Google Blogoscoped

Forum

New Click Survey  (View post)

Jonathan Harford [PersonRank 1]

Monday, July 24, 2006
17 years ago21,376 views

Is it supposed to be blank?

Philipp Lenssen [PersonRank 10]

17 years ago #

Yes, only the first image...

Tony Ruscoe [PersonRank 10]

17 years ago #

Another interesting survey!

How long did I wait for the first image to load before realising that it had loaded? Too long! Perhaps you should be recording the time it takes people to click too... :-)

BTW, those images might be a little too big for most 1024 x 768 users though. The bottom 20-50 pixels or so were off my screen (which meant I thought number 7 was another blank one until I scrolled down).

Philipp Lenssen [PersonRank 10]

17 years ago #

I did some changes to make it easier to understand that the first image is blank...

pri [PersonRank 5]

17 years ago #

you should have a link that makes the user see the next image. that way the cursor doesnt stay in the same place as the page updates. because of the link the user will have to move their cursor and the results might be more accurate and independent of the previous image.

Tony Ruscoe [PersonRank 10]

17 years ago #

(BTW, I'm also getting a JS syntax error because the JS file doesn't appear to exist.)

pri [PersonRank 5]

17 years ago #

im curious to know what you were trying to understand from the 2nd (the half black and white) and last slide (slanted X)?

Philipp Lenssen [PersonRank 10]

17 years ago #

JS bug fixed, and I also decreased the space above the image to increase the chances it fully appears.

Ionut Alex. Chitu [PersonRank 10]

17 years ago #

1. This is where 147 people clicked:

2. This is where 127 people clicked:

3. This is where 124 people clicked:

4. This is where 124 people clicked:

5. This is where 120 people clicked:

6. This is where 116 people clicked:

7. This is where 114 people clicked:

8. This is where 113 people clicked:

Philipp Lenssen [PersonRank 10]

17 years ago #

> im curious to know what you were trying
> to understand from the 2nd (the half black and white)
> and last slide (slanted X)?

The general question is how do certain shapes influence click behavior (which may or may not equal focus/ attention)? The first one is a blank and the rest are structuring the image in different ways. So, for example with the black and white image, we can later on see if there is a higher amount of clicks to the right or left side, or possibly if there are any other significant hot zones.

As Pri mentioned the image order might have some effect on the final patterns due to "lazy" clicking on the same position.

Ionut Alex. Chitu [PersonRank 10]

17 years ago #

I tried not to click on the obvious spots.

Eric Senterre [PersonRank 1]

17 years ago #

I got this message at the 8th image :
Parse error: parse error, unexpected $, expecting '{' in /homepages/30/d39253161/htdocs/court/blog/click2/index.php5 on line 125

:(I'll retry later...

Philipp Lenssen [PersonRank 10]

17 years ago #

Eric, pls try again, I think that was just a briefly available error.

Michael [PersonRank 0]

17 years ago #

Each test influenced the next. I tended to keep my mouse in one spot, and just clicked through them all. I wonder how such a test would be set up in true scientific fashion.

Philipp Lenssen [PersonRank 10]

17 years ago #

Michael, every survey is influenced by such things, scientific or not (the room a person sits in, one's culture, the information given or the lack of information given, the approach of the survey, etc.). But in this case, allowing the sequence of images to influence one another is not in any way an unwanted effect... it just creates additional ways of intepretation of results (and indeed, it would be very unscientific to not take into account image 1 when analyzing click patterns of image 2 to come up with conclusions... though I will leave conclusions up to everyone themselves).

That being said, don't forget that what you specifically did – keeping the mouse in one place only – doesn't directly influence the "hot" zones. Hot zones are only existing where the masses clicked. So, there is a chance your initial decision where to place the cursor is repeated by many "thru-clickers", and only in that case it will end up as being one of the several hot zones that may emerge (a zone just as valid and interesting as other zones)... otherwise it will not create such a zone.

Philipp Lenssen [PersonRank 10]

17 years ago #

Update: Digg this if you like it:
http://digg.com/design/Take_a_Test_Where_Do_People_Click

Matt Morgan [PersonRank 0]

17 years ago #

It would be interesting to see if there is a difference in hot zones when users don't know that it is a test to see where they click. That is, I think it might be a different test if you didn't say "then you'll see where others clicked" first. And you had a bunch of naive test-takers try it out.

The fact you're being told ahead of time that your clicks are being compared to others' must affect click placement.

