We define enterprise search and navigation as the technologies companies deploy to guide a seeker to the information that will resolve her question, as well as support information owners' efforts to deliver adequate quality of seeker experience. Enterprise search is used not only on corporate Web sites, but also on public-facing Web sites, customer and partner portals, and all business applications. Since 2003, we have been using our earlier evaluation frameworks to assess vendors' offerings for search and navigation. To date, we've assessed offerings from ATG, Autonomy, Celebros, FAST, Google, InQuira, IBM, KNOVA, Mercado, Microsoft, Northern Light, Oracle, Thunderstone, and WebSideStory, as well as solutions from EasyAsk (acquired by Progress Software), iPhrase (now IBM), and Verity (now Autonomy).
  
In this report, we evaluate Endeca Information Access Platform 5.1, which was released in April 2007.
  
Endeca Information Access Platform's chief strengths are its upgraded navigation engine, which supports the complex relationships inherent in search over structured data; it's expertise in retail, which has enabled Endeca to win and maintain a strong showing in the U.S. market; and the data aggregation and cleansing capabilities provided by the information transformation layer (ITL), which enable information managers to improve content without incurring the expense of altering source data.
  
Endeca IAP's chief weakness is its management interfaces for business users such as merchandisers and information collection managers. Web Studio and Developer Studio were leading products a few years ago, but competition has outstripped them. Endeca needs to invest in creating a single console for the business user that is designed for all the tasks that user needs to perform. Lack of support for federated search is another significant weakness in Endeca's enterprise and media markets. Federated search, by which we mean the ability to invoke other search engines and incorporate the results into a single set, is increasingly a requirement in ecommerce sites as well. 
  
Keep in mind that we developed our criteria to evaluate search solutions that span the enterprise: ecommerce, customer service, knowledgebase, help desk, intranets, customer and partner portals, and applications that require unstructured data. Solutions that we analyze are often targeted for only one or two of those uses. Endeca is targeted at enterprise search, retail and B2B ecommerce, search on media and publishing Web sites, and knowledge management.

OVERVIEW OF ENDECA AND INFORMATION ACCESS PLATFORM

Endeca Technologies is headquartered in Cambridge, Massachusetts, and has offices in the United States, Europe, and Asia Pacific. It was founded in 1999 and currently has more than 350 employees, supporting more than 400 customers. About 40 percent of these customers use Endeca for ecommerce applications, including the top three United States retailers: Wal-Mart, Home Depot, and Costco. Endeca customers also include 50 of the Fortune 500, among them American Express, IBM, Tesco, Saks Fifth Avenue, Office Depot, and John Deere. Endeca has two U.S. patents, on Guided Navigation (USPTO 7,035,864) and disambiguating multiword search terms (USPTO 7,062,483). Additionally, Endeca has eight patents pending for its Information Access Platform; its guided navigation, query, and spell correction capabilities; and its merchandising approach.

IAP’s new support for complex data relationships enables multi-dimensional navigation in addition to successive refinement across multiple record types.
Endeca has shifted branding, and, to some extent, strategy, from its focus on discrete application offerings for ecommerce (InFront) and enterprise search (ProFind). It now promotes its technology as a platform for search throughout the enterprise and brands its technology Endeca Information Access Platform (IAP). It continues to market solutions for ecommerce and enterprise search, but the old product names, InFront and ProFind, have gone by the wayside.

Perhaps this shift in branding was unavoidable, as Endeca sought to broaden its market to publishing, financial services, and knowledge management. These applications don’t fit well into either enterprise search or ecommerce, and yet adding new products and creating new brands for each customer segment is a very expensive approach. Jettisoning the product brands is not a great loss, since most customers tend to think they bought and use “Endeca.” We’ve never heard InFront or ProFind on a customer’s lips. The downside in the branding shift is the peer group Endeca joins. Enterprise search platforms are the playing field for Oracle, IBM, and Microsoft, heavyweights with immense incumbency in the enterprise. Competing with these behemoths for strategic investments is a tough battle. Fortunately for Endeca, many search technology decisions are made in stealth before enterprise IT and their enterprise software providers get wind of the deal. And, in many companies, search isn’t yet considered a strategic asset. (Go figure.)

