In-Depth

Bay Area Chapter Explores Outsourcing, Visualization

TDWI’s second San Francisco-area chapter meeting attendees heard expert advice on offshoring and dashboards.

A quick show of hands in the audience at the recent San Francisco-area TDWI chapter meeting indicated that most companies still send data work off shore to save money. That disappointed one of the panelists.

"Cost is the wrong reason," said Anthony "Tony" Giordano, a partner at IBM Global Business Services. That advantage is evening out as offshoring intensifies. Talent is spreading out, and the demand for talent is forcing costs higher.

The area’s second chapter meeting, held late last month, touched both ends of the BI data trail—starting with visualization with dashboard expert Stephen Few and ending with a panel on offshoring, one of the most-requested subjects.

A much better reason to send work offshore, Giordano said, is to focus on what each company actually knows how to do. Most don’t know data administration, so why not send it out?

Offshoring (involving a change of place) and outsourcing (involving a change of company) can also be done to achieve the right scale by finding enough talent. One distributor to large U.S. restaurant chains, recalled Shobhit Jaiswal of Wipro Technologies, couldn’t find the required talent pool domestically. They were able to find it offshore.

The most-repeated theme was that offshore operations consist of people who need the same attention and are just as ambitious as your onshore crew. Invest in your offshore workers, care for them, and get them involved.

With rising demand for talent in India and elsewhere, talent is turning over—and fast. Talent is spreading out, too, and competition for staff is strong. "The market is really hot," said Peter Hikido of Franklin-Templeton Investments. "We have people cold-calling our folks every day."

Turnover isn’t just your vendor’s problem, it’s your problem, cautioned Dave Kohn, a partner at Accenture, which helps business intelligence clients with outsourcing.

To slow it down, engage your offshore workers. "People are no different from onshore here. They like to know that they’re valued and that they’re part of the organization." He reports much higher retention when the client is involved.

Giordano added, "It always helps to be personal." He said he tries to send congratulations for things like births and thank you notes to someone who worked over a weekend. "People want to be noticed, to feel like part of the team."

Those little things change your attrition rate from 30 percent to 15 percent just as a result of those personal acknowledgements. What also helps is giving them a career path and to help them grow.

Hikido has replaced the term "offshoring" with "global model" to denote a system in which offshored people have a relationship with onshore companies and want to stay.

When they do leave, make sure your operations are "de-linked" from individuals, advises Jaiswal. Make sure there’s someone ready to take over in case a key worker leaves.

Also, start simply. Establish well-defined tasks with clear handoffs and clear expectations of quality. Make your offshoring safe and easy while you build confidence and competence, Giordano advised.

Start with processes that can be clearly defined. ETL is a good one, Giordano said. One that’s not so suitable is data modeling. It’s "really tough off shore," he said, because of the heavy interaction with the business side.

In particular, he recommends enterprises have a unique set of requirements for each type of activity. For example, handoffs for ETL work are different from database administration, which is different from analysis. Generic handoffs don’t work.

Giordano warned, "Have a really big pipe." People underestimate the bandwidth required—information security, too. You’re going to send some of your most valuable information half way around the globe. The security setup can be very time consuming.

Others cautioned companies to make sure software licenses cover offshore seats.

Enterprises must set up the communication lines and procedures. For example, they need to be able to answer the question, "Who’s going to be awakened in the middle of the night?"

A Shower of Facts

All that data bouncing out of India and other places produces "a meteoric shower of facts." That’s a line from a poem by Edna St. Vincent Millay that opened dashboard expert Stephen Few’s talk. It goes on to say that those facts have "no loom to weave it into fabric."

Few explained, "You can have the most powerful data warehouse in the world, but if you can’t make sense of it, that whole investment has failed."

If you’re looking for patterns, trends, and exceptions, visualization is your tool. Yet visualization is badly underused in BI, he said.

Few showed several ways software can help interpret data: It can filter data dynamically, showing the user results on the fly. A series of charts can, in one glance, show patterns and outliers. Users can zoom in on a subset of data. The user always benefits from flexible navigation.

Few goes into detail on that point, of course, in his two books: Information Dashboard Design: The Effective Visual Communication of Data (2006, O’Reilly) and Show Me the Numbers: Designing Tables and Graphs to Enlighten (2004, Analytics).

As usual at TDWI events, the exhibitors held drawings for prizes. The most fitting that afternoon was an iPod: made offshore and sporting an interface worthy of a meteoric shower of songs.

Must Read Articles