Ramibotros [PersonRank 10]

17 years ago #

the results of image 5 are very interesting .. ppl who were too cool and didn't click the nose, clicked anywhere above it because they somehow did focus on it

Niraj Sanghvi [PersonRank 10]

17 years ago #

Looks like the digg traffic is a bit too much :)

Maybe try caching the results pages? (Assuming you're displaying them dynamically.) You won't immediately see the latest results, but something from a minute ago. But it'll give the database a break :)

big ugly lawyer guy [PersonRank 0]

17 years ago #

it's broken... the digg effect?

Philipp Lenssen [PersonRank 10]

17 years ago #

Thanks guys. I've optimized the script and do some heavier caching now.

Sam Davyson [PersonRank 10]

17 years ago #

Very nice test, I like the first image best such a clear pattern on it.

Mambo [PersonRank 10]

17 years ago #

Who's NOT going to click "dont click here" =P

Philipp Lenssen [PersonRank 10]

17 years ago #

I like the 2nd image too, 'cause it almost splits the first in half... but shows the same cross pattern.

Mambo [PersonRank 10]

17 years ago #

I like the one where the two lines join... because that's kinda of predictable, but could also come up with some interesting results as well.

Jeremy Mikkola [PersonRank 1]

17 years ago #

Amazing – I was surprised buy some of the locations that people clicked.
The click surveys are a brilliant idea. Good job.

Philo [PersonRank 0]

17 years ago #

I would like some way of seeing where *I* clicked on the final results.

Andrew Hitchcock [PersonRank 10]

17 years ago #

This turned out really interesting. I like the image with two lines: you can see how people tried to click in the middle of the blank space between two lines.

/pd [PersonRank 10]

17 years ago #

yes, this survery has a lot of traction..!!

Caleb E [PersonRank 10]

17 years ago #

I thought the volume of clicks in the *opposite* corner from the "click here" was some interesting defiant behavior

Caleb E [PersonRank 10]

17 years ago #

hmm, philipp, did you plan on releasing the click by click data? I bet if you ran a linear regression on the points in the 1st and 3rd quadrant (upper right and lower left sections) on the 8th image you'd get something VERY close to an angle bisector.

Caleb E [PersonRank 10]

17 years ago #

ooh i just figured out why there's a line of dots all lined up on the first image. It's in the same location as the digg link! so everyone is just incredibly lazy. What about hte other line – shorter and higher and to the left?

Niraj Sanghvi [PersonRank 10]

17 years ago #

How are you generating the images of the heat maps? They look very cool.

cactus [PersonRank 0]

17 years ago #

Awsome experiment however like Philo stated previouly you should really be able to see where you clicked. Unfortunitly, I assumed that would be the case therefor was careless to remember where I had clicked, making the experiment, for me, less enjoyable.

Philipp Lenssen [PersonRank 10]

17 years ago #

> Awsome experiment however like Philo stated
> previouly you should really be able to see where you clicked.

You are right Cactus. I'll try to implement that.

Niraj:
> How are you generating the images of the heat maps?

I'm selecting all dots for a specific image. Then I'm loading a blue base image (PNG) that I prepared in PhotoPaint. Then I'm painting small red squares with lots of transparency. Then in a second run I'm painting smaller yellow squares with even more transparency, and then I'm savin the JPG. The base PNG must be high color for good results. I'm recreating the images every 10 minutes after heavy traffic brought the site down when I created every image after a click...

Caleb:
> did you plan on releasing the click by click data?

Good idea! Here you go, the data minus the IP addresses:
http://blogoscoped.com/click2/data.zip

"id" is just for reference, "number" indicates the image number, and "x" and "y" are the coordinates on the image.

Philipp Lenssen [PersonRank 10]

17 years ago #

The result page will now show your click position.

Igg G. Snotzen [PersonRank 0]

17 years ago #

I think you should do this experiment with people, animals, sporting events and the like.

Caleb E [PersonRank 10]

17 years ago #

This definately deserves some follow up, especially now that you've optomized the process, philipp :)

I think you should do one on popular websites. Where are people's eyes (mouses) drawn by the design of certian pages (Google (the search box probably), Yahoo (the content in the center), Digg (the first link and the ads), Slashdot (first story) These are just predictions though, would be intersting to see if they are correct). You could also do new yahoo vs old yahoo, or new digg vs old digg, or regular google vs some of thier design experiments, google SERPs vs Yahoo SERPs vs Ask SERPs vs MSN (Live.com) SERPs. The possibilities are endless :)

Philipp Lenssen [PersonRank 10]

17 years ago #

Yeah, I might do a follow-up with websites!
I saw eyetracking data for Google.com so now we could compare how eyetracking and clicktracking compares, it might be related!

