Background

Sinebrychoff, the oldest brewery in the Nordic region, was established in 1819 and is part of the Carlsberg Group, the fourth largest brewery group in the world.  Sinebrychoff has approximately half the Finnish market for brewery products. As well as its own brands, the company produces, sells and distributes Coca-Cola™ Company products in Finland. Sinebrychoff also offers an extensive selection of international beers and, via its partners, a wide selection of other alcoholic beverages.  The Sinebrychoff facility in Kerava is one of the most modern in the world and produces 400 million litres of alcoholic and non-alcoholic drinks each year.

Starting point

By 2000 Sinebrychoff had developed a forecasting process and implemented its own demand forecasting tool. The precision of the company’s forecasting was already very good! However as its product range expanded it became evident that it needed to increase the efficiency of its forecasting process and, in particular, to automate demand forecasting for products offered primarily as a service to customers. The growing product range also made it more difficult to evaluate and monitor marketing operations specific to individual retail chains when forecasts were made at product level.

“The brewery industry is very much campaign-led,” says Yrjo Nikkanen, Sinebrychoff’s Logistics Manager responsible for forecasting.

“Campaigns implemented together with customers and customer-chains often give sales a substantial boost. There can be several marketing operations in motion at once and they can have a cumulative effect.”

His logistics team works closely with Sinebrychoff’s sales people and with the company’s customers to predict the demand generated by marketing campaigns. Even though this cooperation yielded good campaign-specific results, Sinebrychoff also faced other challenges. Specifically it was very hard to work out how best to take a sales forecast for a promotion by a single chain and to use it adjust overall production given that a promotion doesn’t simply boost consumption amongst the originating chain’s customers but has a ripple effect across the market. “Finally, we came to the conclusion that we needed a good, sufficiently accurate quantitative base forecast that we could edit,” Nikkanen explains.

Towards a new forecast process

Sinebrychoff looked at a number of ways of improving its forecasting, from adopting the SAP solutions used by the Carlsberg Group to developing its own tools further.

However, in the end, RELEX was selected.

“The versatile forecast models of RELEX Forecaster and especially the optimisations that enable good automation were important to us. In addition, the straightforward graphic user interface and flexible possibilities to make forecast corrections convinced us,” explains Sinebrychoff’s Forecast Manager Antti Lappalainen.

Efficient and quick implementation

It took just three months from the point where it was decided to adopt the RELEX forecast system to implement it and bring it into use in the production process. Sinebrychoff’s System Manager Pekka Soininen says the implementation project was exceptionally smooth: “This one was quite painless as IT projects go. Progress was straightforward, and the project stayed on schedule very well.”

More precise forecast with the same workload

In tandem with the project to implement RELEX, Sinebrychoff moved to a product-/customer-chain-specific forecasting process. “Of course it’s a major consideration that we can create ten times as many forecasts with the same workload as before,” says the company’s Logistics Manager, Yrjo Nikkanen.

“The workload is also lightened because special events, such as the launching of new products, production interruptions or product substitutions, can be managed with a few button pushes,” adds the man in charge of forecasting at Sinebrychoff, Planning Manager Petteri Kokkonen.

Following development steps

Sinebrychoff is pleased with the way the new systems are working but their medium-term goals go much further. “At the moment, campaign actions are automatically removed so they do not ‘disturb’ future forecasts, but we still make the actual campaign forecast manually. The goal is – with the help of RELEX – to automate this aspect of our forecasting so that our ‘human input’ is focused on really important exceptions.”