Tracking a user's mouse location

see the results...draw your own conclusions

[ggrot]

In this recent thread:
[webmasterworld.com...]
There is some discussion regarding where a user's mouse is generally located on the screen. It is theorized that this would be the best place to put the navigation structure when considering left vs. right navbars. Nobody really knew except from some simple observations about their own habits...so I decided to test it out. I took a site of mine that had no left nor right navigation for a test. The site in question only uses top and bottom horizontal bars for minimal navigation (most navigation is within the body). The exception to this is that on some pages there is a set of related products links on the far left.

I used a little javascript code to detect the location of the user's mouse, and used a little formula to agregate this data irregardless of browser size. The javascript simply calculated a value 0-100 which represented columns across the browser of width browserwidth/100 rounded. This way I could get an idea of the region the mouse was in independent of screen or browser size.

Then I sampled the mouse location 10x/second and had the javascript code request a 1 byte image from the server whose url was unique and included the x-percentage in it. The server recorded all this data. So far I've collected about 400 pageviews(estimate). The graph still hasn't settled out completely, and I had to remove a couple outliers (out of 101 sample points) to make things look reasonable, but there is an obvious trend. Take a look at the graph:

The red points are 99% and 100%. When testing, I noticed that if you ran your mouse off the screen to the right, that you would still record 99 or 100. So these 2 numbers also include the scroll bar, which explains why they are large.

From the trends, I interpret this to mean that few people put their mouse on the left hand side of the screen, and the same goes for the far right. Apparently, the most mices are in the middle, but it does seem to lean towards the right a bit.

Anyone else care to interpret any of this? Also, if anyone wants to add their site's views to the experiment, sticky me and we'll give it a whirl. It would only take 2 lines to include an external js file and you'll be included in the results. I'll update you as the graph becomes clearer.

[EliteWeb]

I keep my mouse next to that close window box in the corner of the window for whichever operating system I use. Quick Closing ;)

[tedster]

Thanks for the experiment. It's great to see actual data, rather than being confined to anecdotes and impressions.

It looks like something close to a bell shape curve is emerging - and the bottom of the screen looks quite heavy with data points.

I've often appreciated the websites I've seen that maintain a bottom navigation in a frame. Once position:fixed is better supported, bottom navigation will become an interesting venture.

[ggrot]

I should clarify something...vertical location on the screen is not represented in this data, the vertical location on the graph simply indicates frequency of horizontal location. I'm not sure if that was what you were thinking about regarding the bottom navigation option, tedster, but I haven't tracked that yet. The problem is that I can only track pixel location, I have yet to translate this into document height values (taking the scroll bar into account). It would be neat to be able to use this as a logging tool and overlay a color coded grid on every page showing where the mouse hovers the longest on average.

[visitor]

ggrot:

Very interesting experiment:)

Please keep us posted.

[Shane]

I may have missed something, ..... did you do this for one page or for several? If only for one page then might not the layout of the content affect the mouse position (especially if there are help items poping up?).

Doing it for several pages where the content is in different positions and corrolating the mouse position versus content would be interesting. (And some work.....)

Ciao,
Shane

[brotherhood of LAN]

Right now my mouse is near the bottom right corner of the screen, because i clicked a button down there to get here to write this :)

I dont think any small scale experiment could truly be deemed reliable when you take factors like screen resolution/browser type/font settings/type of mouse/navigation structure/quality of webmaster

How about it? :)

[ggrot]

So far the expirement is 1 site, multiple pages. The javascript only runs using IE, and neutralizes screen/browser width by using a percentage width measurement. Type of mouse will play into it, but I don't think that a sample of a few hundred users would be too distinct from the population. As for quality of webmaster...well, I can only say that I do my best.

[ggrot]

[216.114.78.161...] has a larger data set in it. Looks like the trends are disappearing some. Might be that its fairly random after all.

[tedster]

The left to right trend is fading, but the bottom of the screen trend is clarifying.

Again, the idea of static, bottom of the screen navigation looks more and more appealing tome, but I'll be danged if I'll do it with frames!

[bird]

I think you should make this a bar chart instead of a scatter plot, or you'll keep confusing tedster with those "bottom of the screen" ideas... ;)

The results are very interesting nonetheless, and would get outright fascinating with data from more than one site. Seeing this much, I suspect that the useability of any specific navigation menu location depends more on where the user has his eyes focused most often than where his mouse pointer resides at a specific point in time. They can only move the pointer to a location they look at, unless they just want to have it out of the way.

[Shane]

Building on Bird thought:

Would it be useful to look at the distance a mouse is required to move to get to the desired link?