/pd [PersonRank 10]

17 years ago #

eyetracking data for google ?? do you have some pointers on that Philipp ?

Caleb E [PersonRank 10]

17 years ago #

I just thought of some more:
apple.com vs microsoft.com
MSN Search SERPs vs Live.com SERPs
hp.com vs gateway.com vs dell.com etc
gmail vs yahoo mail (new and old) vs hotmail (new and old)
all of technorati's designs (you already have the pictures!)
Google Maps vs Yahoo maps vs Live Local
Google news vs yahoo news vs cnn.com vs msnbc.com etc
amazon.com vs overstock.com vs buy.com vs ebay.com
google blog vs yahoo search blog vs ask.com blog

Philipp, you could make a whole website out of this. It'd be very interesting.

Caleb E [PersonRank 10]

17 years ago #

This is what philipp is talking about i think:
http://eyetools.com/blog/images/google/eyetools_google_search.jpg

Philipp Lenssen [PersonRank 10]

17 years ago #

Hmmmm... I could shuffle the order in which the images appear for every user. This way, those who only look at the first or second image and then click through wouldn't create significant "noise".

/pd [PersonRank 10]

17 years ago #

Philipp, with those 8 pngs is it possible to ahve a pic list to supper impose /overlay one over the other

for eg I want to review Png(3) overlaid on png(7) ??

A Process, where I can pick a base picture and superimpose any other png on it or a number of pngs on this..

This will tell me pattern variations based on liner vectors of the orignal picture..

Philipp Lenssen [PersonRank 10]

17 years ago #

Excellent idea Peter. I created two things:

- A PhotoPaint image with all layers, so you can swich on/ off the visibility of different layers:
http://blogoscoped.com/click2/all.zip

- and a PNG image showing all layers on top of each other:

http://blogoscoped.com/click2/all.png

Niraj Sanghvi [PersonRank 10]

17 years ago #

Philipp, very cool...thanks for the image info!

Dynamic image generation is something I've always been meaning to play around with more :)

Sean [PersonRank 0]

17 years ago #

I like the idea of recording the amount of time it takes a person to click and plotting that in the z-axis. It might give some insight into the distribution of people who click impulsively (right away) and the those that think about where to click and any correlations with where in the (x,y) space their click lands.

/pd [PersonRank 10]

17 years ago #

the patterns are similar to the galaxy systems :)-

nice data culls here!!

Corsin Camichel [PersonRank 10]

17 years ago #

let me guess where the "Click Here" and "Don't click here" texts are :)
And the center has a ball in it

Philipp Lenssen [PersonRank 10]

17 years ago #

> I like the idea of recording the amount of time
> it takes a person to click and plotting that in
> the z-axis.

While I didn't record the time before a click, I recorded the IP + date/time of every click. Using this data, I could generate delays between IP click 1 and IP click 2, making some assumptions about session length and such...
Hmmm.

Kim ki joong [PersonRank 0]

17 years ago #

what a fun survey.

Sunil Nair [PersonRank 9]

17 years ago #

Maybe you ought to have a short time window (say, 5 to 10 seconds) within which to click – the results would be more instinctive, than deliberate.

Clickchick [PersonRank 0]

17 years ago #

Funny thing, as they say:people are all the same. In Holland we call it 'kuddegedrag'.

Mel [PersonRank 0]

17 years ago #

How lame was this?

Billy [PersonRank 0]

17 years ago #

Great idea, awesome. I'm going to feature it on my website. www.helpthebongless.com

Clemens Packheiser [PersonRank 0]

17 years ago #

When i saw the words click here i automatically clicked there. it was so amazing...

errr maybe i shouldn't spend all day long on the internet

[URLs removed]

:D

have a nice day

Nic [PersonRank 0]

17 years ago #

That was fuckin' dope, yo.

John Doe [PersonRank 1]

17 years ago #

Interesting stuff. I expect Derren Brown to be behind this.

Philipp Lenssen [PersonRank 10]

17 years ago #

I just found this bit via Search Engine Watch:

Clickstream Heatmaps Resemble Eye Tracking Heatmaps
"User click patterns appear to follow eye fixations."
http://www.websiteoptimization.com/speed/tweak/clickstream/

Jon Stoneman [PersonRank 0]

17 years ago #

Hi Philipp,

Nice survey. Click distribution is a very interesting and the results are often quite surprising.

