Stopping Brain Drain; Almost a third of bruce power's workers are set to retire in the next three years. The goal: suck their knowledge into a database before they leave.

Twelve hundred of the 3,700 employees who work for Bruce Power, a utility company in Tiverton, Ontario, are eligible to retire in the next three years. CEO Duncan Hawthorne sees that as a threat - and an opportunity.

The opportunity: Infuse Canada's first private nuclear power generator with new blood, thinking and technical skills.

The threat: Retirees leave with more than two decades of knowledge about Bruce Power's reactors, how those units are maintained, and all the little items - such as the quirks of an aging steam generator and ways to weld metals that have been altered by extreme heat - that can't be found in textbooks.

Simply put, the generator of 20% of Ontario's power could come apart almost literally at its seams.

Hawthorne can, of course, afford to lose the retirees' salaries. What he can't afford to lose is what's in their heads.

"These workers were here in the mid-1970s when the site was commissioned and the boilers were installed," Hawthorne says. "They have the full history. We can replace the certifications, but not the tricks of the trade and the skills of a craftsman. We don't want to lose that corporate knowledge."

His answer? Making "sure we've sucked dry the experience of workers before they leave." His tool? A knowledge database created using Kana Software's IQ application. When finished, the company will have access to information on everything from how to weld steam-pipe fittings and their supports, to how to perform an exit interview properly, to workarounds for repairing an office printer.

A knowledge management project might seem like a no-brainer, but it's hard to quantify, according to Jim Murphy, an analyst at AMR Research. First, a system has to be developed so it can apply to multiple parts of the business such as human resources and plant maintenance. Then you need employees to use the system and deliver their knowledge by inputting tips into a Web form. Even if that goes well, measuring returns on sharing experience and saving time isn't easy.

Nevertheless, Murphy says knowledge management software is being used in businesses that have a large number of pending retirees, and where operating efficiencies or a competitive advantage such as a novel R&D technique can be promised. A pharmaceutical company, for instance, can't just let the scientist who has developed a patented drug defect to a rival.

The rub: "It's an inexact way of approaching the problem, but it's better than nothing," Murphy says.

Hawthorne says most of Bruce Power's returns are anecdotal - a crew, say, that used information from the knowledge base to save an hour on a reactor repair. But there aren't any hard numbers linking knowledge management to the end result. "We track work crew tasks, compare crews and look at errors and time to completion, but it's hard to say how much can be attributed directly to knowledge management," he says.

However, Hawthorne sees knowledge management as a necessity. And supervisors can create and maintain metrics that will give some idea of the benefits produced.

Murphy says most companies using knowledge management software try to calculate returns based on phone calls saved, time saved - say, 15 minutes a day - or lower training costs.

But that assumes you could put a value on saved time, and employees used the time for work. "If I had an extra 4 minutes a day, what would I really do with it?" Murphy asks.

Hawthorne went ahead with the knowledge management project because it serves as an insurance policy. "We can never be in a position of not knowing how to do something," he says. "This allows older workers to pass along skills and help train future workers."

Christophe Michel, manager of technology solutions at Bruce Power, says the company began its knowledge management rollout slowly in early 2002 by focusing on human resources. The problem: Bruce Power had too many phone calls about policies governing vacation time and benefits. The fix was developing a portal that could handle the most basic questions and free up managers for more complicated queries.

The project, completed in November 2002, served as a template for a dozen areas, ranging from welding to strategic issues such as company positions on the environment and technology support.

According to Michel, each knowledge base has the same look and feel, and includes Web forms to allow workers to input "knowledge deliverables" about a specific activity. For instance, a report on a turbine repair would include the job, time to complete it, and details on any hurdles such as metal that unexpectedly fused and how they were overcome. The data is shared across the company and, in some cases, the nuclear power industry, which shares information about plant operations since the Three Mile Island meltdown in 1979.

The challenge is convincing employees that sharing knowledge about something like maintaining the brackets that hold steam pipes in a generator is important. "The success of knowledge management might be 20% to 30% technology, and the rest is process and culture," Michel says. "The most significant hurdle is changing the attitudes toward daily work."

Hawthorne says the goal is to capture data on jobs that don't occur daily, say, restarting a steam generator after an outage. He says that knowledge is the most likely information to be lost as retirees walk off into the sunset.

If all goes well, Hawthorne says Bruce Power will retain its corporate culture and provide road maps to do both routine and once-in-a-decade tasks. Hawthorne likens knowledge management to the difference between getting directions from Mapquest and a local truck driver.

"You can print a map that gives you a route; that's easy," Hawthorne says. "But that route won't tell you where the construction is unless you get directions from someone who has traveled that route. We're trying to retain that knowledge."

Bruce Power collects employee experience with knowledge management software. Quantifying returns is the hard part.

