. For example, Company X sees that 75% of last weeks hits were
related to generators, so they push more generator-related links to all of
the users. This is an effective means to periodically manage your overall
userbase's needs, but should we address our individuals needs with as much
diligence as we put in making our pages accessible to everyone? After all,
if we want to draw more potential customers to our sites, shouldn't we also
want to take great care in satisfying the needs of our current customers?
Complete tracking and realtime statistical analysis to actively present
information more relavent to the user seems almost too obvious an answer.
There are downfalls to this as well, your users concerns about privacy are
probably at the forefront of those. The only extent of difficulty in the
project is the decisions you make about analyzing the data - thorough
analysis requires thorough work, but some basic statistics could be derived
quickly. What do you think about the sites you frequently visit analyzing
your usage and modifying its content to better suit you? Do you think this
is something your company or your customers would be interested in, or are
they already using a simlar system? What do you feel are the positive and
negative aspects of user tracking?
--
Kevin Vaughan
ktva...@unity.ncsu.edu
Jacqui caren
@typhoon.southeast.rr.com>:
>hat do you think about the sites you frequently visit analyzing
>your usage and modifying its content to better suit you?
I think that they are still in the stone age.
Most web sites (not all) consist of content accessed via
a number of navigation systems.
what you are proposing is that a site be restructured
to better fit the most popular visitor trends.
would it not be better to have a site that allows a
user to navigate in whatever manner they prefer
with the ability to tailor every content and navigation
page to bestfit that users personal profile?
I detest the idea of using site analysis tools to try and
tailor a site to the most comon denominator. It is very
similar to the argument by some browser officianados that
since they redesigned thier site to look better for browser
X because they got a lot of hits from browser X. They now
get a lot more hits form browser X so that must be good :-)
The fact that not all of thier target audience may use browser
X with all of th features they use may lead to a lot of lost
customers seems to be irrelevant to such folks.
I think stats analysis tools are a waste of time for this purpose
for a large number of reasons.
I prefer to think that some decent maket research and a well
planned site designed for the selected atrget audience is a must.
I have no problem in using tools to identify how to optimise navigation
etc however this is only a problem with static (template engine)
based systems. True dynapage engines can provide more
flexible tailored naviagtion and content presentation schemes.
Jacqui
also I think thirdy part stats pakages are a waste of time
unless you can build up context about the users by tying
session tracks to users at some point.
Simply inferirng that the majority of hits form aol.com
go this way has veyr little business value. however knowing that
IT staff all (start or) navigate to specific point is
invaliable.
Kevin Vaughan
> unless you can build up context about the users by tying
> session tracks to users at some point.
But this is exactly what I was saying! Keeping track of individual users,
rather than groups. Not distributing X page because 75% of my users like X
related things, but displaying X page to Y user because they happen to have
shown previous interest in similar pages, while still allowing them plenty
of options to get out to other areas of the site. The statistics gathered
from this information would be very helpful in defining interaction and
navigation with the user.
Jacqui caren
I prefer to empower the user.
offer users as many 9sensible) navigation options as possible.
do not tie them to one or more fixed navigation heirarchies.
integrate 'search engine' (mutlidimensional navigation models)
with strctured navigational heirarchies.
provide tailoring options to allow users to select
what they are interested in
how they want it presented
how they want it structured
All of the above impact the business model and are driven by the
target audience. market Research can be wrong and you this is
where user feedback comes in. However assuming that the limited
amount of hits that reach your server are a definitive cumulative
suggestion can be misleading at best and deluded at worst.
for instance www.gojobsite offers a feature when you can
build a search query from a vacancy profile by clicking upon a
'see jobs like this one' link.
This (IMHO) is far better than trying to second guess which are the most
comonly accessed vacancies and trying to offer a generic or perosnalised
hotlist.
GJS already provides the ability to provide personalised search
profiles - and will even email you info from thedse (on request).
however I would hate to be seocnd guesed about my personal preferences
upon any internet site...
GJS atte,pts to sell things (vacancies to job hunters) etc but
not all web sites have this sales purpose.
There is nothing wrong in having sites for the purpose of
providng information (www.w3.org, www.cnn.com)
providing centralised resources (project management, bug tracking)
market research - something we are planning to do ourselves
very soon.
etc.
All of these can benefit from personalisation however
this means different things in each case.
Applying generic stats analysis to all cases is nonsensical at
best.
Jacqui
p.s.
I should register an interest. Having worked in the web-database
arena for many many years and being involved in the production of
high-end reports for telco customers upper mangement I know just how
'tacky' some of the web stats packages appear. Most of them do
very little other than provide summaries of raw data.
We considered building a set of web log info analysis components for
our DataPublisher product but soon realised that there is no real
market for this sort of high-end product (no one want to pay for
this sort of thing) and the analysis we could being part
functional while eliminating manual procezsing can reduce
the posy procesisng that ofen occurs by customers we could not see
how to jump from raw information to suggestions or recommendations
given the lack of quality of information present in logfiles.
Of course if someone did come up with a need for such reports then
I would be interested in it however I hate the idea of doing something
that seems to have value but is delusionary at best.
kulkar...@gmail.com
URL - http://courier-tracking.org "