Start mail me! sindicaci;ón

Search and Browse

Today I watched a really great presentation by Peter Morville and Mark Burrell at UIE discussing search patterns. I have to admit that the only reason why I attended is because Peter was speaking and I always love what he has to say, because I very rarely have to actually design search interfaces.

After the presentation I actually started asking myself why the hell is it that I so rarely have to design for search behaviors. The reality is that oftentimes I’m designing for existing services where search is an existing capability and iterating it is never in scope.

One of the problems with that, which became more apparent to me after the presentation, is that treating search as a separate behavior from browse is really misguided. I thought about this problem before but could not quite articulate it very well until today.

Historically I had been taught and understood search and browse as distinct elements – which they are visually and from a UI elements standpoint – but from a behavioral perspective, they really are not, rather, they are part of a continuum. A spectrum of discovery behaviors if you will.

Browse-search spectrum

If we think, for example, about how faceted classification emerges in search interfaces and in browsing interfaces it becomes really clear how intertwined they are.

One of my questions to Peter during the presentation (which unfortunately did not get addressed but hopefully will be part of the UIE follow-up podcast) was if he had identified patterns of use of faceted search and if there were any emergent patterns that could help answer if faceted search is more appropriate for a particular kind of content or context — and when it might not be appropriate.

Faceted browse/search is a hot topic at work and I feel like it’s been historically a random requirement that ends up on a project brief because of process inertia. Someone saw it somewhere and thought it was cool so decided that it should be applied to the kind of content we are surfacing for our audience.

I have no good evidence to substantiate my hypothesis at this point (unless lack of examples in the wild is enough), but I suspect that for our content – namely video content, generally in the entertainment realm, frequently movies, series and other TV programs – having faceted search as a primary tool for discovery is really inappropriate.

I have definitely seen and appreciated the application in e-commerce and feel like there is a prevalent pattern there for its use. But on the content I design for, I just don’t know. If I am to rely on what I know from user behavior learned observing people try and get to the video content they want (across different platforms in a number of distinct scenarios of use) the attributes they need to make decisions are frequently few. The variation in behavior is little in terms of user motivation, and greater in content type (i.e.: people look for movies differently from how they look for series).

How can I make a compelling argument that this particular pattern is not the right fit when I am not sure what is? I’ve seen it fail in usability tests but that only makes people try to fix it and improve it, not to try a completely alternate solution that might be appropriate. Any ideas out there?

Also, I’m not on a crusade against faceted search, I am just looking for ways to 1) articulate that there might be a problem picking this particular pattern 2) explore other ways to do it (both in the context of use and content I described). Any ideas are welcome.

Regardless, I think it will help me in the future to frame the scope of what I need to design for when dealing with content discovery behaviors by thinking about them in the browse-search spectrum. At least I expect that to give me a better argument to combat feature requirements void of context.

Gravatar

Erin Lynn Young said,

January 12, 2010 @ 7:22 pm

I sympathize with your hypothesis about faceted search. I’ve had many opportunities to design faceted (or otherwise filterable) searches, and I think the difference you mention between users exploring content those in search of items is fair.

Facets support refinement. They help users take a long list and narrow it by attribute to something small enough to analyze. Into something that meets the user’s specifications — Shoe size. Lens speed. You name it.

In consuming content, I think users with attribute specifications are far less frequent. We saw this as tag clouds were all the rave there for a while, but got little use.

Which is not to saw that the user’s don’t exist. It really depends on the application. Users who return to find previously accessed content might filter by specs. Design pattern libraries are another example of a list where facets are beneficial. And, are there any users for whom only a specific type of content is relevant? (For instance, press may wants to see only news.)

If your use case analysis didn’t lead you to any user types who are looking to find a piece of content against certain specifications, I’d feel safe in your assumption that facets are not of value in the experience you’re planning.

Hope that helps.

Gravatar

Renato Feijó said,

January 12, 2010 @ 8:08 pm

I guess the best way to understand whether faceted navigation (or search, if you prefer) is appropriate or not to your design is to look at the problem from your audiences’ perspective. This may sound obvious, but faceted navigation make task completion for targeted, carefully chosen user groups more streamlined. It does that by creating high-speed lanes through which people get to what they are looking for, usually a small set of common tasks. It’s also very helpful when you need to cater for very different use of terminology, a trivial thing in government web sites, for instance.

To be successful at this exercise, you need to thoroughly understand your audience segments (relevant to interaction design purposes) and their goals. The best way to do this is to model personas (which will roughly represent one of those audience segments), collecting and documenting attitudes, behaviours and goals and other design-relevant information.

