You’ve done your research. You understand how consumers search for your products and services. Now, what do you do with this insight?
In my previous article about Giving Customers What They Want, I outlined an approach to creating a search behavior model that can be used to develop your information architecture and your website content strategy. In practice, these are important first steps required to develop a world-class website. Now that you have this insight, what do you do next?
A critical task is the evaluation and selection of a technology that enables the behavior model and maximizes the user’s search experience. There are tools that excel at maximizing the best human experience. These are known as faceted search engines.
You’ve seen them at work at Amazon.com and dozens of other shopping sites: search results are clustered by price, feature, color or other facet. These tools do a very good job with structured data (information about products), but they also do wonders with all that unstructured content on your website (i.e. primers and reviews). They can extract entities from content (i.e. people, places, things, titles, address etc.) and make them a navigational option during the search experience. These tools do a good job of blurring the distinction between browse and search, giving consumers the best of both worlds.
Last time, I built a consumer search behavior model for Home Improvement where I showed how to develop a content architecture using search behavior? I will use this model to illustrate how a landing page can be constructed to enable the power of faceted search. Here’s what the content architecture model looks like:
To illustrate this process, I’ve created three prototypes of a fictional landing page to show how this behavior model can be converted into navigational options for use by a faceted search engine.
In this first prototype, we introduce three functional elements—two of them are organizational in nature. The first element is the traditional search box that will let search relevancy dictate future navigational and content experiences. The second, which is an organizational view, allows consumers to drill down by clicking the terms Remodeling or Repairing to see contextual content from either perspective. These two terms are widely used in search phrases. The third organizational element allows you to expose content, products and services based upon the experience of the customer. This is known as faceted authority.
In the second prototype we introduce a navigational left-hand menu bar that organizes all the topics identified in the search behavior model into four major high-level categories. These categories provide the consumer a good feel for the products, services and content the website contains in a single glance.
- Browse Projects
- Browse Products
- Browse Information
- Company Directory
Not only is this menu organizational in nature, but these navigational categories can be manipulated in real-time. For example, a product manager has the option to reorder the display, or not display categories (or attributes) for any particular consumer search. The product manager can also define and add new attributes for categories in near real-time. I’ll talk more about this topic in the business rules section later on.
In this third prototype we introduce the final elements which are conversational entry points to the four major content categories. The user has the option to drill down into one of the four conversations that focuses content and the navigational choices around a single broad topic such as New Remodeling Projects.
- New Remodeling Project is an informational front to the Browse Projects section in the left hand menu. Since bathrooms is the number one search topic, it makes sense to call it out.
- Tools and Appliances is an informational view of the Browse Products category.
- Do It Yourself is a focused slice from the Browse Information category.
- Getting It Done is a call to action to use the Company Directory.
When you look at the above prototype you see that we have tied this landing page’s architecture to the search behavior observed in the previous article. This page reflects:
- The option to browse or search.
- Options to display content according to the user’s ability to handle DIY projects.
- The consumer’s wide-spread usage of Remodeling and Repairing in large numbers of search phrases.
- A menu structure that covers the entire range of topics found in the search behavior model.
- Information modules that expose content to aid a purchasing event across the four high-level categories of behavior. These modules provide options for experimenting with content / conversion ideas.
This is a data-driven approach to developing a landing page based on search behavior, but I caution you not to stop at this point. I do recommend that you conduct usability studies as a routine part of your website development process, because poor UI design can undo the advantages gained through understanding search behavior. In my experience, usability always uncovers a couple of real serious problems and a half dozen minor ones. Usability will help validate your architectural model.
A deeper look at faceted search
Now that we have examined an approach to using a search behavior model to develop a web page, let’s take a look in a little more detail what faceted search is, and why it’s not only valuable to consumers, but to you as well.
Faceted search, also known as guided or product search, is a technology approach that combines the best of browse and search into a singular experience that gives consumers complete control over search behavior. Users find information by doing a search, then navigating the results by category (e.g. Projects), by topics (e.g. Bathrooms), or by attributes (e.g. color, price or feature). And of course, users can continuously refine search results by using any combination of the above, while moving from browse mode to search mode and back to browse. This technology effectively exposes the most contextual information to consumers when they are in research / buy mode.
Next, the user can slice and dice the search results. They can search within a results set, they can sort by meta data attributes, or they can filter results by text, categories or attributes. This allows a consumer to look at your products from several vantage points and discard what they don’t care about.
Finally, the user has social options. They can share search results with their friends, save searches for future use, sign up for email alert notifications for price changes or when new products are introduced. These provide options for repeat engagements.
Bottom line: when customers use faceted search they get an organized search result that offers excellent contextual buying and product research experiences.
