How to Increase the Reliability of Your IT Infrastructure Using Predictive Analytics REGISTER >
At Austin, Texas-based National Instruments Corp., a $514 million manufacturer of sensitive digital instruments for engineers and scientists, enterprise search is critical on several fronts, says Jeff Watts, the company's search and community manager.
National Instruments' products are highly technical, so call-center support plays a big role in maintaining customer satisfaction. In fact, all of the company's frontline tech support reps are degreed engineers. To help keep callsand costsdown, National Instruments created a massive database with hundreds of thousands of documents, all of which can be searched and accessed online by any of the company's 25,000 customers (in 90 countries) in a matter of seconds.
"We make every effort to push customers to use the Web first," says Watts. In fact, he adds, customers who want to speak to an actual call-center rep are encouraged to first qualify their questions online and get a support number, thus giving agents time to find the information (using the same search tool, by the way, as the customers) they need in order to successfully field the call. That's significant, considering the company averaged 350 support requests per day in 2004.
But the search engine, from Oslo-based Fast Search & Transfer (FAST), wouldn't be worth a dime without the intense content management effort Watts oversees.
"Creating the content is a massive companywide effort," Watts says. As many as 1,000 employees at NI are involved in writing content, from application tutorials to troubleshooting articles. Each document is carefully classified and stored so that users will be able to find it easily.
Tagging and classifying data is an expense that many companies overlook when it comes to search, says Joseph Busch, principal of Taxonomy Strategies, a San Francisco-based consultancy.
While companies such as Google make searching for information look easy, "it would be a real disservice to imagine that this is going to be a plug-and-play solution," he says.
Check out eWEEK.com's for the latest news, views and analysis on enterprise search technology.
"It needs to be painstakingly configured in order to really make it work." For example, software programs don't always associate files with the proper keywords, which forces companies to sift through documents manually and add keywords file by file.
Furthermore, to create a solid taxonomy for datawhich is key to ensuring that documents can be searched properlythat data needs to be tagged with "metadata," which assigns content and context information to any given file, including subject matter, type of file, last modified date and other information.
Ananthan Thandri, vice president of IT at Cadence Design Systems Inc., a leading maker of CAD-CAM software, agrees that search alone won't solve the problem of getting relevant data.
Cadence, which has been using search products from Verity Inc. since 1998, puts all employees involved in creating content to work on properly assigning keywords and affixing meta-tags to each piece of datamore than 1.1 million documents.
"We spend a lot of time on tagging and making sure we have our taxonomies set properly," he says.
While you're sorting out the content management issues, be sure to consider sensitive and personal data as their own separate categories, protected from general searches.
National Instruments found that employees often save personal information to internal databases without knowing it. "Before we had search, people saved a lot of things on the public intranet that shouldn't have been there, but it was next to impossible to find. Now our new search engine has crawled all of that.
So it's not difficult for one employee to accidentally find personal information about someone else," Watts says. Getting people to be more mindful of where they save their data has been a challenge, he says, and Watts and his team are drafting guidelines.
Because of that, the company has been wary of adding more searchable databases until it can be sure that sensitive documents are saved to the proper servers. Watts estimates that at least half the company's content is still not on the system as a result.