Ask HN: What data about users and how they use your website do you wish you could collect?
I'm interested in what sort of user browsing/usage data you folks wish you could collect from your website (numbers, visual data, etc). What features can't you currently get from another service, or could be made more useful?
5 comments
One thing that I can't get via Google Analytics (the only one I use) is how long a person stays on one page when they "bounce" immediately. I have often thought there should be an easy way to build a little javascript ajax call that polls every 5 seconds or something, and whenever they leave - it would stop polling and you'd know they were gone.
There are so many details we already have (keywords, page of entry, etc) - but the bounces frustrate me a lot. If they are bouncing after 2 seconds, that's a problem. If they are bouncing from my info page after 60 seconds, that's a totally different problem: it means they liked what they read, but I somehow couldn't entice them to stay.
There are so many details we already have (keywords, page of entry, etc) - but the bounces frustrate me a lot. If they are bouncing after 2 seconds, that's a problem. If they are bouncing from my info page after 60 seconds, that's a totally different problem: it means they liked what they read, but I somehow couldn't entice them to stay.
I have often thought there should be an easy way to build a little javascript ajax call that polls every 5 seconds or something, and whenever they leave - it would stop polling and you'd know they were gone.
That would cause a DDoS.
That would cause a DDoS.
This could be done with a synchronous ajax request attached to the (nonstandard) onbeforeunload event.
Something else I'd like to see is a measurement of average page load times. In practice the best you could do is time from tracking script load to dom ready and/or body load. So your numbers would be off, but still useful for spotting trends.
Something else I'd like to see is a measurement of average page load times. In practice the best you could do is time from tracking script load to dom ready and/or body load. So your numbers would be off, but still useful for spotting trends.
I believe analytics shows you average time on site...
I think this measure time between pageviews. IE bounces recorded as 0.
I don't think it is between page views.
My site is a one page ajax heavy site, so I can only have analytics on the landing page.
The time on page shows around a minute average (changes per day), while the bounces shows nearly 90%. Doesn't fluctuate as much.
Not sure how they do these measurements. But what is really needed is a tool that can measure on ajax sites.
For now i've started counting server side requests, and it really is showing how unreliable Google Analytics is.
Not sure how they do these measurements. But what is really needed is a tool that can measure on ajax sites.
For now i've started counting server side requests, and it really is showing how unreliable Google Analytics is.
http://www.google.com/support/analytics/bin/answer.py?hl=en&...
and
http://www.google.com/support/analytics/bin/answer.py?answer...
Doesn't work to track Ajax? I've never used but was thinking to apply for some of my projects.
and
http://www.google.com/support/analytics/bin/answer.py?answer...
Doesn't work to track Ajax? I've never used but was thinking to apply for some of my projects.
I think that after X amount of time the page 'reloads.'
<begin brain dump>
I've never really seen a good visualization of a site's link graph (with focus on link text & page rank), probably due to the sheer size of many sites.
I'll second other posters that time on site for each page is a biggie.
Anything you can do to expand on the 'funnel' model is solid. (Clicktracks treatment of this is especially good IMHO)
Just to throw it out, I'd love to know more about if/how my users use other sites (so i could see if they've been on competitor.com or wherever recently =P), Although I can't possibly think of a way to do this that isn't a blatant privacy violation.
I'd also like more clustering options by language or country, so i could see if US/English users use the site differently than UKers or German speakers, although I realize that for most sites that are international enough to care, they have multiple language sites, there's still visitor cross-over.
This is a stupid one, but automatic FTP downloading of log files is something I seem to have to set up all the time, so a mechanism for aggregating logs would be nice =).
All just random brain noise, hope it helps =).
I'll second other posters that time on site for each page is a biggie.
Anything you can do to expand on the 'funnel' model is solid. (Clicktracks treatment of this is especially good IMHO)
Just to throw it out, I'd love to know more about if/how my users use other sites (so i could see if they've been on competitor.com or wherever recently =P), Although I can't possibly think of a way to do this that isn't a blatant privacy violation.
I'd also like more clustering options by language or country, so i could see if US/English users use the site differently than UKers or German speakers, although I realize that for most sites that are international enough to care, they have multiple language sites, there's still visitor cross-over.
This is a stupid one, but automatic FTP downloading of log files is something I seem to have to set up all the time, so a mechanism for aggregating logs would be nice =).
All just random brain noise, hope it helps =).
Really depends on the site. But the list is endless.
There are lots of things that would be useful to know.
Here is an example
So, say you want to know how regular users of feature X, to get to it. There are lots of questions you might want answered.
Say you find 5 - 10 'paths.'m (eg google search for 'site.com'>homepage>pagex>login') You might want to see the general pie graph of which paths get used. Google analytics can probably handle this. But what if you want to know about 'users' using this path. Do users migrate from path A to path B over time? Do users form their path in the first visit & stick to it? Are users erratic in their choice of path? Do users of the site that joined after redesign (sub segment) use a different path. Are users of more advanced features (sub segment) more likely to use path X.
There are lots of things that would be useful to know.
Here is an example
So, say you want to know how regular users of feature X, to get to it. There are lots of questions you might want answered.
Say you find 5 - 10 'paths.'m (eg google search for 'site.com'>homepage>pagex>login') You might want to see the general pie graph of which paths get used. Google analytics can probably handle this. But what if you want to know about 'users' using this path. Do users migrate from path A to path B over time? Do users form their path in the first visit & stick to it? Are users erratic in their choice of path? Do users of the site that joined after redesign (sub segment) use a different path. Are users of more advanced features (sub segment) more likely to use path X.
I wish you could get proper demographics of your visitors who decide to bounce. Right now you can figure out exactly who your users are, but there is no easy way to figure out what demographics don't get convinced by your message
i would like to know
1. how many sites before mine they were at
2. how happy i made the user
3. how angry i made the user
4. a full detail of their computer specs. so i can set my expectations on site performance
1. how many sites before mine they were at
2. how happy i made the user
3. how angry i made the user
4. a full detail of their computer specs. so i can set my expectations on site performance
Ask the users! I'd answer a poll at the bottom of your page... Where is your page?
Answers: 1. Way way too many (I'm to the stage of browsing that I'm posting comments! ehhe). 2. Great answer to the initial question so I'm pretty happy now. 3. Very very far from angry (see 2) 4. MacBook Pro (10.5.5) 2.4Ghz Core2 Duo 4GB RAM.
Answers: 1. Way way too many (I'm to the stage of browsing that I'm posting comments! ehhe). 2. Great answer to the initial question so I'm pretty happy now. 3. Very very far from angry (see 2) 4. MacBook Pro (10.5.5) 2.4Ghz Core2 Duo 4GB RAM.