How faceted search benefits your organization
Faceted search technologies give product managers contextual control over the consumer search experience based on real-time behavioral feedback. Faceted search engines provide merchandizing tools that allow product managers to:
- Continuously fine tune the consumer experience with real-time analytics.
- Enable custom content and product offers based upon search behavior / history.
- Create new navigational paths and offer substitutions and promotions based upon search behavior.
- Provide more opportunities for up, down and cross-selling based on behavioral context.
An added benefit is that facets are so flexible that you do not have to invest in developing a large and bulky taxonomy to organize your products and content.
Business rules govern the consumer experience
The heart of engaging consumers in a two-way conversation is dependant upon your search tools having business rules. These rules govern how consumers interact with search results. Some of the more important functions are:
- Rules that trigger an action to be taken on a schedule. These are defined by the product manager. For example, you can define a different set of navigational options if the search happens over the weekend.
- Rules that define special offers tied to specific search terms, or to navigational events (e.g. pre-defined product categories). Rules can also be triggered when consumers view particular information such as an introductory primer for aspiring handymen.
- Rules that are tied to individual landing pages or classes of inventory. These are great for A/B testing.
- Rules that modify attribute weighting based on a category, product or context. For example, a rule could display only the music categories a customer has browsed in the past and emphasize music selections similar to what they purchased before.
- Rules that can present variable pricing based on customer history or profiles. For example, if the customer has never paid full price for anything, offer a 10% discount.
What to look for in faceted search systems
Besides the ability to search and navigate across a category and attribute, a good faceted search solution will also support natural language, wild card, fuzzy and phonetic search. They should also support searching on SKU and item numbers without prompting the user for this information. A good faceted search engine will know a SKU code when it is typed into the search box.
Faceted search tools should also support a full range of optimization options that allow product managers to improve search results. This includes the refinement of linguistic information working in conjunction with relevancy, precision and relaxation mechanisms. Let’s take a high-level look at some of the more important functionality available to you for improving search results.
- Question analysis supports query expansion by making suggestions and offering associations as navigational choices (synonyms for example).
- Real-time spell correction seems pretty obvious, but a typo is not a typo if the consumer does not realize that they have misspelled the word. In the consumer’s mind, the crummy search results are your fault.
- Ambiguity resolution is where you offer search results that are disambiguated as navigation choices. For example a search on “flooring” would produce navigational choices such as bathroom floors, kitchen floors, living room floors, wood floors, tile floors and carpeting. These choices can be serendipitous, exposing consumers to options they may not have thought of.
- Finally, when confronted with meager search results, product managers can manipulate thresholds for increasing quality results by:
- Relaxing minimum word frequency requirements.
- Expanding the proximity range (how close words are to each other).
- Relaxing the quality of associations (increase the pool of related terms).
- Exposing the “did you mean?” option.
- Exposing more navigational attributes.
There are a few things that you should understand about faceted search technologies. First, consumers who just use the search option can see results that are different than users who use only the navigational options. Also, these tools can be less effective if you do not have a critical mass of content. If you sell a very limited range of products with little supporting content, then these search tools will provide very little value.
In theory everything can be a navigational facet. When you first parse your content through the indexer, the search engine will choose facets for you. You will have to review this list, delete what you don’t want and add facets you designate as important. You will have to experiment for a period of time to develop the right mix of facets for your customers.
There is a lot more to these faceted search engines than I have covered here, but this provides a good high-level overview of a process to map your search behavior model to a custom landing page supported by faceted search tools. This technology will be a good fit for you if:
- You have a good handle on your customers’ search behavior and search history.
- If you want to empower your customers.
- If you have critical mass in product and content.
- If you have the staff to implement and manipulate your business rules.
Several vendors have very robust feature sets that give you very precise control over the consumer search experience. The tools from Endeca, Vivisimo and Microsoft Fast are well known, but can cost you hundreds of thousands of dollars to license.
There are other options that are more cost effective. You can download Lucene and Solr for free. This highly regarded open source solution gives you the basic faceted search technology, but you don’t get any merchandizing tools. If you go this route you will need engineers that really understand search technologies.
There is a third option to consider. There are a number of vendors who have added faceted navigation to their search tools over the past five years. Two vendors that have very good technology and are reasonably priced are Attivio.com and LucidImagination.com.
Lucid Imagination has taken Lucene and Solr and has productized the technology and offers professional services in support of the license.
Attivio has developed a very unique approach to dealing with content in interesting new ways. Check out their CIA Fact Book or Baseball demos. These demos will illustrate how they are pushing the boundaries of facets in search.
Opinions expressed in the article are those of the guest author and not necessarily Search Engine Land.