You might be interested in our 'clickdensity' service. It records clicks on web pages and generates image maps from the data.

http://www.clickdensity.com/?refCode=wjsoc

It's a monthly subscription service, but you there is a 30 day free trial and a demo where you can view the images for the clickdensity website.

You can filter the clicks shown on the images by time since page load and various other attributes.

Cheers,

Jon.

Roger Browne [PersonRank 10]

17 years ago #

Clicks on the blank image are concentrated on very precise diagonals; much more geometric than I would have expected.

Even if we don't click near the center or the edge, it's likely to be on a diagonal. Interesting!

pradyumna [PersonRank 0]

17 years ago #


It would be good if you can also display the total number of persons who have clicked that particular image.....

Philipp Lenssen [PersonRank 10]

17 years ago #

> It would be good if you can also display the
> total number of persons who have clicked that
> particular image.....

I do, above the image...

GDR! [PersonRank 0]

17 years ago #

Interesting :)

krul [PersonRank 0]

17 years ago #

nice, but not very usefull.....may be next time some more interesting pictures =]

Marc [PersonRank 0]

17 years ago #

Very nice survey, it's funny because even i got fooled by the "click here" thing :D and most did :P
then the "don't click here" showed and i clicked on that heheh

Heather [PersonRank 0]

17 years ago #

Great tool! Like some of the folks were suggesting, it would be an interesting usability testing tool as far as successful page layout and information hierarchy are concerned. I'm still not sure how you're generating the results dynamically, but would love more info on that.

Philipp Lenssen [PersonRank 10]

17 years ago #

Heather, I'm using PHP5 + GD library – for more see:
http://blogoscoped.com/forum/46688.html#id46816

Helen Griebeling [PersonRank 0]

17 years ago #

Too obvious where the hot spots will be.

Unmesh [PersonRank 0]

17 years ago #

I Agree with u helen

shubalub galoomp [PersonRank 0]

17 years ago #

weird but fun and good and different and interesting

Imogene lewis [PersonRank 0]

17 years ago #

toatly RANDOM!!!! See YA!!! ;~)

The Bob Unit [PersonRank 0]

17 years ago #

How come, on that first picture, where ever I click it says that "about 65,000 people clicked there"?

Philipp Lenssen [PersonRank 10]

17 years ago #

I turned it static today.

sean howell [PersonRank 0]

17 years ago #

that was good to see how many other people clicked in the same spot

Gordon Luance [PersonRank 0]

17 years ago #

This survey rocks!! Looking through the comments I spotted one from Jon Stoneman who includes a link to a tool called clickdensity. I quickly checked this out and it seems to do exactly what people are suggesting here. This seems to be a whole new approach to usability analysis, as Heather suggested in a comment above. I've signed up for a free trial, and already have 340 clicks. Worth looking at it: http://www.clickdensity.com

Maaike risterveldt [PersonRank 0]

17 years ago #

Cool!

dposse [PersonRank 0]

17 years ago #

hey, that was pretty cool. Good job. ^_^

Mina. F. Little. [PersonRank 0]

17 years ago #

That is a verry cool survay and surprizing too!
I dident expect so manny cliks in some places and i thought that more people would have cliked somplaces...
I nearly cliked on the "CLICK HERE" but decided against it Maybe it is also a personality test?

noname [PersonRank 4]

17 years ago #

I got one more idea, how to get rid of false clicks when someone clicks on the same position:
1. make a blank page with "click here" link on the top. You can first clear the page from the address bar and status bar if the browser allows it, to have a maximum area
2. make a second page, where there would be nothing in the position where was the "click here" before
3. put all the pictures on that second page and make e.g. 3 in a row, 3 in a column, each with width: 30% or height: 30% so they will all fit the page without scrolling and when you want to click them, you need to move your mouse.

You can stay with only putting the "move to next" link somewhere outside of the pictures frame, but in that case, some people will click the area close to that link.

P.S.: you may put there images of some default web designs – google search, company web, blog – just you should use some Lorem Ipsum text.

photoDude [PersonRank 0]

17 years ago #

at first i thoght it would be shite. I was intrigued by the results. nice work.

Roderdron [PersonRank 0]

17 years ago #

superb idea.

Tru [PersonRank 1]

17 years ago #

Very interesting. As were some of the comments that followed.

