In the past two years, a host of events created unprecedented challenges for the global supply chain. Between the ongoing COVID-19 pandemic, 2021 Suez Canal obstruction and global chip crisis, companies struggle to keep up with changes in inventory supply and product demand.

Although inherently unpredictable events are difficult to factor into sales & operations planning (S&OP), it highlights the importance of companies anticipating supply and demand changes to maximize productivity. The past few years have encouraged many companies to critically consider their supply chain resiliency and invest in artificial intelligence (AI)-driven S&OP solutions to exploit their data.

Recently, flexis AG released S&OP 4.0, cloud-based software designed to help companies with S&OP at both product and network levels. The new software uses AI and company data to identify patterns in sales to help companies maintain inventory and maximize profits. The goal is to make intelligent ordering and forecasting accessible to large companies without requiring a data science team.

"With the new version, we continued our investment in cloud technology, increased user comfort and added further AI components. In this way, we support our customers in making their sales and production planning as accurate as possible and adaptable in the event of market changes," said Robin Hornung, managing director of flexis Consult GmbH and Division Manager S&OP.

S&OP Aims to Develop Supply Chain Resiliency

As the saying goes, the only constant in life is change. However, inherent unpredictability in supply chains can make it difficult to maintain operations over long periods of time. To develop a resilient supply chain, companies need to anticipate changes and avoid expensive delays. To some extent, this anticipatory planning relies on real-time data and historical data to make informed decisions about ordering and manufacturing.

S&OP traditionally integrates information about demand and supply with a company’s larger financial plan. Usually, a company uses aggregate demand data and compares that to available product supply. This informs inventory purchases and manufacturing goals while impacting the eventual profit margins. Conventionally, companies use data spreadsheets to track sales and make predictions for S&OP. But this process does not take advantage of recent advancements in AI and machine learning. Like many processes, S&OP stands to gain from the big data revolution as companies learn to use their data to make more informed decisions.

“Sales planning currently uses many different models, but at the end of the day, it is still like reading a crystal ball. We try to improve this using an additional scientific approach,” Hornung explained.

Currently, most companies try to forecast country or regional sales while manufacturing products in a handful of facilities. With variant lead times and unpredictable supply chains, mistakes in S&OP can result in months-long delays. If a company overestimates sales, it is left with too much inventory and eventually outdated products. But underestimating inventory can lead to costly setbacks, tied-up capital and loss of sales. This is where resilient supply chain management can support sustainable manufacturing without affecting the bottom line.

AI-driven Intelligent Ordering and Forecasting with S&OP 4.0

S&OP 4.0 is a cloud-based software designed specifically for large companies in automotive, consumer goods or other manufacturing-based industries. The software is optimized for companies with complex products delivered through international networks.

“The two new pieces we’ve developed over the past year take advantage of the power of AI in S&OP applications,” Hornung said.

An overview of the flexis S&OP solution. (Image courtesy of flexis.)

The software aims to help companies answer simple questions, such as “Which product do we produce, and when, where and at which volume?” and “Where are demand and capacities out of balance?” Using AI, S&OP 4.0 uses historical data to help companies answer these questions and improve their supply chain resiliency.

According to flexis, the use of S&OP can lead to up to a 15 percent increase in sales, a 20 percent reduction in associated sales costs and a 20 percent increase in profitability.

Two New AI Tools Integrated in S&OP 4.0

With S&OP, the first step is to set up and configure the software to a company’s individual world. The software models existing networks down to the point of sale if necessary. Both sales and production structures are modeled within the network and integrated with the planning framework of the company. For example, the software can integrate production and logistics plans into the framework to help business decisions inform the data outputs.

An example of the S&OP 4.0 user interface showing forecasted sales overlaid with ideal inventory values. (Image courtesy of flexis.)

Once the software is configured, standardized functions are available to assist with S&OP, including demand capacity balancing, sales planning, inventory planning and more. The two features use AI and include a forecasting tool and an intelligent ordering tool. Both AI-driven functions use company data to facilitate S&OP decision-making.

The sales forecast, SmartProc AI (Prophet), uses historical data to identify patterns in sales. The process smooths the data to avoid predicting extreme spikes or other inherently unpredictable events. The goal is to identify overall trends and use those to inform long-term strategies.

With forecasted sales, a second AI step can then be used to make intelligent supply orders to keep up with demand. The new tool is called SP Supply Request, and it can balance the fulfillment of ideal inventories with forecasted sales predictions.

The forecasted sales and supply numbers can be manually edited depending on a company’s final decisions. The user-friendly interface allows anyone to toggle between different charts to see projected sales over time overlayed with ideal inventory values.

The main benefit of the two new AI functions is to deliver increased flexibility and rapid response to changes in supply or demand. At the same time, the software can help reduce risks and decrease production costs by anticipating predictable patterns in sales. The software is intended for long-term forecasting of anywhere from three months to 15 years as opposed to predicting day-to-day changes.

Companies can also plug and play with several APIs that integrate with third-party and vendor-specific software. Flexis uses a multi-tenet architecture that allows a company to call a given API and deploy it locally.

Both AI tools are based on an optimizer, such as CPLEX, ExpressMP or a flexis homegrown LP. The software can be hosted on any cloud environment, including Microsoft Azure, AWS, Google or company-specific internal clouds.

Innovation Drives Predictable Cycles in Sales

The consumer response to innovation is highly predictable. Usually, a new device or product is met with a period of hesitancy. This is eventually followed by a growth period in sales as consumers learn more about a new product. Using this information, Hornung highlighted the applicability of S&OP 4.0 to the automotive and telecommunications industries, where innovation drives cyclical sales following lifecycle patterns of products. Flexis currently partners with a telecommunications company for beta testing of the S&OP software to help it anticipate these trends and patterns in technology sales.

Hornung also presented an interesting challenge faced by the company. “When I present the software to new customers, many companies want to try it out right away. This usually means they want to predict future sales based on the last four years of data.” There is just one problem with this excitement: the last four years have included some of the most significant events of the previous few decades. The pandemic, chip crisis and more make it difficult to use 2020 and 2021 to forecast future sales. Instead, Hornung said his team relies on data from 2015-2017. The data is used to predict the 2018 and 2019 sales numbers and then compared to the collected data to confirm the model’s accuracy.

In addition to using their data, companies can also integrate competitor data as a reference if requested. For example, it is common to purchase forecasts for competitor vehicles in the automotive industry to integrate into S&OP decisions.

Next Steps for the Future of AI-Driven S&OP

In the immediate future, Hornung and his team are working on a process they have termed “automatic refinement.” The goal is to have the S&OP software compare the forecasted data to real-time sales to assess the model’s reliability. The model will then automatically reconfigure to continuously improve as real-time sales are collected.

Moving forward, Hornung and the flexis team are also looking at the big picture of sales in the context of a company at large. They are interested in the effect of advertising and marketing on sales outcomes and overlaying that information within the S&OP software. For example, how can a company determine if an ad campaign was successful and led to real-time increases in sales? By overlaying marketing information within the software, the flexis team will help companies better evaluate the efficacy of ads and better predict spikes in sales due to campaigns. The result will be two-fold: a more robust sales forecast and a real-time evaluation of marketing strategies.

Overall, it is an exciting time for S&OP as companies take advantage of AI to improve their supply chain resiliency and better anticipate changes in sales.