The Key to Collaboration

Most manufacturers will recognize the forecasting tool that was employed up until the late 90s at First Alert, manufacturer of consumer fire and carbon monoxide detector equipment. It was an Excel spreadsheet, a giant, unwieldy one, probably 30,000 lines long with lots of subtotals, requiring manual updates every time there was a tweaking of the product line or a new customer. And most of the data came from 50 different sales rep organizations -- sometimes very optimistic ones -- that would fax in forecasts once a month for someone to key in.

First Alert executives knew they had to do something to improve their demand visibility and forecasting initiatives and trim their inventory position.

"We were running about the same amount of sales as we are today, $150 million, with $49 million in inventory," says Frank Finneran, forecasting manager for First Alert, a consumer products arm of BRK Electronics, which is now a part of Sunbeam Products.

First Alert reached its forecasting and visibility goals with roughly the same sales volume but now operates with just $17 to $19 million in inventory.

How They Did It

Getting to that level of forecasting took some time. First, Finneran and his colleagues attempted to build their own forecasting tool based on Microsoft Access.

"But we were still relying mostly on sales reps," says Finneran. Essentially, the improvement was not as vast as he had hoped.

First Alert then deployed a competitor's solution that knocked inventory levels down to $30 million just by having a dedicated forecast manager and a tool that could be easily plugged in. The tool produced a statistical forecast based on historical data, but didn't always produce the accurate buying patterns of today's fickle consumer.

First Alert has a level of seasonality in its products, such as a surge in smoke alarm sales during October, which happens to be Fire Safety Month. Business increases significantly in the fall as people begin closing up their houses and turning on the heat. "Seasonality and customer promotions creates variability in demand," says Finneran.

First Alert sells 30 to 40 base products in its three main product categories of smoke alarms, carbon monoxide detectors and fire extinguishers, but produces about 350 SKUs for these due to retailers' special packaging demands. Its top 10 customers, large retailers, deliver about 85 percent of its business, with the next 20 or 30 customers bringing that number up to 95 percent. Altogether, this equals a database of about 2,000 items by customer by SKU. Alas, First Alert wanted to forecast that business more accurately.

"What would take us to the next logical step was collaboration, a capability the current tool lacked," says Finneran.

Database Boost

That's when First Alert deployed John Galt Solutions' ForecastX EngineP2 and ForecastX CollaboratorP2, in concert with its existing Microsoft Access front-end database, which held product history as well as current budgets and forecasts.

First Alert also built "opinion lines" in the product suite containing data from various sources, including forecasts from internal sales staff and product managers, sales and order history and the current consensus forecast. Then it added POS sales and retailer inventory through collaborative forecasting, planning and replenishment (CPFR) programs with retail partners such as Wal-Mart.

"Now for 85 percent of our business, we're able to look at their order history for the last three years, the POS data, their inventory level, their POS forecast going forward, all on one screen for that SKU," says Finneran. "It's Internet-based, real-time and configurable, so we can assign different people different views."

Users can attach notes to explain data, such as a chain's planned store openings, or the fact that a pack size changed during the sales history, skewing results. New notes are pushed out to subscribers. Tools also allow "what-if" analyses.

Now, instead of increasing production purely on a salesperson's opinion, First Alert incorporates the sales projections into the tool as an opinion line available to a range of staff members. Order history is imported each month, filtered, then posted for all to view and respond. Once a month, middle managers in finance, engineering, marketing, production planning, sales, customer service meet to create a consensus agreement for the next segment in its rolling 12-month forecast.

Collaboration at Work

First Alert's accepted forecasts are uploaded into its JDA Software application and ultimately its manufacturing planning system, where it is factored in with capacity restraints, inventory levels and other influences to execute production. The improved forecasting has enabled a dramatic reduction in inventory levels, boosting turns to nearly eight times a year and improving cash flow available for marketing, research and development and other improvements.

Winning Formula

Despite the trimmed inventory, First Alert raised its customer service performance from 94 to 98 percent, according to Finneran. In addition, sales calls to retailers are now fueled with better data. The toolset definitely helps upstream as well, through purchasing, procurement and the people at the plants who buy according to plan.

Tools that increase accuracy and visibility of forecasting, according to Finneran, are essential to retain retail customers.

"Overall people with more efficient supply chains win," says Finneran.