YELLOW LIGHT Evaluate. The value of encoding knowledge is hard to quantify.

GREEN LIGHT Go

YELLOW LIGHT Caution

RED LIGHT Stop

CEO: Duncan Hawthorne Decision: Whether to use knowledge management software to ease a retirement wave.

How Do You Know If You're Getting a Return?

Decision

*license supports 100 concurrent users at $1,500 each in 2003, 2004. Company didn't disclose costs. **first-year development cost $500,000, with a full-time and part-time developer in future years. ***assumes 1,850 workers use software after two years. one employee accounted for $74,971 in profit before taxes in 2004. Assumes an 8-hour workday; 7.5 minutes saved is used on work. Sources: AMR Research; Baseline; kana software

EXPENDITURES 2002 2003 2004

Software costs* $1,000,000 $150,000 $150,000

Development costs** $500,000 $120,000 $120,000

TOTAL $1,500,000 $270,000 $270,000

SAVINGS

15 minutes a day*** $0 $0 $2,160,800

$0 $0 $0

TOTAL $0 $0 $2,160,800

NET ($1,500,000) ($270,000) $1,890,800

NET in first-year dollars (2002) ($1,500,000) ($245,455) $1,562,645

Discounted to present at 10% a year. Calculations in U.S. dollars

Dashboard

NET PRESENT VALUE: ($182,809.92) Three years of net values

FreshDirect chairman Rick Braddock, 63, isn't going to dazzle you with technology know-how, but he has a knack for using systems to make commerce a little easier.

At Citibank, where he served as president from 1990 to 1992, Braddock helped double use of the bank's automated teller machines. As chief executive of Priceline, he brought reverse auctions to the travel business. Now he leads New York-based online grocer FreshDirect as it marches toward an initial public offering.

Experience has taught Braddock not to look at return on investment as the magic metric, because it doesn't foretell success. Instead, Braddock focuses on sales metrics such as average order size. His take: Monitor how you're doing with customers, and the rest will take care of itself.

Baseline news editor Larry Dignan recently spoke with Braddock about FreshDirect's prospects and metrics.

Why does the FreshDirect model work where Webvan [an online grocer that went bankrupt] didn't?

Look at the business model of FreshDirect. We have a 300,000-square-foot facility and buy product directly from the source, turn it around and ship it directly to customers. We get [groceries to customers] at least seven days faster than a traditional grocery store supply chain. We have no stores, so we can reduce costs and have higher margins [and can pass that] along to customers. Since the product is directly shipped, it's higher quality and fresher. That's the reason we exist. Webvan didn't have that model with direct purchase and turnaround. They didn't develop a consumer proposition and had a low average order size. They used their capital to build out their warehouse and manufacturing network around the country before they validated the proposition for the customer.

What are your best performance metrics? What's the big number?

I'm very numeric the way I manage, but I believe you can have big problems when you pick out one big number. I think naming one metric would be a mistake as a leader. Because if I did that, it would convey that I only care about one thing. I'm 63 years old and have run businesses for 30 years, and I've learned that what I look at as a businessman influences the development of the people who work with me. If I take a simplistic view of the business, which one number would imply, I do a bad training job, a bad development job and a bad management job.

So what measurements are important to you?

For FreshDirect, trial and repeat customers matter. You have to get a new customer to try FreshDirect, but that doesn't always mean he'll become a loyal customer. So frequency matters. There's also a set of customer service metrics you have to track. Early in a customer relationship, you have to be strong in your service delivery. In all Internet businesses - and perhaps more in this one - you need to replace a [person's] face with reliable, high-quality service. You also have average order size - which is what harpooned Webvan. You have to have a nuanced view of the customer from a lot of different dimensions.

Why not focus on returns on investment and capital?

To me, those types of return metrics are lagging indicators of business performance. If you don't get your business right from a customer perspective, you're not going to have a fun time looking at returns on investment. In my view, the first thing to do is work the customer dimensions and then look to the ROI and similar measures. Usually when a company misses its financial metrics, it has been in decline for a while. And the reaction to fix it is usually a financial reaction when it should be to go back and look at the business proposition to the consumer to see what's wrong.

So, the behind-the-scenes technology - bar codes, manufacturing lines and distribution centers - doesn't interest you?

I don't consider myself good at that kind of thing. I'm more engaged strategically with technology. The kind of issue that interests me with FreshDirect, Priceline and the Internet in general is the conversation with the customer where you have a real-time relationship. The problem with most Internet applications is that there has to be a stake in the ground between a dumb application - the historical way people bought books, music and airline tickets - and a conversational one-on-one approach where you're pivoting off the knowledge you gain from e-commerce transactions. Just like a credit card company, FreshDirect has a ton of information on its customers, but I'd call it data, not information, at this point.