| |
Figure
1 : Sample poll with pouroucontre.com
Despite the immediacy of online polling and
however nice and democratic they may look,
a serious webmarketeer will be after something
a little more elaborate. Beyond the problems
of representativeness of your 'samples' which
we have studied in part
I, one may rightfully wonder how pertinent
Internet polling may be. This type of questionning
actually involves that your items be brief
and straightforward in a 'for or against'
format (in my above example, there is no way
of knowing for sure whether the respondent
is a survey-pro or a simple surfer, let alone
whether he or she will ever envisage posting
a survey online, etc.). Let us then focus
on fully-fledged, professional Internet surveys
in order to gather more precise data (about
a website, product or service) and make decisions.
Designing and administering an 'IRL*' survey
requires that one follows a few rules. These
rules are now well known for quantitative
surveys even if our experience in the field
shows that they are sometimes ignored. The
methodology for building questionnaires in
12 steps will provide you with the basics.
I have adapted that methodology to online
surveys in part II of this article. In the
third and final part, I will recommend our
visionary marketeers a choice of available
software online. This latter part will come
with real-life examples. METHODOLOGY We have
already covered the difficulties generated
by the weakness of online sampling in part
I. However, the fact remains that the
questioning mode still does matter anyhow.
In other words, one should not add more biases
to those already brought in by the non-representativeness
of the sample. At any rate, online studies
remain the best possible way of measuring
Website audiences because of the permeability
between site and survey interaction.
Stage
ONE :
The Audience |
More than sampling, I
like to mention the volume pf websurfers
that you will be able to attract on
your survey main page. For this is problem
number one of the Internet marketeer.
To me it's on top of the agenda, even
before the setting-up of the objectives.
Indeed, a lack of respondents will render
the whole endeavour entirely useless.
Whereas interviewing 1000 people in
the high-street could prove long and
tedious (if not costly), when it comes
to measuring online services it will
also prove inappropriate to do it offline.
Yet, recruiting a 1000 interviewees
on the Net is a more difficult business
than it would seem at first sight. Now,
there are a few long-lived myths on
the Net : People will tend to make you
think that clicks will come to your
pages by millions. One should make it
clear that it's untrue. If you are not
one of the 'majors' or one of these
'dot-coms' with millions to spend on
advertising you will soon find out that
attracting new visitors and building
a Wensaudience requires a lot of effort
on your part.
To overcome this problem,
an attractive banner should be installed
on your pages (at the end of the purchasing
process for instance) and mainly, you
will have to chase your users via e-mail
while repeatedly reasserting that the
analysis of their responses is key to
the improvement of your online service.
The latter method is also the most effective
I have found. It also enables you to
target your respondents hence minimise
the so-called 'non-representativeness'
problem we have studied in part I.
Allocating a present to
your respondents (via some sort of online
contest or raffles) can be rewarding
too, but this method may be less accessible
to smaller outlets with limited budgets.
On the other hand, I would
advise against online surveys at all
cost if your target is a panel of experts
(as is the case with Delphi surveys).
If such is your need, do revert to face-to-face
interviewing or individual e-mails/faxes
sent after a telephone call. Indeed,
the probablility for knowledgeable experts
to visit your pages by mere chance is
feeble and don't assume that they will
be ready to cooperate anyway.
|
|
Stage TWO :
The Objectives
|
- With the preparation of your first
online survey, you will logically
wish to try and test whatever goes
through your mind. The contents
of your Website, its user-friendliness,
its navigation, the quality of your
products, the reliability of your
delivery process, etc.
It's a common pitfall of online
and offline surveying. Trying to
fathom the whole Website perception
in one go is in fact a very bad
idea, unless you manage to be brief.
But then the risk for you to gather
very vague answers is high. Cut
down your objectives to paper, then
your whole survey will be at risk
and it is very likely that your
question will be jumping from one
subject to another and also come
out in random order.
|
|
Stage THREE :
Pre-testing the Survey
|
- In addition to all the inevitable
general remarks on that theme ( see
the building
questionnaires in 12 steps ),
pre-testing Net-based questionnaires
is a tough job because you will
have to convince a few volunteers
and it's not a piece of cake, mainly
outside the corporate context.
- If it comes to the worst, test
yor survey on your own, while trying
to put yourself in the shoes of
an ordinary customer. I would be
surprised if you didn't find many
of your questions need rewording.
|
|
Stage FOUR :
Assumptions
|
- Based on stage two, spell out
the statements that you wish to
confirm or contradict,
- Do not filter your assumptions
based on personal prejudices such
as "I don't need to check that
point since I think that ..."
unless you have gathered facts on
the subject.
|
|
Stage FIVE :
Qualifying / filteringinterviewees upfront
|
|
|
Stage SIX :
Writing/editing your questions
|
- Design your questions so as to
be understandable (to anyone) unequivocal
and non-suggestive,
- Avoid provoking your respondents
with your questions,
- Avoid circumlocutions and redundancies
which may tire the interviewee (he/she
will then be tempted to fill-in
your questionnaire at random),
- Be concise, since the Internet/html/screen
format forces you to,
- Edit your question according to
your pre-test (stage three),
- Don't be afraid to suppress all
redundant/useless questions,
- Do add questions before it's too
late (when relevant and useful).
|
|
Stage SEVEN :
Installation/
administration
|
- Install your page and reference
it - if necessary - through submit
it,
- Launch an e-mailing campaign to
warn your users about the availability
of your online survey,
- Include links to your survey on
your Website whereever possible
such as on the purchase-confirmation
page for instance.
|
|
Stage EIGHT :
Analysis & report
|
- Launch a flat file analysis and
add filters to refine it,
- Favour graphical analysis tools
in direct html format which can
be uploaded online on an Intranet
or the public Internet,
- Paper-based reports as far as
Internet surveys are concerned are
less useful than online ones.
|
|
|
