Thanks for the feedback! In its current implementation, the widget actually only displays other articles on the installed site (that way it's more useful to the site owner). And each visitor's arrival only affects the widget as it displays to them -- they don't affect how it appears for other users.
edit: couldn't resist temptation to play around with it :)
I see how it works now, that's nifty. So, basically you're taking the search query, you retrieve the search terms and then you fish up a bunch of urls on the local server to try to capture the user longer.
This should have a measurable effect on time-on-site and bounce-rate stats.
You should do an A/B test for sets of pages with and without the widget to see if those variables change measurably (outside of statistical noise range).
That way you can get yourself a unique selling point.
Comments
Thanks for the feedback! In its current implementation, the widget actually only displays other articles on the installed site (that way it's more useful to the site owner). And each visitor's arrival only affects the widget as it displays to them -- they don't affect how it appears for other users.
There's a demo here: http://www.searchekko.com/demo.
Ah ok, that makes it a bit more solid.
edit: couldn't resist temptation to play around with it :)
I see how it works now, that's nifty. So, basically you're taking the search query, you retrieve the search terms and then you fish up a bunch of urls on the local server to try to capture the user longer.
This should have a measurable effect on time-on-site and bounce-rate stats.
You should do an A/B test for sets of pages with and without the widget to see if those variables change measurably (outside of statistical noise range).
That way you can get yourself a unique selling point.
Yep, I definitely hope so. Just emailed you too