meteolytix generates precise sales forecasts for bakery branches

To read a German version of this case study, please click here.

Instrumented

• Data is collected from a range of sources: weather data from worldwide sensors and systems, daily sales figures from store POS systems and historical sales figures from ERP systems.

• Data flows into a customer-specific calculation model. The system determines daily sales forecasts for each branch and each product and dispatches them to the customer’s systems.

• From the differences between the actual and projected results the model learns to assess the different factors of influence and improve over time. The insights enable more exact material requirements planning, production and logistics optimisation as well as considerable reduction in returns. Better control of stock levels creates increased sales and greater customer retention, and less waste of valuable food.

meteolytix GmbH builds statistical models that provide daily sales forecasts for the retail and service sectors. With six employees, this innovative enterprise unites the know-how of statistical data analysis, high-quality weather forecasts and enterprise consultation service in a unique manner. Its aim is to save costs for its clients by making intelligent use of available knowledge to increase service levels and improve customer retention.

The company originated from a joint venture of WetterWelt GmbH, a special supplier for accurate, location-specific weather forecasts, and analytix GmbH, an institute for quantitative market research and statistical data analysis. It all started when a bakery inquired whether WetterWelt could provide exact sales forecasts for the products in its outlets based on the weather forecast, since experience had suggested to the bakery’s management that its sales varied widely as a function of the weather condition. Drizzle, for example, is typical cake weather. With a heat wave, however, grilled sandwiches tend to see higher sales. And with continuous rain the sales of one branch can decrease dramatically while they rise in another – depending on how the customer flows behave as a function of the weather.

In the beginning of 2009, Dr. Meeno Schrader, chief executive officer of WetterWelt, contacted analytix GmbH with this question of concrete ‘branch weather sales forecasts’. To answer this question WetterWelt and analytix combined their know-how and found an innovative solution which ultimately led to the foundation of meteolytix GmbH. At the ‘GründerChampions 2011’ competition, which is sponsored by the German Federal Ministry of Economics and Technology, meteolytix won the champion award for the Schleswig-Holstein region for its original and highly innovative business idea, as well as for its environmentally aware approach.

Calculating piece-precise sales forecasts for every outlet

Using IBM SPSS Statistics, meteolytix has developed a comprehensive model which calculates a daily sales forecast from a huge number of influencing factors for each branch and product. Both weather-dependent and weather-independent data (such as historical sales figures of the branches, vacation dates and holidays, the local competitive environment or actual marketing activities) are included in the calculations.

Dr. Björn Christensen, Chief Executive Officer of meteolytix, explains the reasons for the use of IBM SPSS Statistics in this solution: “We had already been using IBM SPSS software for a long time at analytix because of its ability to process millions of records with many attributes very quickly, and because of its rock-solid stability. It provides comprehensive algorithms for practically every statistical demand, and on account of its good market penetration it is already well known and very popular with many customers.

“It’s also a uniquely flexible solution, since you can purchase individual SPSS modules according to demand. Moreover, the support for R integration is extremely valuable for statisticians. We are entirely satisfied with this solution.”

The high precision of the sales forecasts has a number of positive effects for large bakeries. Considerable amounts of energy and raw materials can be saved on production and distribution, since the company can estimate exactly how many of each product will be sold. Experience has shown that the average number of returns for all items has decreased by approximately one third since the solution was introduced. This means there is less need to throw away high-quality food, and also reduces the likelihood of selling out of the items that customers want. Staff planning also becomes more exact, helping to meet demand and provide higher levels of customer service.

In addition, manual effort for processing daily product orders in the headquarters and in the branches has considerably decreased. Employees in each branch now only spend about half an hour per day on managing their orders.The automated and precise order proposals from meteolytix have reduced the order management process to a simple day-to-day comparison, plus the consideration of any special factors that might affect sales. For example, if roadworks are in progress in front of the branch, it might be the case that there will be fewer walk-in customers than usual.

“Some of our customers have already completely switched to the central order process because they have had such a positive experience from it,” reports Björn Christensen. “Every branch saves about half an hour per day on its order process. If they have 40 to 60 branches, this adds up to very considerable monthly savings.”

Developing individual forecast models for each customer

Before all the benefits of meteolytix’s daily predictions can be realised, the customer goes through a consultation and adaptation process that takes approximately six months. First of all, meteolytix needs to clarify whether the forecasting system is generally suitable for the customer’s specific conditions. Generally, this type of solution is most appropriate for a company with between 20 and 40 branches.

This first project phase relies on an intensive information exchange which serves to adapt the statistical forecast model to the customer’s individual conditions. For example, the regional and local circumstances of every single outlet must be considered. Further complexity is added due to the fact that many products need to be made in batches – single pie or cake slices, for example, cannot be produced. Production and default costs need to be determined to enable the derivation of perfect production and distribution plans.

“Theoretically as many factors as desired can be added to the model,” Björn Christensen explains. “The system itself learns to assess the importance of single influencing factors over time.”

In this consultation phase a customer-specific sales forecast model is gradually developed, and then submitted to a manageable number of branches for testing. In addition, the IBM SPSS software is integrated with the customer’s production and sales systems, because a steady backflow of detailed sales and returns data from every branch is a key requirement for the daily delivery of exact predictions.

The longer it is used, the more the system learns. The model constantly imports actual weather data from WetterWelt’s sources. Supra-regional influencing factors like holiday dates, vacation times or big events are added separately to the calculations. The constant input of new data creates a closed information loop between meteolytix and the customer. The aim is a system for production and distribution optimisation that learn independently, automatically adapting to the actual trends.

In the final phase, the meteolytix forecast system is connected to all of the customer’s branches. From this point on, the solution runs almost autonomously. The only additional factors that need to be announced to the system are marketing activities or specific situations that affect single outlets

The potential is nearly limitless

The solution that was originally developed in response to the inquiry of one large big bakery can now easily be applied to a whole series of other retail sectors.

Björn Christensen enthuses about the nearly boundless usability of this innovative solution: “Our forecast system is suited for a wide range of other applications. For example, cash register planning in the retail sector can considerably be improved with it, and we are currently in negotiations with a big retailer on this topic. A similar model can also help hairdressers plan staffing requirements much more exactly than ever before thanks to the small-region weather forecasts that are available at 15-minute intervals.

“Our latest talks with an energy provider are very promising. The precision of our predictions of sunshine periods or wind force in specific locations allows precise forecasts of wind or photovoltaic power over the whole day, and simplifies the planning of conventional energy production. Our SPSS model is very versatile and can cover practically any individual requirement.”

In the future, it is likely that meteolytix’s solutions will be used in other countries across Europe and worldwide.

“We can use our forecast model where weather measurements are available. And there are hardly any blank spots left in the world,” explains Björn Christensen. “Up until now, our customers have only come from Germany and Austria, but, in principle, with our SPSS model we could optimise the planning of thousands of enterprises worldwide.”

About IBM Business Analytics

IBM Business Analytics software delivers actionable insights decision-makers need to achieve better business performance. IBM offers a comprehensive, unified portfolio of business intelligence, predictive and advanced analytics, financial performance and strategy management, governance, risk and compliance and analytic applications.

With IBM software, companies can spot trends, patterns and anomalies, compare “what if” scenarios, predict potential threats and opportunities, identify and manage key business risks and plan, budget and forecast resources. With these deep analytic capabilities our customers around the world can better understand, anticipate and shape business outcomes.