Enterprise search is one of the most powerful but underused technologies available to IT. Many companies aren't buying search at all, and many that do have it aren't implementing it on a broad scale.

It isn't the fault of the technology or the companies that sell it. The products for the most part work. What's getting in the way are internal politics, understaffing, and an unwillingness on IT's part to tackle the bigger requirements that could truly change how employees find information across a company. Too often, enterprise search technology ends up pigeonholed in a single department or used on a single data set--a big but underutilized investment.

That's exactly what happened to one U.S. high-tech manufacturer that two years ago started using SharePoint as the central hub for its product documentation, discussions, and collaboration. The approach caught on with pilot users, but they had problems finding information that was split between the SharePoint portal and the network.

The company deployed Google Search Appliance, which could search SharePoint and other systems. It was a nice fit, and the pilot was expanded to include main databases, file shares, and e-mail.

Flash forward to today: SharePoint never caught on at the company beyond the initial group, and neither did enterprise search, which now helps only 5% of possible users search about 10% of the systems they use. This happened despite the fact that the company has the technology and licenses it needs to expand search to the entire company.

Is this an oddball case of a project gone bad? Hardly.

Here's why enterprise search usually fails, and a three-step approach to making search succeed.