Re: Measurement & tracking method: eyetracking shows contributing factors to an ad's visibility

From: Greg Edwards <gedwards1_at_eyetools.com>
Date: Fri 08 Oct 2004 07:47:20 -0500

Peter Faber said:

> Here you make a good point: "Because of this, attention tends to
> follow the fixation. Or, more appropriately, the fixation tends to
> follow the attention (in general)."
>
> Definitely it is more appropriate to say "The fixation follows the
> attention." the other way around is absolute wrong in my opinion.
> (it's like saying the horse moves because of the wagon it is pulling).
>
> How does your tool deal with the fact that attention depends on the
> personal preferences of the visitor?

The short answer is:
--------------------
You can segment the visual interaction data by the responses given to any
free response, multiple choice, brand-lift, or demographic question, which
can be asked before and/or after each task. So, for instance, you could
determine for each individual whether a site or advertisement has positive
or negative brand-lift and then use that to segment your data (and maybe
additionally segment on "income" or "gender"). This approach is also really
useful when you want to improve the experience for your new potential
customers demographic without harming the experience for your repeat
customers.

The more deep answer is:
-----------------------
that behavior is affected by three things:
1) personal preference and individual variance in approach
2) visual presentation of content, and layout
3) psychographic or demographic "behavioral buckets"

Focusing first on (1) and (2):
Personal preference (effect 1) and natural variance will vary from person to
person, yet as a group these divergent "random variables" work the way that
any random sample in a statistical distribution works: the one constant that
affects all of them and which pulls them towards a "mean" is the visual
stimuli they are presented with -- your website or your advertisement
(effect 2). As a result, you see "hotspots" in areas where the page focuses
people's attention -- good designs and copy writing have hotspots in the
areas that satisfy your business needs and the user's needs, bad designs and
bad writing don't.

Trends in what is seen and not seen (and therefore what people "get" from a
site) exist because the effect of the individual's personal preference is
actually embedded within a largely unconscious reaction to the visual
presentation and choice of text wording. I'm struggling with the degree to
go into the explanation, but fundamentally, to use your analogy, "the horse
*believes* it's moving the wagon, and it is, but the wagon is moving the
horse just as much."

The final variable is (3): psychographics or demographic "behavioral
buckets"
The question is, when does "natural variance" actually become a completely
different behavior or set of drivers, and that's where all of the things you
already know about marketing apply: some people will never be interested in
your product, some you'll get no matter what, and most you need to
convince -- you actually segment the trend data to be able to see how your
different demographic or psychographics groups interact with your pages.
First, you see how "everyone sees things", then you break it down so that
you see how "repeat visitors" see things vs. "newsiest", or "people who
LIKED the site or who liked the brand more after exposure to the site"
viewed things vs. those that didn't.

-Greg

Greg Edwards
Chief Technical Officer
Eyetools, Inc.
gedwards1_at_eyetools.com
http://www.eyetools.com






Received on Fri Oct 08 2004 - 07:47:20 CDT