When designing (UX/IA) a property search system what are some important things to remember?

Question

I'm designing the UX for a property search system. The platform is FSIS. I am basically looking for best practices in doing a Faceted Search while being able to incorporate many refiners based off the numerous data points available to me.

How do i validate what criteria should be a 'filter/refiner' or what should be left out? Likewise how do i determine what data point should be arrive from dropdowns, checkboxes (parallel) or links (filter)?

I've referenced Best Practices for Designing Faceted Search Filters, by Greg Nudelman, which is insightful however many things are evovling in search practices - likewise i am open to breaking search pattern paradigms when beneficial. For example, when is search criteria appropriate to place above results, ie. Yelp and when should it sit on the left or right of content, ie, google or kayak.

Answer 1

I guess the top positioning feels more like it's controlling the results from first use, where as left positioning feels more like filtering as an afterthought if you don't see what you want to start off with.

According to the Faceted Metadata for Information Architecture and Search ( http://flamenco.berkeley.edu/talks/chi_course06.pdf page 54) the lesson learnt at eBay was that facet controls placed along the top of the page are used more than when on the left side. There's a LOT of other interesting stuff in that presentation, but it's a bit short on detail.

This paper: UIs for Faceted Navigation Recent Advances and Remaining Open Problems ( http://people.ischool.berkeley.edu/~hearst/papers/hcir08.pdf ) may be relevant - but I'm not sure how recent is recent, nevertheless it's interesting reading and the WeFeelFine.org app is a bit funky, but ultimately useless :-)

Combining data visualisation and faceted search is an interesting development - via the concept of Elastic Lists ( http://moritz.stefaner.eu/projects/elastic-lists/ ). Interesting because they provide a visual representation of the weight that a facet carries in relation to the size of the data set, as well as other characteristics.

I think it's also very worthwhile looking at the mspace demo (Use the Try Demo button at the top right of http://mspace.fm/ ). It has some very interesting things going on. I like the ability to add/remove/reorder columns and drag in a new facet to the columns, and really tailor it to the way I want to search. The Back Highlighting feature is nice too and the ability to add items to a scratch pad might really suit your purpose?

Answer 2

I can speak from my own frustrations with certain faceted search engines out there. The first that comes to mind is NewEgg. I love NewEgg for the most part, but their guided search could be smarter. The number one complaint I have is the way they handle numerical data, and I've seen this elsewhere too.

When looking at SSD's on NewEgg I see these options...

This is almost useless. Any time you have a range of possible numerical values, the search engine should allow you to somehow specify the range you're looking for. There are three possible solutions which would all be better.

1. Create a min/max text entry or some sort of range control.
2. Create checkboxes next to each item so the user may select multiple options.
3. Change the options to 8+, 16+, 32+, and so on. This would be based on the basic assumption that the customer always wants more for less money. So if they're looking for 32GB drives, they could potentially buy a 40 or 48GB drive assuming it's not much more expensive. Another example of this can be seen at realtor.com where the number of bedrooms and square footage criteria are done in a similar fashion.

Links vs Checkboxes vs Dropdowns

In the simplest examples links will work just fine. They're simple for the user and even have advantages in SEO.

Checkboxes are useful like in the example above where it's desirable to select more than one option in a grouping to filter by. If you go down that road, I think it's helpful to commit to it like Realtor.com did. Their search is composed entirely of checkboxes, radio buttons, and textboxes (for a price range). They chose not to add links to the mix, as they don't really fit in with the other form elements.

Dropdowns should typically be a last choice when you're desperate for screen space, as they require two clicks to select anything.

P.S. Check out the Pivot Viewer for Netflix. It requires Silverlight. Checkboxes and numerical range controls are utilized where appropriate, and it works quite well.