Clickstream/heatmaps (was Eye-Tracker software/hardware recommendations)

26 Aug 2009 - 10:00am
5 years ago
2 replies
959 reads
Adrian Howard
2005

On 25 Aug 2009, at 19:18, Nick Gould wrote:

> *As an aside, I think it's interesting that many of your arguments
> against eyetracking could also be leveled against clickstream
> analysis / clickmaps, etc... I am amazed at how willing clients are
> to believe that this data is meaningful on its own.

Yeah. I've had "discussions" with a few people recently about
clickstream/heatmap results:

Them: "Look everybody's clicking here where Y is - we want them to do
X. We should move X to Y's location"
Me: "Erm.... maybe it's because they want X? Or the language talking
about Y is wrong? Or..."

Them "Look - this person has been on the verge of clicking on this
area for ages. Look at them wiggle the mouse around".
Me "Did you ask them?" (turns out the person in question was stopping
his over-eager laptop screen dimming kicking in while he read the text
in the body of the document)

Then again - at least clicks are (usually) instances of somebody
actually wanting to interact with something. They've been a great tool
for helping demonstrate that bits of the interface that look
"clickable" should actually do something useful (or not look like they
should be clicked on.)

Cheers,

Adrian
--
http://quietstars.com - twitter.com/adrianh - delicious.com/adrianh

Comments

26 Aug 2009 - 1:51pm
Jared M. Spool
2003

On Aug 26, 2009, at 11:00 AM, Adrian Howard wrote:

> On 25 Aug 2009, at 19:18, Nick Gould wrote:
>
>> *As an aside, I think it's interesting that many of your arguments
>> against eyetracking could also be leveled against clickstream
>> analysis / clickmaps, etc... I am amazed at how willing clients are
>> to believe that this data is meaningful on its own.
>
>
> Yeah. I've had "discussions" with a few people recently about
> clickstream/heatmap results:
>
> Them: "Look everybody's clicking here where Y is - we want them to
> do X. We should move X to Y's location"
> Me: "Erm.... maybe it's because they want X? Or the language talking
> about Y is wrong? Or..."
>
> Them "Look - this person has been on the verge of clicking on this
> area for ages. Look at them wiggle the mouse around".
> Me "Did you ask them?" (turns out the person in question was
> stopping his over-eager laptop screen dimming kicking in while he
> read the text in the body of the document)
>
> Then again - at least clicks are (usually) instances of somebody
> actually wanting to interact with something. They've been a great
> tool for helping demonstrate that bits of the interface that look
> "clickable" should actually do something useful (or not look like
> they should be clicked on.)

People click on things all the time that are *not* things they
inevitably want. Thus the problem with pogosticking (http://is.gd/
2AiMU).

The problem with clickstream analysis is you can't tell the misfired
clicks from the desired clicks.

Nick is right -- all the problems I have with eye tracking, you have
with any instrument where you don't know "why" something is happening.
It's all about the inferences you make from the raw observations.

Jared

27 Aug 2009 - 4:06am
Adrian Howard
2005

On 26 Aug 2009, at 20:51, Jared Spool wrote:

> People click on things all the time that are *not* things they
> inevitably want. Thus the problem with pogosticking (http://is.gd/2AiMU
> ).
>
> The problem with clickstream analysis is you can't tell the misfired
> clicks from the desired clicks.

I agree completely :-)

What I was trying to say (obviously badly) was that there is (usually)
a deliberate intent to "click" - most of the time they didn't hit the
button accidentally. Not that the user wanted to go wherever that
click led (or not). Or that you could necessarily infer anything about
the reason for that click without further investigation.

So when I have a chunk of interface that is purely informative but
looks like it's something that should be interactive, and the designer/
PO don't believe me, looking at a nice fuzzy heatmap of hundreds of
users clicking on the darn thing is trez useful.

Of course this is nothing that you wouldn't get from a five minute
user test - but sometimes it's politically easier/cheaper/faster to
get a lump of javascript added to a web site and look at those results
rather that do some quick guerilla testing. Annoying I know. I can
then use those results to push for more pre-release testing to catch
those issues earlier.

Make vague sense?

Cheers,

Adrian
--
http://quietstars.com - twitter.com/adrianh - delicious.com/adrianh

Syndicate content Get the feed