Ask HN: Best way to visualize / analyze website stats in realtime?
Say I'm running an e-commerce website.
I want to be able to (data updated once a minute):
- See how many orders we're getting. I want to be able to answer questions like "are we taking more orders this evening than normal?" and "what hour do we make the most profit each day?"
- How many people people did X? (where X could be commenting, playing games, registering, tweeted about us, etc)
There's a lot of analytics software up there, but not many let you break data down into minute or hour chunks of time. Also, we'd want to be able to customize what data is being monitored.
We already use Chartbeat and Google Analytics. Chartbeat doesn't let you customize data (so we couldn't see how much profit we make per hour, for example), and Analytics isn't realtime at all.
15 comments
[ 4.0 ms ] story [ 43.8 ms ] threadi.e. we'd want to ask "how many t-shirts did we sell yesterday".
Needing to wait a day (as in Google Analytics) would be too much.
You can use a script on your own server or a YQL query to create the file for custom data and we pull it at regular intervals into your status board. Earlier this week we added support for custom chart widgets.
We're in an invite freeze period for our private beta right now, but if you mail me (details in my profile) I'll send yours out.
Just to re-iterate; we don't help you gather stats; we display stats you've already got and want to visualise in near real time.
However, we don't store data (besides temporarily caching it) so if you want to view history it has to be present in the existing feed.
The first type use logic which we have defined up front to extract the appropriate information for display. We apply little or no logic on the custom data feeds as that is assumed to have been carried out by the provider, usually when the XML/JSON file is generated.
If you have an obscure source of data, or it's sourced from a service we don't yet extract from, then I'd suggest writing a script to extract and massage the data into a format we accept and point your dashboard at it. We have definitions and example scripts in our support forum and I'm more than happy to give you a hand setting it up if that would be useful for you.
Also, check out http://www.metricly.com and http://chart.io.