Jennifer Zaino
SemanticWeb.com Contributor

Zoom Information is finding more business potential in the semantic indexing capabilities behind its ZoomInfo business information search engine. The company this week unveils a new service that helps sales representatives mine small and mid-sized business prospects -- an increasingly coveted audience given the downturns in the fortunes of many bigger companies.

"Particularly in this economy our clients tell us they need coverage of smaller companies where there is lot of innovation and change," says Zoom Information VP of Enterprise Products Chip Terry.

They want to reduce the hours they spend manually searching for these prospects, which is a challenge given the volatile nature of the space. According to the U.S. Small Business Administration, there were 637,100 new businesses in 2007, and 560,300 business closures. But even with the ups and downs of players in the SMB space, they still represent a great deal of untapped potential and a viable alternative even as massive and long-established companies go down in flames.

Terry points as an example to a media company that's using the Zoom service to find little-known startups with VC funding and spin-offs to take conference sponsorships as big players in its industry pull back on spending. Another example: the biggest custom-made clothing business you've probably never heard of -- the Tom James Company, which sends its representatives directly to senior business executives in their offices. It's finding opportunity in selling to executives in smaller companies now, where there are fewer gatekeepers to deal with and less of a bomb shelter mentality these days, Terry says.

"In mid-market companies people are still excited, so they're a great place to find new business," Terry says.

For ZoomInfo, the task was to develop algorithms to discover and label companies already in its database so that it could find them in more ways. It considers, for example, the fact that the number of people in its database connected to a company, though only a partial list, gives some indication of the size of the business, as that number tends to scale up in bigger companies. It also considers the number of industries an organization is represented in and the number of products it sells.
Also on the list is an organization's web footprint, using its semantic searching capabilities to match references to a company on the Internet; millions of references indicate it is not likely to be a small player. On the other hand, it might be -- a media company, for example, may have a huge web presence, but ZoomInfo can extrapolate from its own data set that shows three people connected to the company in its database that the company likely has somewhere in the neighborhood of 30 employees. And, using its data as well as other sources integrated into its data set that give the average revenue per employee for a company of that type, it also is able to come up with an idea of that company's overall annual revenue.

"This is a model-driven algorithm where the model represent all these little levers and weights," says William Wechtenhiser, VP of engineering. ZoomInfo is defining small to mid-sized companies as having less than $20 million in annual revenue and under 20 employees. Data is kept fresh thanks to ZoomInfo's system that constantly crawls the web and refreshes information regularly. A small company that outgrows ZoomInfo's definition of small would be found as the service sees a steadily increasing number of people connected to it within its database, more web references overall, and then perhaps particular references to something like an IPO.

"That shows up and immediately and gets included in this model," says Wechtenhiser. Companies who can rely on their semantic prowess to drive new services may have a bright future. Essentially the same model and algorithm is behind another service currently in beta at Zoom Information and expected to come out shortly. This time it's aiming at helping to tell difference kinds of organizations -- profit, non-profit, government and education -- apart from one another, as sellers generally tend to focus on one or two of these verticals. It's harder than it looks, as there are lots of odd government entities and educational institutions out there on the web that might not be so easily distinguished from public companies.

"How to recognize those and sift them into different categories -- we'll use essentially the same model and algorithm and build up the semantic index to then slice and dice that way," says Wechtenhiser.