Much has been said about the benefits of faceted navigation in e-commerce. Less has been said about the search marketing ramifications a�� most recently touched upon over here by Rand Fishkin at SEOmoz.
Many of the “big boys” have already rolled out their facet implementations. This includes such familiar brands as Home Depot, B&H Photo/Video, and J&R. NewEgg was one of the earliest to implement this advance. With good reason. Users don’t enjoy navigating through hierarchies of organizational pages on large e-commerce web sites a�� where, at any point they might click the wrong link. Retailers may enjoy this just as little. More and more web surfers using the hierarchical web directories on Web 1.0 sites aimed for the search box, while computer scientists created a working implementation of keyword search.
Unfortunately, even a well-implemented internal keyword search may leave the user with search “noise,” perhaps a bunch of quasi-relevant accessories, and a vague feeling that this may not be the complete list. When and if keyword search does find relevant products, it does nothing to further aid the user in the decision-making process. What if she wants a red one a�� does she modify her query? Will doing that show her only red ones via “boolean and or all results”? Did she even use the correct words? Maybe not.
In other words, the search box does less for the “long tail” aspects of products. After establishing that she wants a digital camera, the user might wonder about brand or price as the next deciding factors. Other factors include color, resolution, LCD size, waterproofing, etc. Most websites offer us price and brand as facets, but indeed they are only two of many facets a user might want to help her make a decision. Increasingly, websites are offering more than just price and brand filters to the end of improving user experience.
It comes down to this a�� faceted navigation makes it easier to find things; and consumers buy more when they readily find them.
But what about the bots?
As search marketers, we’ve probably considered exposing at least the brand facet as a landing page of the form: [brand] + [category]. What we may not have thought about is that “[brand]” is one of myriad facets. Both users and bots would benefit from simple, well-presented paths to explore these facets.
Almost all websites use categories and subcategories in their navigation scheme. Navigating to the rather general category of, for example, “cameras” on the B&H website presents a decision between a few subcategories like so:
Unfortunately, since B&H’s inventory fills up a city block, it has a total of 558 digital camera models to choose from. This leads to information overload a�� something that we learn early on is not conducive in converting our window-shoppers to buyers.
This is where the category-based navigation ends for B&H. Categories are just as stone age as the “Internet Phonebook” in flexibility for some aspects of decision-making. For those aspects, combining a series of dimensions a�� or facets a�� may work better.
For example, if you are looking for a yellow digital camera with 8-9 megapixel resolution, you can find it with the facets on B&H:
This type of decision-making process is not facilitated by the rigid tree structure of the typical category system. A tree presumes the exact process by which the consumer makes her decisions. It must be traversed in a straight line: for example, “Cameras A� Digital Cameras A� 8 Megapixel.” But what if the potential buyer has placed color before resolution in their priority list a�� yellow or bust? Categories also cannot gracefully handle the decision regarding two resolutions: a union a�� combined with the rest of her selections.
Facets rise to the task. This synthesis of a shallower category tree with a series of facets may be the best analog of a knowledgeable salesman discovered yet. He doesn’t even take lunch.
Unfortunately, he’s also a potentially giant spider trap.
The debate between category classification and faceted classification is actually a very old one, and stems from the discipline of library science. Dr. Shiyali Ranganathan introduced a largely ignored alternative to the categorical approach of the Library of Congress and Dewey Decimal systems of classification in 1933. He used 5 fixed facets: personality, matter, energy, space, and time. He suggested his system adapted better to information than a category-based system. This more elegant system still dominates in India. More than 75 years later the debate lives on outside the library, on the internet a�� an even larger collection of information. Faceted navigation in e-commerce is, in fact, a modern adaptation of Dr. Ranganathan’s contributions to library and computer science.
Facets can help to create a set of legitimate landing pages that may complement subcategory pages a�� for users as well as search engines. In fact, one could flip-flop on whether some facets are subcategories or vice versa.
Implemented carefully, exposing some facets will create effective landing pages for longer tail search queries that categories would address in only contrived ways. However, attention must be paid to the word “some.” Without care, faceted navigation may create a giant spider trap with seemingly infinite permutations of the same products. In fact, some faceted search implementations a�� B&H included a�� address this concern by excluding their faceted navigation pages from search engines entirely. Rand proposed just that as a solution over here. Unfortunately this is not really a solution, and anything but optimal a�� as one might expect after noticing the overlap in function between subcategories and facets.
The next part in this series will address the relationship between facets and search marketing, the long tail, and duplicate content. A final article (to be published in the upcoming Summer issue of Search Marketing Standard print magazine) will address other optimizations and implementations. Finally, a comprehensive treatment of these topics will appear in the 2nd edition of Search Engine Optimization with PHP.