Search is one of the most common activities that users will carry out on the organization's web site. In fact, many predict that search will surpass all other forms of navigation as the method of finding information. One day, a company's intranet may resemble Google's home page.

Indexing Documents
The selection of a search engine is critical for more reasons than one. At the top of the list is support and the strong foundation of the company selling the system. Things like index speed, scalability and query speed are quickly becoming commodities and all but the largest of organizations will be more than happy. But, it's the sustainability of the company you pick that is most important. Switching costs for search on the internet are non-existent. Don't like Google? Within seconds, a user can switch allegiance to Yahoo! Companies, on the other hand, face incredible costs when unhappy with a search engine solution, both in software costs and lost productivity.

EDM Logic recommends a search solution that handles multiple word phrases in the index. A system that understands that New and York together have a different meaning that the word new alone and that New York Yankees is different from New York is the latest innovation in this space. While this is not natural language exactly, a lot of the benefits from natural language engines come down to understanding not the query "sentence" structure, but multi-word phrases within the query text.

Querying the Engine
Querying the system is uneventful, but 95% of the user feedback will be based on lack of search relevance for the query entered. There is only so much tweaking an IT department can do to the index — and again, no search solution is perfect, even Google on the web. So EDM Logic recommends a "hedging" strategy. Develop some extra usability by adding refine search, best bets and other tools to the search solution.

Many will argue that the user is not willing to participate in finding information beyond the initial query. They are wrong. It's actually the fact that users can not "iterate" with the engine that causes the grief. All users are willing to actively participate in finding useful information, but become irate when the system gives them no options to advance the search.

Searching is something users do all the time, so a search box should be located on every page of the organization's web site(s). So information delivery should come down to one thing — the results page. But delivery is one area where user's are demanding more and more convenience.

Web Portal
Beyond the search results page, users are starting to ask for the ability to not only save searches, but monitor new documents in the index via the intranet site. For example, a company may have a language set for documents dealing with risk management. Anytime an employee adds a document anywhere in the organization that has is highly relevant (language set thresholds) to this subject, risk managers are automatically notified.

Mobile Device
While mobile device access to company information is a "next" generation phenomena, it can't be discounted considering the innovations in cell phone technology. While no one will want a full document before reading an extract or automatic summary of the file, users will demand a different result delivery from current browser-based searches and companies need to start preparing for this next generation.

Business Partner
Often, our clients have the need to share information not only with employees and clients, but with business partners. In fact, for most this is a stream of revenue. For example, JP Morgan sells analyst reports to companies like Bloomberg and needs an effective way to transport the content as it becomes available. Knowing what the business partner has at any one time and how it is being used (or misused) is very important to the content authors. There are special mechanisms that can be used to validate information usage by the business partner.

Today, there is no debate among CEOs that information management contributes directly to the bottom line and delivers increased business performance and responsiveness. Yet, there is no tried-and-true approach to integrating information management into an organization. Far from a commodity, it has alluded the best of companies who have tried everything from knowledge management teams to extending ERP systems. At EDM Logic, we have tremendous respect for the fact that there is no "silver bullet." The answer is in the approach — the framework for understanding our client's needs and demands for effective information management.


Most information management malfunctions come from the lack of coherent policies — not recipes or stringent outlines, but documents that create procedures to keep information relevant in changing business conditions. For example, what happens when the company makes an acquisition? What are the procedures for integrating the new company into the acquirer's information systems. While that is a specific example, what about the everyday functions of document retention and disposal? Governance covers everything from release management to handling user feedback. It also encompasses all legal requirements for information and often these policies find their way into a courtroom.

Stewardship is equally important. We call it "Tuesday-morning quarterbacking." After everyone sits down and comes up with a brilliant game plan on Monday, who is in charge of fixing it on Tuesday. Yes, even the best laid plans are full of holes, especially when user acceptance and engagement, factored into the plan, are impossible to predict until actual launch. It is up to the stewards of information management to listen, tune, fix and improve the enterprise information system. And these same resources steward the governance policies.

Many ask how content creation directly impacts information retrieval. Besides the obvious "garbage-in, garbage-out" explanation, content creation processes directly impact how information shows up in search results and information navigation. And we are not limiting our discussion to the manual creation of metadata by authors. Case in point is CAD drawings, notorious for having very little usable text to index, making retrieval a great challenge. There may be an algorithm that sits between the CAD drawing and the search engine that expands the description of the drawing using file location, author and department information.

Sources are important in one respect. The more, and the more diverse, the better. Companies believe they have an advantage over web search engines. Instead of billions of heterogeneous pages to index, companies have much smaller, homogenous indexes. But in reality, it's the differences between documents that make retrieval easiest. For example, an oil and gas rig company cannot rely on keywords drill, rig, oil or equipment as they most likely appear in most of the company's documents. The challenge is to look for subtle differences in documents so they can be organized, navigated and searched.

Absolutely the most important element to a successful information management system is the ability to measure that success and tune the system when it's looking more like a failure. Case in point — the design and implementation of information collections. Inevitably, some collections will be perfect, some will be overused and other collections will get no traffic at all. It is important to not only know this, but split apart highly trafficked collections and fold in less visited collections. And that's just to start. One of the most critical measures is whether users are leaving the search system from the search "results" page. This tells the information managers that users are giving up on searches before reaching actual documents. This measurement alone is worth gold to any successful deployment.

Once measured, how can the system be improved? 100% of the time, it's through user feedback. This is why EDM Logic encourages clients to put a feedback link on every page, including results pages. In fact, better than feedback, we label the link, "Did you find what you are looking for?" This encourage users to interact regularly with the stewards of the information management system. But, stewards must reply to every user suggestion and do so in a timely fashion to encourage and incentivize users to keep talking.

This gets a lot of "press" and cost associated with it, whether listed as usability, use cases, GUI design or user interface on the invoice. EDM Logic again believes there is too much emphasis on predicting the needs of users before deployment and under less than ideal conditions. We would much prefer setting aside the user interface budget as the "user reaction" budget. If this is too risky, still deploy, but with a smaller, trusted group. The feedback from real users, under real conditions will fix a system much quicker than any upfront plans. And, of course, the simpler, the better. We often encourage companies to emulate the Google user experience, which requires no training to use.

Access controls present an interesting challenge. While most search engines fully adopt an organization's existing rights management software, EDM Logic encourages clients to start with documents that can be seen by all employees. Yes, release one of the information management system is accessible and readable by every employee. If it's sensitive data, like HR, do not include in the first release. It's asking too much of IT to release a new system and not expect some hiccups along the way. But revealing sensitive data is one of those unforgiveable mistakes where heads roll. Plus, we find HR is not interested in sharing data until they see the system up and running. And often, we have been asked by clients not to add sensitive data to the main index, but make a copy of the search engine only for HR.