From the manual collection and entry of data into Excel, the world of data management and accessibility has grown tremendously in the past 30 years. Gone are the days of clipboards and calculators and mainframes. As information technology took hold, data collection and analysis moved from manually driven processes into the digital age with increasing speed.
Today, enterprise resource planning (ERP) systems are essential tools for the metals industry, particularly for producers, processors and distributors. But some ERPs are production-process centric and may not encompass the “front office” functions such as planning, budgeting and forecasting that are essential to an accurate, real-time view of the business.
“Accessing data with Excel resulted in creating data warehouses around a specific function or machine. From that data, company-specific business models were built,” according to Andrew Zoryk, Accenture’s managing director and enterprise business services lead, chemicals and natural resources, based in Stuttgart, Germany. “Now, our complex businesses need enterprise-wide solutions that are largely being met with ERPs. But the acceleration of technology and the need for access to real-time data are driving the focus not only on the production side of the business but the business planning processes as well. Increasingly, companies must be integrated horizontally along the entire value chain that includes not just sales and operations but financial planning.”
Raj Chhabra, director, Deloitte Consulting LLP based in Detroit, has observed that “despite massive investments in ERPs, Customer Relationship Management and other enterprise systems, many companies do not have the information they need to make good business decisions. In most cases, the problem is not a lack of information but, instead, a lack of insight and accuracy of information.” Sources of inaccuracy in planning, budgeting and forecasting include lack of focus on the income statement, forecasts that are not based on consistent or relevant business drivers and an absence of an integrated data modeling framework.
The advent of ERPs did allow companies “to bring the data for all facilities on to one common platform” and make that data accessible throughout the company, according to Andrew Callaghan, senior manager of performance consulting with Crowe Horwath LLP, which has headquarters in Chicago. “Now the challenge is to get insights from the data. To expand into forecasting and planning, companies are starting to invest in business intelligence (BI) and machine learning software that identify patterns among vast sets of data that can lead to more accurate planning and forecasting,” he said.
Crowe Horwath views machine learning’s application for forecasting as a three-step process: 1. Gather historical transaction data from an ERP; 2. Generate baseline forecasts using predictive analysis techniques to establish data patterns; and 3. Review and adjust the forecast before submitting data to the business planning schedule. “Machine learning software provides the analysis through algorithms to create the predictive factors that will influence forecasting and therefore the sales and operations expectations for the entire business,” Callaghan said.
ERPs frequently are business-division or country specific, thus creating data silos that need to be pulled together, SAP AG’s Jennifer Scholze said, global lead, mill products and mining, based in Boston. Tools, such as SAP’s S/4HANA, offer a more integrated comprehensive solution to “pull together the data silos into one spot as well as providing standardization of data throughout the company. And by utilizing cloud computing, the data is presented real time and accessible across the entire business and throughout the world,” Accenture’s Zoryk said. “Increasingly, companies need to be managing their data with greater agility. They need to disaggregate data from its structured hierarchy so that you can easily see when you need to change your operations.”
Many leading software vendors have introduced “cloud-based technologies for deployment, advanced functionality, faster computing engines and purpose-specific forecasting models and accelerators,” Deloitte’s Chhabra said. “Access to large computing power, data storage and specific applications with leading process practices enable improved accuracy while reducing the overall cost of ownership. Access to cloud-based platforms for budgeting and forecasting will also broaden the availability to mid-market firms,” he said.
Using the cloud and other IT tools on an as-needed basis can reduce the costs of computing while expanding access to them within an organization. “The concept of ‘software as a service’ allows a company to pay for its computing time and power only when accessed. That model not only lowers capital expenditures for IT and software but provides up-to-date technology for all users,” SAP’s Schloze said.
The digital glue afforded through data unification platforms like SAP’s S/4HANA can be extended to a firm’s supply chain. Accenture’s Zoryk sees the “extended scope of vision beyond the silos to the broadest view of your business. For steel producers, it is possible to see the schedule of incoming ships carrying raw materials to determine the impact on future prices before docking and more accurately determine input costs that will affect final selling values. This type of data integration can help establish a more accurate and frequent view of the business cycle on a rolling basis and not just quarterly. This change in forecasting demonstrates a stronger commitment to sales and operations planning.”
SAP’s Scholze pointed to an increasing trend whereby automakers throughout the world are “sharing their market forecasts with key supplying groups like metals. That information based on real-time data allows metals companies to adjust their inventory levels in anticipation of forecasted demand, thus improving their overall forecasting strategies. We have also seen a metals company physically locating its employees at automakers’ facilities in the capacities of strategic planning to benefit both the company and its suppliers and not just for the purpose of product quality control,” she said.
Metals companies are increasingly relying on automakers’ forecasts “not just for the tier 1 suppliers but for those tier 2 and 3 suppliers as well,” Crowe Horwath’s Callaghan said. “Metals companies need to make the necessary IT investments to keep their automotive business. There is a delicate balance between turns and production, and we are seeing increased acceptance of forecasting as a critical element of achieving that balance.”
Forecasts are also positively influenced by historical patterns identified through machine learning. Service centers are able to increase revenue by having the right balance of material on hand based on the current and projected demand forecasts. “Traditionally metals service centers develop forecasts by looking at historical records and talking to their customers. Both of these activities are useful but machine learning can add a third dimension by combing through large data sets and identifying patterns and correlations. Machine learning models can help build a more accurate forecast based on historical data points and identify correlations between disparate data,” Callaghan explained.
Deloitte’s Chhabra agrees: “The concepts of mining data and data discover are not new but the use of historical data for predictive planning and forecasting purposes is in its early stages of adoption. Predictive planning capabilities can generate reliable statistical predictions based on data over periods of time and also capture product or time cycle impacts so that forecasting becomes more accurate and reliable.”
The next frontier in IT and software application builds on existing ERPs, which, if updated frequently, remain the critical foundation for further data integration. “ERPs are a longer-term investment of 15 to 20 years or more. By working with the right technology partner and updating on a frequent basis, a metals company can stay with the same ERP for a long time and avoid the significant cost of wholesale replacement,” Crowe Horwath’s Callaghan said.
Investments vary depending on the “level of complexity of the financial planning models and the landscape of integration requirements with source systems,” Deloitte’s Chhabra said. “Investments can range from $1 million to $5 million for the initial costs of hardware, software and training of employees. But investments in technology have to be layered on top of process improvements, and development of these systems for planning and forecasting is an ongoing process.
“Organizations need to realize there is no silver bullet to improving forecasting accuracy. Some of the guiding principles for improving accuracy include prioritizing for simplicity, planning for delivery in steps, maximizing shareholder buy-in and keeping a realistic focus. Improving forecasting accuracy requires changing a company’s culture along with its processes and information technology. This can only be achieved through improved transparency and a wider deployment of metrics,” Chhabra concluded.