use of hit counters etc.
Nicoll, Mark
work for the Government of Alberta Canada in the area of Performance
Measurement. We have just launched a new web page which pulls together
links to all performance measurement related documents issued by this
government. Having done so, we now want to know who is using it, how
frequently it is being used etc. We would also like to survey every
"x"th visitor to find out how useful it has been for them etc.
I'm wondering if anyone else has experiences in this area that they
might share with us. Thanks very much!
Mark Nicoll
Performance Measurement
Alberta Treasury
Room 455, Terrace Building
9515 - 107 Street
Edmonton AB T5K 2C3
ph: 403-427-8345 nic...@treas.gov.ab.ca
Find the Government of Alberta's performance reports at:
http://www.treas.gov.ab.ca/comm/perfmeas/index.html
Mark Nicoll
Performance Measurement
Alberta Treasury
nic...@treas.gov.ab.ca
ph: 403-427-8345
Carl Hage
> We have just launched a new web page which pulls together
> links to all performance measurement related documents issued by this
> government. Having done so, we now want to know who is using it, how
> frequently it is being used etc. We would also like to survey every
> "x"th visitor to find out how useful it has been for them etc.
The short answer is you can't.
Server access logs do not have individual userids, only remote site IDs (the
source network, i.e. which ISP or corporation). Also, keep in mind access
logs do not count all accesses. For example, if 1 AOL or Prodigy user
accesses your page, then the next 100 users might get it from the internal
cache.
Since you can't send random surveys, rethink the implementation so you
make it easy and encourage users to submit feedback or have other kinds of
interaction. If you have a mailing list, then you could send surveys via
that, for example. You might use the mailing list to notify users of
updates, etc.
Create a feedback form with general comments, plus you can include a form
with specific questions. Some places, e.g. customer support pages that answer
questions, have a short form at the page bottom, "Did this answer your
Question".
Also, don't rely on hit counts-- it should only be a rough guide. If
management asks for accurate usage statistics, tell them that it's not
technically possible. When you mention the number of hits, always include a
disclaimer that not all accesses are included.
--------------------------------------------------------------------------
Carl Hage C. Hage Associates
<mailto:ca...@chage.com> Voice/Fax: 1-408-244-8410 1180 Reed Ave #51
<http://www.chage.com/chage/> Sunnyvale, CA 94086
Dana Noonan
> From: "Nicoll, Mark" <Nic...@treas.gov.ab.ca>
>
>> We have just launched a new web page which pulls together
>> links to all performance measurement related documents issued by this
>> government. Having done so, we now want to know who is using it, how
>> frequently it is being used etc. We would also like to survey every
>> "x"th visitor to find out how useful it has been for them etc.
Although I agree with Carl Hage that encouraging all users to
provide feedback makes sense, a simple script could work.
The automatically generated access logs are not the best solution
for selecting every ""xth visitor. I would use a script to log
and count visits to specific pages. The same info available in
the access logs (or a subset) could be saved in a database for
later analysis.
Adding a simple counter so that every 20th visitor would
be asked to fill out a survey or see a special "click here to take
survey" image would be pretty easy to do.
One site I designed uses a simple CF script at the top of the
page to log document access and specific search terms. We used the
data to check to see if the instructions on using the site search
engine were understood by the users in various categories (by domain,
time, type of document, etc.).
Tracking referrer log info at the same time would tell you
if most of the folks reading page_X came from Yahoo or from
a link on your site.
A note on surveys: Most folks resist long surveys on the Net.
A quick one question survey on a page works best. However, it
might work to ask folks to volunteer to be part an online
focus group or site advisory group. These folks would be likely
to be the most committed users and be willing to take time to
review a variety of site and topic related issues.
Last issue: Please make sure that method of survey selection
and the content of your surveys conforms to your site's privacy
statement. Pillsbury has a simple but clear privacy
statement at: http://www.pillsbury.com/privacy.html
Dana Noonan
Piper Resources
noo...@piperinfo.com
http://www.piperinfo.com/