Given the short history of mashups in the enterprise, IT departments likely don't entirely understand the technology's role in the business. Organizations can fall victim to mashup mistakes given the markedly different approach necessary to deploy and maintain them across the organization.

To start, the mashup is not the enterprise application. The latter, typically a large-scale long-term project led by the IT department, requires extensive designing, coding, deployment and ultimately user training. By contrast, a mashup is not built by IT. Rather, it's owned and created by a small group of business users who want to combine data in a customized fashion and consume it in small doses.

For that reason, mashups could number in the thousands within a single enterprise alone, said John Crupi, chief technology officer at JackBe, a mashup platform vendor based in the US. "It's not about creating something that serves a thousand users, it's about a mashup that may serve 10 users but you could have hundreds of thousands of mashups," he said.

The concept of building things for one or two people is largely foreign to IT, said Crupi. And although it may appear initially trivial in the eyes of IT to create a mashup by simply writing some code, that process will inevitably be multiplied by many small groups of users who each want their unique version of a mashup.

Mashups should therefore be driven and created by the business, and IT's role should be relegated to advisory support. "If you go to IT and say you want to do mashups on this data, they'll say they are moving to services-oriented architecture (SOA) in 2011. But the business needs it now," said Crupi.

By design, mashups are intended to be very light-weight and ad-hoc, without the "overhead and heavy-weight style" that has traditionally characterized the IT environment, said Ed Julson, senior director of product marketing with mashup technology provider US-based Kapow Technologies.

A mindshift is required of IT for mashups to be successful, said Kapow's founder and chief technology officer, Stefan Andreasen. Specifically, there needs to be recognition of the fact that knowledge workers are increasingly using desktop applications to optimize their work. "Mashups is the natural follower of the evolution of what is happening there," said Andreasen.

US-based IBM recently released the Mashup Center, a platform enabling business users to create mashups on their own. The company's global vice-president of portals and mashups, Larry Bowden, said there are a couple of advantages to letting the business user own the mashup. First, a group of business users with the skillset to assemble mashups brings power and innovation in numbers compared to the limited resources of the IT department.

Second, users in a particular department know their business better than any other, said Bowden, and mashups should be placed "in the hands of the business users to really lets them iterate and improve upon on a daily basis, as they learn in response to what's happening in the marketplace."

But placing the mashup in the hands of the user still means that IT has to ensure the right environment to facilitate the mashup, like appropriate authorization against certain data sources, said Bowden. Some organizations allow mashups to freely combine elements from the Web with corporate data, and "you'll have other companies that are in regulated environments, like banking, and the last thing you're going to do is let a piece of information about a person's account or hard numbers ever get mashed up in a market intelligence study."