Astrid:) [PersonRank 0]

17 years ago #

i gave the smiley face a nose...well, i think i did.:?

Not special [PersonRank 0]

17 years ago #

Wow, we're not as original as we think we are....

Zoidberg [PersonRank 0]

17 years ago #

That's because as everyone tries to be original, they modify their original choices approximately the same way, leading to similar results

mememe [PersonRank 1]

17 years ago #

most excellent survey. it's very interesting to see that most people will click in the obvious places. very cool i must say!

Chimera [PersonRank 0]

17 years ago #

Every spot was pretty random except the smiley face, i wanted to give that a nose

Boo [PersonRank 1]

17 years ago #

pointless.

SKINSMAN [PersonRank 0]

17 years ago #

This survey kind of reminds me of football – that you can pretty much click anywheres (in football throw anywheres) and there will have been someone else doing the same thing. And just like football, clicking on these things is sort of a metaphor for life – that we're not alone, and that we pretty much have no originality at all – or are at least limited in our originality... though football isn't necessarily non-original. I mean, there's the West Coast offense, and the East Coast Offense, not to mention the mid-west Defenses and the upper mid-west special teams. Those are kind of original. Anyway, enough said.

nk [PersonRank 0]

17 years ago #

very interesting.

Melanie [PersonRank 0]

17 years ago #

Are there any more surveys besides the one with the women playing cards? Oh and why does it always say 40,000 people clicked here?

Eve [PersonRank 0]

17 years ago #

Wow that is so cool!

Monicah [PersonRank 0]

17 years ago #

like.. wtf.

Philipp Lenssen [PersonRank 10]

17 years ago #

> Are there any more surveys besides the one with
> the women playing cards?

Not yet, at least on this site... maybe in the future.

Tanner Thomas [PersonRank 0]

17 years ago #

Interesting

Sasha Sasha [PersonRank 0]

17 years ago #

Wow...that was so odd....like, I don't get the purpose, but I do agree with some of the other people about the whole "we're not as individual as we'd like to believe..." thing. Makes sense. Good Stuff.

me [PersonRank 8]

17 years ago #

that was neat.

That One Chick [PersonRank 0]

17 years ago #

That was really interesting. I was a bit surprised how much they differed...

Ashley Bryne [PersonRank 0]

17 years ago #

That was a really interesting survey! It's cool to look at where others clicked and compare it to your own.

princess-angel [PersonRank 0]

17 years ago #

what the?

stinky [PersonRank 0]

17 years ago #

that woz soooo much fun hahahahahahahahahhahahahahhahahahahah

[anonymous] [PersonRank 4]

17 years ago #

that wasnt detailed enough, a true waste of time.

just a guy [PersonRank 0]

17 years ago #

i also found that i was VERY unoriginal in my choices . ----HMMM i wonder what that means???

da_marb [PersonRank 0]

17 years ago #

it just shows how common u r lol

Sassy Saxon [PersonRank 0]

17 years ago #

thats pro please make more!!!!!!!!

Wibawoba [PersonRank 0]

17 years ago #

Why?

catt napp [PersonRank 0]

17 years ago #

I found it strange to see how many people clicked in the circles, the mouth, nose or eyes and in places where there was no rectangle. Enjoyed though.

Rie [PersonRank 0]

17 years ago #

Yay I'm different, I clicked in the colder areas :)

Eva [PersonRank 0]

17 years ago #

This is LOVELY!!

I approve!

Good work, friends!

some guy you might know [PersonRank 0]

17 years ago #

this is so .... stupid, fun ,weird,piontless i love it

Dan [PersonRank 0]

17 years ago #

I hardly think it's stupid or pointless.. It illustrates a really good point: i.e. that the mind is attracted toward detail.

KOIKOI [PersonRank 0]

17 years ago #

FAKE

Crawly [PersonRank 0]

17 years ago #

Wow... interesting to see how even blank images produce lines and geometrical figures. We are just small parts of the big equation...

carrettop [PersonRank 0]

17 years ago #

i think thats so cool
i wana but my own pic in

Jethro Wilbury [PersonRank 0]

17 years ago #

Wow...that was entertaining...but I'm still kinda bored.

zoe [PersonRank 0]

17 years ago #

i liked it! you should have more!

metches [PersonRank 0]

17 years ago #

that was awesome

Forum home

Advertisement

 
Blog  |  Forum     more >> Archive | Feed | Google's blogs | About
Advertisement

 

This site unofficially covers Google™ and more with some rights reserved. Join our forum!