Last year I designed a prototype of an advanced search system for Student Job Search (SJS).
I’d taken inspiration from the iTunes ‘smart playlist’ feature and created an ultra flexible tool for students to add multiple layers of search phrases and criteria.
Then a few months ago SJS asked us to build the search for real.
I took another look at the prototype and considered our student primary persona (we call her ‘Mira’).
I also looked at the SJS job listings which usually number under 5000 at any one time (SJS have a constant churn of temporary jobs, which makes them unique from other job sites such as Seek and Trade Me Jobs).
There’s no doubt the search tool I’d originally envisaged could be very cool, and based on our previous research I had little doubt that most students would work out how to use it.
But what bugged me is that the tool was perhaps too powerful – students would be able to create such refined search criteria that they would narrow their results too much and therefore be blinded to potential jobs (which might not quite meet their exacting criteria).
Then I did something I rarely do – I told the client “I was wrong”.
…I explained that I my design was too complex and that I could design them an even better and equally powerful search experience for students…
So they left me to it.
My new ‘advanced search’ design doesn’t include any additional search boxes, dropdowns or widgets.
You don’t even see an ‘advanced search’ link!
My solution uses one-click filters which can be accessed from the ‘quick search’ results or the ‘browse jobs’ feature.
The student can perform a search in the usual way and then incrementally refine the search results – for example…
-
Search by keyword, category or region
-
Filter above results to view jobs paying $25/hr and over
-
Filter above results to view jobs 10-20hrs per week
- Filter above results to view jobs starting within 7 days
The system works beautifully, is more usable than my original design – and you can never get a null result when using the filters…
It almost makes we wish I was a student again! (it’s interesting to see which jobs pay over $25/hr – those students are in hot demand!)
Other examples of similar filters in action can be viewed on wine.com and amazon.com
3 responses to “Filtering Out Advanced Search”
This is a pretty perfect feature, well designed. The key it for me is that you give information on the number of search results before clicking on a filter. The user can do a cost benifit analysis of a click.
This could be improved by reducing the cost end of the equation; filters should not have to reload the page, most of the data should come down the line with the first set of search results.
On an aesthetic note, that yellow banner ad on the left is overblowing the cool new functionality.
It’s interesting to see that SEEK (http://www.seek.co.nz) have just launched the same feature as the one we did for SJS a month or so ago – I don’t know if this is just coincidence, but someone out there is obviously tapping to same funky filtered beats…
How do you keep this fast? Essentially aren’t you re-querying your database on every filter click and also getting the counts for every permutation? How do you keep it fast?