Enhancements Since V.4.7

Our last review of an Endeca product was InFront V.4.7 in October 2005. Endeca has significant new features in three releases since then, IAP V. 4.8, IAP V. 5.0, and IAP V. 5.1.

Enhancements to IAP V. 4.8 include support for creating merchandising rules with multiple triggers, improved keyword redirect configuration, a more flexible crawler for Web content, and performance improvements for incremental indexing.

Enhancements to IAP V. 5.0 include a Web Services API; Relationship Discovery to automatically extract entities, terms and clusters; new and also improved adapters for Documentum, salesforce.com, WebSphere Product Center and Alphablox; and improved workflow in Web Studio.

Enhancements to IAP V. 5.1 include major enhancements to Guided Navigation, which now accommodates more complex data relationships; 64 bit support in the platform to increase scalability both in number of records and speed of processing; new deployment templates, which include the operational components such as scripts for Web or file share crawling and indexing; and Web Studio improvements for business rule configuration options and management of multiple deployments through one interface.

The most significant of the V 5.1 enhancements is in Guided Navigation. IAP’s new support for complex data relationships enables multi-dimensional navigation in addition to successive refinement across multiple record types such as products or stores. In successive refinement, each choice causes items to drop away from the set of results. Endeca’s updated navigation allows you to navigate from any item through any of its relationships, enabling a more complete view of the data set. For example, a user can navigate to a subset of products of interest, then pivot to see related reviews, and then refine the reviews themselves by their characteristics, simultaneously refining the product results still further. It also enables fast retrieval of results for very complex queries. These last two capabilities are fundamental to supporting search over relational data, or, as it’s often called, business intelligence search.

Enterprise search, in our parlance, is search and navigation technology deployed within a corporation: portals; intranets; public Web sites; ecommerce sites; knowledgebases; and customer service, CRM, and other applications.

EXAMPLE: CIRCUITCITY.COM. I’ve been getting a nasty fuzzy sound from my stereo. Not being a technophile, I conclude that I should replace the oldest piece of equipment in the stack, in this case, the amplifier. At Circuitcity.com, I find a very well-organized array of choices. Home theater, accessories, satellite radio: no, no, no. Home audio: looks just like the thing I want to replace. There are seven items in the home audio category. They fit on a page, which means I can look at them all together. I make a decision in minutes. It’s a fabulous experience.

There are a few elements that make this experience work so well. Most important, in my view, is the way search results are grouped and presented as navigation options, and the way these groupings are illustrated and repeated. Presenting navigation options in the left hand column is a familiar convention, and one place the eye immediately goes when presented with a busy page. The pictures which illustrate each category, presumably the best-seller in each category, offer an alternative target for your attention that presents considerably more information than the list of words in the left hand column. Here are not only picture of products to help you see what the categories mean, but a very easy way to either look at the choices in the category or to select the featured product and skip the comparisons. Circuit City does a great job with search and navigation. (See Illustration 1.)

Circuit City Example

Illustration 1. At circuitcity.com, search results are presented as a set of examples for each of the product categories represented. The seeker can click on the navigation options listed in the column to the left, click on the same options presented as pictures in center of the page, or select one of the products illustrated. The seeker thus has a streamlined path, both in terms of the number of clicks and in terms of understanding his options

RECAP OF ENTERPRISE SEARCH DEFINITION AND REQUIREMENTS

Definitions

Search, in our view, includes navigation. To the seeker, they are inseparable tools, and, from an information manager’s perspective, both techniques must be employed consistently to deliver high-quality results. From a technology standpoint, navigation should—or at least could—be driven by search.

Enterprise search, in our parlance, is search and navigation technology deployed within a corporation: portals; intranets; public Web sites; ecommerce sites; knowledgebases; and customer service, CRM, and other applications. Enterprise search solutions also provide tools for people who manage findability for information collections and the quality of seeker experience for customers, employees, partners, and suppliers. The solutions offer approaches to organizing search results, tuning the relevance of search results, and customizing the presentation.