I think what's missing in the formatting is the dimension of "time" - I noted this on the article comments as well.
I'd love to see an animation that reflects the visual attention of multi-user studies across time span.
Throw out the noise, give an "average" of how and when users see each page element and then you're solving not only for the "where" but also the "when".
That's a crucial piece of data in trying to solve for user abandons, etc.
(GazeHawk intern here.) That's definitely a good approach if you presuppose the existence/obviousness of distinct page elements, which is a whole beast in and of itself.
Right now we allow our customers to define "Areas of Interest" in our reporting system, and we immediately display statistics like average user time on area, average rank of area (ie how many other areas, on average, did users look at first) and so forth. However, it's unclear to me how to represent these numbers in a visually intuitive way in the case where there are 50 distinct areas.
One possibility we're exploring is coloring different areas with different intensities based on some index composed of these features. However, it's not clear what the right formula for this index is, or if it's the same in all situations.
4 comments
[ 2.5 ms ] story [ 11.9 ms ] threadI'd love to see an animation that reflects the visual attention of multi-user studies across time span.
Throw out the noise, give an "average" of how and when users see each page element and then you're solving not only for the "where" but also the "when".
That's a crucial piece of data in trying to solve for user abandons, etc.
Right now we allow our customers to define "Areas of Interest" in our reporting system, and we immediately display statistics like average user time on area, average rank of area (ie how many other areas, on average, did users look at first) and so forth. However, it's unclear to me how to represent these numbers in a visually intuitive way in the case where there are 50 distinct areas.
One possibility we're exploring is coloring different areas with different intensities based on some index composed of these features. However, it's not clear what the right formula for this index is, or if it's the same in all situations.
I didn't do much HCI but worked with some HCI labs that did saccade analysis with photography and photo manipulation. Fascinating stuff.