Your IA facets should support key goals of the various personas. As for the patterns that support faceted navigation, many types of fly-out menus and combos can be used. I don’t think that there is a standard or conventional implementation at this point in time, if that is what you’re looking for.

Designing faceted navigation is a difficult endeavour – and can be tricky in politically charged environments – but definitely worth the effort, if you are dealing with massive amounts of content that is constantly updated, and very distinct audiences.

Boa sorte.

Gravatar

Stephanie Lemieux said,

January 13, 2010 @ 9:59 am

I’m currently working on a digital asset management project, and we are definitely seeing an application for faceted search for images and videos. It is a global pharmaceutical, so these brand managers are interested in browsing marketing and training resources (including videos) by various aspects: brand, market/country, language, subject, etc.

So, while I agree that faceted search is not appropriate for every context, I would not generalize that it’s not appropriate for videos. But I also agree, having built some retail taxonomies that touch on movies & tv shows, that there are few truly meaningful facets for most users in those categories. Genre (action, drama, etc.) tends to be the primary element, or recency (coming soon, new release, etc.). If it’s tangibles, then format reigns supreme (DVD vs. Blu-ray).

Gravatar

Livia said,

January 13, 2010 @ 10:23 am

@erin “In consuming content, I think users with attribute specifications are far less frequent” That is my understanding from observing people look for things to watch across platforms. What has been difficult though is to segment the audience by behavior because pretty much everyone shows all the core behaviors at one point or another. It’s a problem I need to solve, it’s just been hard.

“And, are there any users for whom only a specific type of content is relevant? (For instance, press may wants to see only news.) ” – I’m designing for a general audience, so basically anyone who watches TV, uses a DVR and watches movies and tv shows online. It’s really too broad (which is why I am having trouble segmenting).

I do know that in the scenarios I have identified there are distinct behaviors and expectations based on content type (for example, when people want to go online to catch up on a tv show they missed, they expect to be able to look for it by the day of the week that it airs, while someone who is looking for movies to watch from their OnDemand library is more likely to look for genres or the premium channel they subscribe to. These are clear attributes that make sense to allow people to filter the content by. I guess my concern is how to not overwhelm the user — because even if I use the faceted browse/search pattern with these attributes, they are REALLY not scalable. Take the channel filter for example: 200 channels! This and other reasons are why I’m debating this approach.

And great point about tag clouds!

Gravatar

Livia said,

January 13, 2010 @ 10:29 am

@renato I see your point, and as I stated above, it’s been difficult to segment this audience’s behaviors because so many people express so many of the same behaviors. There is almost too much commonality. In a way, that should make life easier because it would allow me to tackle the most common needs, at the same time, the attributes that people need in their decision-making process to choose this content are too numerous, which is why I’m debating HOW to implement them (what pattern is most appropriate). That and the fact that some people are enamored with faceted search/browse but can’t articulate to me why it’s the best fit for this context.

Gravatar

Livia said,

January 13, 2010 @ 10:41 am

@stephanie You bring up a good point. When I started working with this I thought that “video content” was the distinct way to look at this, but the reality is that the fact that the content type being video does not matter – what matters is the actual content genre – tv shows x live sports events x movies, etc. So I agree, I am definitely not generalizing that faceted search is not appropriate for videos. As for “there are few truly meaningful facets for most users in those categories” I could not agree more. And perhaps the answer to my problem is to argue against introducing so many attributes and keeping only core ones we observed as key to the decision-making process (as I noted in the comment above). I still am struggling with UI aspect of facets because of the scalability issue.

Gravatar

Dan Blaker said,

January 13, 2010 @ 1:42 pm

Faceted search is useful mainly (exclusively?) in cases where the user needs to exclude entire categories of search results (e.g., all shoe sizes other than 11). For finding a particular item (i.e. needle-in-a-haystack searches), it’s detrimental compared to browsing (because you’re actually hiding part of the haystack) and slower than full-search (because it takes more time and thought to think about facets than it does to scan and click “next page”). For discovery-oriented searching, facets are less interesting than excerpts and once again you might be hiding something (based on your categorization choices) that the user actually wants to see.

Another effective argument against it (in some organizations) is that Microsoft has focused heavily on faceted search in the new version of SharePoint (2010). In my experience, when MS chooses the left fork in the UI road your best bet is to take the fork on the right…

Gravatar

Greg Nudelman said,

January 13, 2010 @ 3:20 pm

I hope my new Integrated Faceted Breadcrumb design might help take a step at unifying search and browse. Look for the upcoming Boxes & Arrows article — I’ll post a link as soon as it is available.

Great Seminar!

