SAN ANTONIO—Soft skills are as important as analytical expertise when it comes to solving problems in large enterprises. That’s the message distilled from a selection of presentations by corporate analytics experts at the recent INFORMS Conference on Business Analytics and Operations Research in San Antonio.
INFORMS considers soft skills so important that the organization requires applicants to demonstrate a list of them when applying for its Certified Analytics Professional designation, said William Klimack, a vice president of INFORMS. Klimack, a decision analysis consultant at Chevron, gave a presentation titled “Soft Skills in Your Organization: The INFORMS Perspective” citing the ability to frame a business problem, conduct interviews, and elicit information.
These skills surfaced in executives’ presentations, with lessons ranging from finding the right communication style to adjusting an analytics implementation to fit the organization’s culture.
Intel Adapts Messaging on Demand Forecasts
For instance, at Intel they discovered that giving out specific numbers from the results of a supply chain forecast is a mistake because the recipients over-emphasized them, recalled Tom Rucker, director of supply chain strategy and development there. But if analysts expressed their forecasts in terms of ranges, the recipients often used them for intelligent planning, Ricker said during a presentation titled “Concurrent Multi-Phased Development to Ensure a Vibrant Supply Chain.”
“We would give suppliers a number but the only sure thing with the number was that it would be wrong,” he recalled of the process of telling equipment suppliers what orders to expect.
Forecasting had become an issue at Intel because advances in technology force the company to build a new factory about every other year at a cost of several billion dollars, he explained. But sales forecasts could be off by a factor of 10 either way, and change orders are common up to a week before the ship date, he said.
Improving this situation required sophisticated analytics that were not possible until Intel cleaned up its data, which was done by renaming its products based on a common nomenclature, Rucker said. “That [process] took two years, and to go through it amounted to open heart surgery since the existing systems had to continue working,” Rucker recalled.
“We now have five years of data and can track how we are doing compared to forecasts, and can put in new applications in a few days thanks to the cleaned data,” Rucker said.
‘Explain, Explain, Explain’ at Mars
Downplaying numbers and giving the clients green, yellow, or red icons was found to work at Mars Inc., said Sorin Patilinet, marketing services manager at the food and candy vendor, who gave a presentation titled, “Analytics to Measure the Impact of TV Advertising on Sales.”
He said that Mars uses various sources of consumer survey data to determine the effect of TV commercials. The resulting reports show an icon associated with each ad.
“We don’t like to show numbers, as the icons are easier for marketing people to understand. They want to be able to say that they are as good as their competitors, or better, even if they doubled sales. Red means the ad is a waste of money, yellow means you need to try to develop something new as you can do better,” he said. He said it takes 10 weeks of data to see if an ad is any good, but that good ads can be used for years as they really don’t wear out.
As for other soft skills, “Take your time and learn what you are doing,” Patilinet said. “Find a person who can be your client but is your friend. Learn, and then expand. If you expand too early your legs will be broken by those who do not want you to succeed because they have their own agendas.”
And since there is so much turnover in marketing, every presentation must include a review of the methodology. “Explain, explain, and explain, and never stop,” Patilinet said.
Moen Takes Care with Pilot Projects
At plumbing fixtures vendor Moen the lesson was, “Crawl. Walk. Run. Make recommendations and don’t do them all at once,” explained Mark Golan, the firm’s global supply chain materials manager who gave a presentation titled “Multi-Echelon Inventory Optimization Helps Moen with Inventory Rebalancing.” He launched multi-echelon inventory optimization after two successful pilots.
A multi-echelon approach was needed because the finished goods and components were being gauged with different measures, he recalled. “The key lesson was the commonality of components,” and the discovery that some parts went into 50 different SKUs, he said.
“We saw what we needed and were able to leverage the IT department to combine individual reports, and that saved a lot of time,” he said. “The key metrics we monitor now are inventory and service.”
In the end, Moen was able to achieve a 10 percent reduction in safety stock parts and more stable service targets, Golan said.
Ford Adjusts Analytics Approach
At Ford Motor Co., the soft skill lesson involved organizational issues. “We tried to map the analytic process onto the conventional IT process, and it failed miserably,” said Gint Puskorius, senior technical leader for business analytics at Ford, who gave a presentation titled “Dealer-Specific Vehicle Inventory Replenishment Optimization.”
“You need smaller teams, an iterative development method, blurred roles and responsibilities, and analytical skills on the team,” Puskorius said.
He explained that three years ago Ford unveiled a new system that lets dealers order vehicles on the basis of recommendations derived from advanced analytics. Dealers can ignore or modify recommendations, but having the right inventory can let them get better prices since they need to offer fewer incentives.
“After 14 months of profitability we would like to think that analytics has something to do with that profitability,” he said.
The system, based on about a hundred trained neural networks, seeks to balance the inventory of all 3,000 Ford dealers in the U.S. across 14 product lines (including Lincoln). He defined balanced inventory as the situation in which every car has an equal chance of being sold in a given time. Total sales volumes may change from season to season but the sales mix does not, so the balance holds steady, he said.
“Prototyping is crucial — you don’t know what you don’t know, but you need to transition to the production version. The pilot program is critical to getting stakeholders involved,” Puskorius concluded.
Lamont Wood is a freelance writer based in San Antonio. Reach him via email at firstname.lastname@example.org.