"Dear Mike," began an email from a co-worker, "I'm trying to find the latest information on how we support ISO standards. Do you know where this information is located, and can you direct me to it?"
This is probably something you've heard before, and it seems like a no-brainer. But in all likelihood, it's not. Even if you were able to locate this information, it would probably take you just as long to find it as your inquisitive co-worker or anybody else in the company would. Why? Because all of you rely on the same, poorly tuned, ineffective search engines.
But, imagine for a moment, that the information you were looking for was easy to find. Searching on your intranet or extranet should be as simple as something you probably already take for granted: online banking. When you log on to your banking site, you can instantly view your account, and with a couple of clicks "magically" view everything you want or need to know about your account, including your balance, standing orders, and the last three checks you wrote. That's because your banking institution was kind enough to build a system that enables you to access the information you need at your fingertips. Why? Because your bank recognises the importance of delivering relevant, timely information to you, their valued customer, in the right context when you need it. Why? Because if they didn't, they would risk losing your business. For today's banks, personalised search is not an option; it's a requirement of doing business.
This leads us to the obvious question: Why then are our average, everyday search attempts so frustrating? The answer is more complicated than you may think.
To begin to unravel the conundrum, we need to understand the source of the problem better. Some pieces of information are easier to find than others because some content is structured, and some isn't. There are also different ways of data structuring that are more effective than others. Keep in mind that when we consider the vast expanse of information that exists in the world out there, our expectations are actually pretty high.
The reality is that search technologies must be able to decipher an extremely large amount of unstructured content, which may exist anywhere on any computer in the world, including our own laptops. Nevertheless, we, as users, expect immediate, accurate results when we type a simple question or keyword. When we take into account the complexity of this process, it's actually amazing that we are able to find anything. And in most instances, the content we're looking for has no information or metadata attached to it. In other words the search engine often has no way of knowing the origin of the information, who published it, what key topics it refers to or what its relationship is to other content. Clearly, the task of finding and delivering unstructured information is considerably more complicated than accessing our banking information.
This, in a nutshell, is why structure is so important. Once structure has been applied to content, the content can be searched more effectively. That's where metadata comes in. Metadata provides structure by acting as information tags that are applied to content. And a well-designed metadata framework is key to improving the performance and accuracy of search engines, not to mention all of the other delivery platforms, like portals, personalisation engines, syndication engines, and other content-rich applications.
An example of how metadata works: Say a sales manager for a large manufacturing company realises he needs to get a better handle on his customers. He logs on to the Web site of a software vendor and types "customer tracking software" in the search box. Although, the software vendor he is considering doesn't offer anything by this name, its metadata-driven search engine understands that this request is conceptually the same as a type of software that it does offer, customer relationship management (CRM), and provides him with a link to its CRM product section.So how do you move your organisation to this kind of framework? First you have to know what an ideal system and process look like. There are four different things you should consider. The prescription for search success 1. Identify your goal. As with any project, clearly stated goals are necessary. You need to define what a better search is. What type of content are your users trying to find? How have you tried to find it in the past? What's worked, what hasn't? What do you envision the user experience to be? Once you answer these questions, you're well on your way to defining a clear set of goals. 2. Define a common language. At some point along the way, it's critical that you establish and apply a common vernacular to your products, services, programs, and other important subjects that appear on your Web site, or within your content management system. This will most likely manifest itself as a taxonomy or thesaurus down the line. 3. Choose the right technology. There are two primary areas you should consider: search and metadata systems. Search systems should enable users to enter keywords or natural language queries, such as "Who is a local reseller for your products?" They should facilitate the ability to switch between searching and browsing to find content. Metadata systems fuel search engines by providing rich, contextual metadata around content. When assessing this technology, you should ensure that this system is driven from a controlled vocabulary, that it has a flexible import mechanism to support frequent vocabulary changes, and that it delivers accurate classification results. It's important, however, to also ensure your metadata system can support other critical functions, such as metadata storage and retrieval, concept and term extraction, and custom business logic. These features, which extend beyond pure classification, will tee you up for longer-term success for all your future initiatives. 4. Process, process, process. These three words do not have to be synonymous with frustration and pain. Process is your ally. It ensures that the goal you've identified in step one can be reached by taking your common language from step two, and applying it to your content through the software system from step three. Three key process points are:
- Provide the opportunity for manual review of content and metadata. Your software solution must support this by bringing a human into the process as needed based on exceptions or accuracy checks. It's best to do this early in the content lifecycle, when the original author of the content can be easily identified. You might also opt to provide an opportunity for review later in the process to meet the needs of content and metadata managers.
- Keep the impact to content contributors to a minimum. If it's hard to tag content with metadata, authors won't do it.
- Ensure that the process ultimately optimises the value of metadata. If you're capturing metadata to improve search or fuel your portal, make sure that the process carries the metadata all the way through. If the process stops short, you'll find it far more challenging to demonstrate ROI, because you'll find yourself stuck in "what could be" land.