Gravatar

Livia said,

January 13, 2010 @ 4:01 pm

@dan I can certainly appreciate your humor with Microsoft’s choices :) Thanks for the argument about applicability of facets, it’s helpful in terms of articulating pros and cons for my context.

@greg Excellent, looking forward to it!

Gravatar

Search Facets » How to sell out a virtual seminar said,

January 13, 2010 @ 4:03 pm

[...] when you look at software features instead of user tasks.  Livia Labate has a good follow-on discussion. Posted on January 13, 2010 at 8:08 pm by PGusBell · Permalink In: IA, [...]

Gravatar

Erin Lynn Young said,

January 13, 2010 @ 4:08 pm

Livia,

Consider the place of the global nav in addition to the facets.

You could align the global nav with main use cases (ie, Recorded TV) and then surface the most commonly used specifications for that use case at top with a +more to unfold additional specs.

Or you could align your global nav with most popular facets.

Just a thought, not understanding the details.

Gravatar

Livia said,

January 13, 2010 @ 4:25 pm

@erin yeah, this discussion is definitely happening in context with what is going on with the global nav.

We unfortunately suffer from accessities, which is a condition that affects product manager and makes them think that a user should be able to reach everything all of the time from anywhere. We are working on a cure. In the meantime, yes, I am trying to articulate the scenarios of use (which is being very difficult because of the breadth of audience and services) so I can map the specific (facets) for content discovery in a way that it complements the global nav.

I’m frustrated with this challenge because the more I need to change globally to fix this locally, harder it is for me to actually get this done, so I was trying to make as much improvement as possible without proposing much change to the global elements, but that is starting to work against me.

Good times!

Gravatar

dan klyn said,

January 13, 2010 @ 5:14 pm

anecdotally, the usefulness and usability of facetted navigation is proportional to the number of facetting dimensions your content can be mapped to. facetted nav is so powerful that it can sometimes reduce the universe of results too much and too soon. an ecommerce site with 10s of facetting dimensions gets narrowed less powerfully and less rapidly than a site with only a handful of dimensions.

there’s definitely a problem when you present too many facetting dimensions and/or when you present the sequence of facetting dimensions wrongly… but fwiw I think the pattern fails more and harder when there are too few dimensions than when there are too many.

if original air date, channel and genre are the primary (or only) relevant facetting dimensions available in Comcastland, the facetted nav pattern may be an unwise choice. things could get too narrow too quickly. there’s something about a gridded view of what’s on and being able to see adjacent options that you’d lose pretty quickly after allowing customer to narrow too much too soon.

i’m curious to know: is facetted nav being discussed as a replacement for a fixed categorical sort of navigation scheme?

the other thing that looms large for me when I think about facetted navigation is whether or not the site operator has a tool to sculpt the results set that comes back when user clicks on a given facet. if the system being considered allows business rules to govern the ordering of the results returned for a given facet, that’s ideal. one of the arguments for a fixed categorical nav scheme is that it allows us more and better control of the choreography of a user’s visit. bizrules-governed results sets give site operators almost as much control of that choreography of what users see and when as fixed categorical navigation …

Gravatar

Tweets that mention I think therefore IA (Livia Labate) » Search and Browse -- Topsy.com said,

January 14, 2010 @ 1:46 am

[...] This post was mentioned on Twitter by Jared M. Spool, Livia Labate, Kurren, UIE, Dr.Enton and others. Dr.Enton said: RT @jmspool: Nice blog post about using faceted search by @livlab, inspired by yesterday's #uievs with Peter Morville. http://bit.ly/575XIi [...]

Gravatar

When Is Faceted Search Appropriate? said,

January 15, 2010 @ 2:26 am

[...] reading reactions to the seminar, I was particularly intrigued by a post entitled “Search and Browse” by Livia Labate on her fantastically named blog, “I think, therefore IA“. She [...]

Gravatar

Livia said,

January 15, 2010 @ 1:34 pm

Check out this great post by Daniel Tunklang reacting to this discussion: http://thenoisychannel.com/2010/01/15/when-is-faceted-search-appropriate/

Gravatar

Ricerca e navigazione: due facce della stessa medaglia - Alberto Mucignat said,

January 17, 2010 @ 9:53 am

[...] Labate ha scritto un post molto stimolante a proposito di Ricerca e navigazione. In particolare, mi è rimasta impressa questa frase: Historically I had been taught and understood [...]

Gravatar

Browse Is The New Black « Experiencing Information said,

February 7, 2010 @ 10:31 am

[...] seems to agree with this thought in her post “Search and Browse.” And of course, this is a key goal of faceted navigation: the increase a person’s [...]

RSS feed for comments on this post · TrackBack URI

Leave a Comment