5 ms·
Move beyond behavioral targeting: using mouse movements to read visitor’s mind
- ankneo 16y agoLooks cool :) looking forward to use it !!
- rgrieselhuber 16y agoI remember GazeHawk had done some research about this, I'd love to hear their thoughts.
- bkrausz 16y agoI was preparing my "this is bullshit" face, which involves aggressively linking to our rebuttal ( http://www.gazehawk.com/blog/eye-tracking-vs-mouse-tracking/ http://www.gazehawk.com/blog/eye-tracking-vs-mouse-tracking/ ) Then I looked at the calendar. Well played Paras!
- whathappenedto 16y agoIn the non-calendar related world, some of the cutting edge research on using cursor movements is done by the University of Washington and Microsoft Research: http://jeffhuang.com/Final_CursorBehavior_CHI11.pdf http://jeffhuang.com/Final_CursorBehavior_CHI11.pdf The article shows how movements and hesitations can be used to figure out which links are relevant to you. I believe this is the largest scale study to date (38 subjects on eye-tracking + cursor-tracking, thousands of subjects on cursor-tracking only). Also, this article from Emory University: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.169.5079&rep=rep1&type=pdf http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.169... uses cursor movements to figure out if a user is browsing/researching a product or ready to buy.
- kingsley_20 16y agoPlease make these an independent post?!
- bkrausz 16y agoIt's funny because that paper and our blog post cite the same Google paper. There's some level of interpretation of data, but some of his cited papers offer strong evidence contrary to his point. From the Jeff Huang paper: The mean Euclidean distance between cursor and gaze is 178px (σ = 139px) and the median is 143px. The problem here is that there's a long tail of completely inaccurate data: the average distance between eye and mouse is 178px. I believe there are some people who tend to follow their eyes with their mouse, and some people who don't. Google research suggests the ratio is about 1:2. That being said, I am a bit off topic here: there is most definitely value that can be extracted from mouse position. Can it be used to tell where the user is looking? Unlikely in the common case. Can it indicate purchase intent? I wouldn't be surprised. My rebuttal was mostly premeditated by the amount of false interpretation of research about mouse-eye correlation. There's a lot of really good research out there, but it's often mis-cited to support an opposing claim.
- pkaura 16y agoThis is kool.
- logjam 16y agoRead my mind. Also stole some of the change I had laying beside my laptop. Impressive.
- ankneo 16y agohey one question though.. is it related to the day?? :D
- paraschopra 16y agoWhich day are you talking about ;)
- meadhikari 16y agoApril 1st
- ankneo 16y agoyou understood i guess :)
- oscardelben 16y agoNope, 4chan confirmed their involvement.
- matthewslotkin 16y agoI heard Color was using this technology to determine whether people were angry, disappointed, or just waiting for something interesting to happen when using their app.
- petenixey 16y agoI heard the overwhelming outcome of that experiment was "perplexed"
- jdp23 16y agoActually Color used an early version of this technology to determine how VCs would react to their pitch and tuned it appropriately. This really helps explain both the amount they raised and the confusing quote from Sequoia that "we haven't seen anything this rainy since Google!"
- shadowpwner 16y agoThe ten test subjects is hardly a representative sample.
- paraschopra 16y agoDon't forget all of them are from 4Chan. It is as representative as it can get.
- shadowpwner 16y agoWow, I missed that completely. Tags: april fool joke, behavioral targeting, joke, lol, sophisticated
- andy_boot 16y agoSurely the raininess factor is an April fools at least? The happyness factor from mouse movements? Well I'm unsure but I can see how someone could argue that could be calculated. But raininess? Please no.
- paraschopra 16y agoIt is not about whether it will rain or not. It is about whether a user thinks it will rain or not. Surely it can be predicted, isn't it?
- antihero 16y agoIs that an MCP neuron??
- Silhouette 16y agoCan it also tell whether I am thinking of pink ponies while viewing your site, so that an appropriate visual style and colour scheme can be presented? I think that would be a valuable addition to an exemplary toolbox for professional web developers.
- lwhi 16y agoThere are mathematics symbols, it looks scientific. Must be true :)
- dancunningham 16y agoOne step further... using body movements... http://mail.google.com/mail/help/motion.html http://mail.google.com/mail/help/motion.html "Create a flowchart with ease" :-) But is the undertone a suggestion that Kinect just isn't going to make it out of the world of gaming?
- combiclickwise 16y agoalmost fell for it.. well done
- amnigos 16y agoMouse tracking only solves 20% of the problem.