An increasing number of businesses are exploring digital transformation initiatives as a way to strengthen their footing in a volatile supply chain climate.
Emerging technology can support this initiative with next-gen demand planning as a foundational pillar of the strategy. Advanced solutions, powered by artificial intelligence and machine learning algorithms, promise greater speed for data aggregation and processing while offering computing power to analyze a larger data set.
The ability to integrate real-time market conditions and their impacts in combination with historical data significantly improves forecast accuracy and creates a solution offering that will deliver the visibility and insight necessary to achieve resiliency.
But are these intelligent, machine learning-powered solutions the answer to all our planning problems?
How classic demand planning systems are structured and why they don’t work
There is a universal truth when it comes to classic demand planning; it’s a cumbersome process. It affects not only demand planners, but multiple parts of the business such as sales, marketing, finance, and supply chain. The global impact demand planning has within an organization, along with the unpredictable nature of demand, means that when the inevitable request to adjust and readjust plans and forecasts comes, the impacts are complicated and the processes to resolve them are lengthy and often siloed.
This is the case in a typical demand planning situation, where the onus to make demand forecasts generally falls on the shoulders of demand planners. However, to predict demand accurately, information from the sales pipeline, such as pricing and promotion to name a few, needs to be incorporated to get an accurate demand plan. Yet this information is not always available or shared across the organization. Without these critical inputs, it is harder for demand planners to collaborate with other departments like sales or supply planning to reach a consensus in decision-making, even within their own supply chain.
Traditionally, planners have to dig through dozens of datasheets to search for and extract historical data, and then run separate algorithms to generate demand forecasts. These forecast numbers are then filled into demand reports and handed over to supply planning colleagues. This makes demand planning systems more or less a manual entry record for final predictions, which are prone to errors due to incomplete, lagging data or manual errors when entering numbers.
Slow market change responses. Rigid processes. Outdated data
If demand planning follows a waterfall process and uses solely historical data, it faces a considerable risk of being offset by demand surges. Once a plan has been set into motion, companies have little time to react and respond adequately in case of a sudden demand change.
Take COVID-19 as an example: either with demand surges or dips, companies were caught off guard and hit hard as the first wave of infections traveled the globe. No historical data was available to support the predictions and impacts of a fall in fuel demand, an unprecedented escalation in demand of medical supplies, or supplies and equipment to enable remote work due to lockdowns.
We can now see that relying solely on historical data for forecast demand is not a best practice. Rather, incorporating lagging and leading market drivers can help companies monitor, respond and anticipate more effectively. These are the reasons.
Misalignment across planning functions
Misalignment between different parties involved in the demand planning process is a common pitfall. As mentioned above, demand planners may produce their plans on one platform, then put the final numbers into a separate system. In common cases, these forecasts are shared in the platform by planners without providing other departments with all the necessary documentation and explanation on how they reached those figures.
Business units like supply planning, which did not have all the processed data and information used to create the original plans, often dispute these figures if they believe they have better insights. This lack of transparency, visibility, and collaboration in communication between parties is what we usually call a “silo”.
Addressing existing silos and why they need to be removed
Siloed processes hamper corporate progress by causing delays in a supply chain planning process. When silos are removed and collaboration is done correctly, these processes become seamless and help companies achieve their potential. However, resistance to change culture, IT’s support of legacy systems, and hesitation to invest financial and capacity resources into a transformation project prevent many companies from starting on a path to remove these silos. While it is understandable that businesses have some trepidation embark on a journey that requires:
- Evaluating and selecting the right vendor software
- Obtaining strategic guidance from external consultants
- Employing data scientists to develop in-house algorithms for process refinement
- Adopting machine learning for continuous and autonomous implementation
These tasks should not prevent leaders from taking the leap.
Modern silo: Why are we reverting back even with modern systems?
The evolution of connected planning systems that heralded the demise of silos was great in theory, but the execution has proven to be a false start. While on a surface level the platforms provide holistic visibility, when we pull back the curtain we find a staggering amount of technological incompatibility that continues to hinder the acceleration of corporate performance in the supply chain planning market.
Usage of unconventional algorithms which lack compatibility
There has been a push by supply chain planning vendors to convert existing data sets into actionable insights, however, this approach only addresses the known variables that can impact a supply chain. Often, these software vendors have not invested in functionality in their road map to address:
- The exploding number of data sets
- IoT variables
- Other structured and unstructured data that impacts supply chain forecasting
To address this shortcoming, many companies are developing ad hoc formulas to try and capture the intent of the data. However, this approach results in new silos being created that are unable to address the original problem at hand and break the seamless collaboration touted by many of the existing options in the market.
This disconnect will cause new and often deeper rifts and mistrust between teams as the planning solution deployed to address the pre-existing silos can’t be trusted to give the right results. The solution is to embrace a flexible, scalable planning solution that can evolve with the inclusion of new data sets and provide verifiable numbers that the entire business can create a strategy around.
This further fuels other departments’ distrust of the results. They relapse into the same old confusion: “How did they come up with this output?” This lopsidedness continues to blur the planning loop for those who are involved and makes the realization of consensus decision-making harder than ever. This resembles the modern silos problem seen in legacy systems.
What happens if you let modern silos run free?
When planning processes and technology functionality aligns, accurate demand predictions follow. This accuracy allows you to deliver a better customer experience, higher SLAs, and a more profitable inventory forecast, all of which drive better top and bottom-line results.
When planning processes remain disconnected and ineffective, even after a costly and resource-intensive deployment of the wrong software, companies often fall into the same problems as legacy systems. The inability to deal with new data means poor inventory management resulting in a direct impact on operational costs (i.e. transportation, storage, and production) lackluster sustainability initiatives, increasing interdepartmental friction, and a loss of trust with both employees and customers. These impacts, while clear, are often untranslatable in regards to the exact financial impact but clearly do damage to the health and reputation of the misaligned company.
Further issues with in-house and vendor built planning systems
Besides silos, companies have further problems with modern planning systems, both with legacy in-house systems as well as deployed vendor systems.
- While good in theory, in-house developed software solutions often create more problems than they solve. While integration often requires less of a lift than an outside solution, in-house developers are often inward-looking and lack the ability to truly understand the scale of the problem and ideate a solution that will drive the best results. In short, they lack the expertise to build the best solution, and developing in-house will only delay the need to adopt a built-for-purpose solution that has the necessary scalability, functionality, and power to solve complex problems.
- In the case of adopting vendor software, where internal developer resource allocation is less troublesome, in-house data science teams will face the challenge of implementing, maintaining, and integrating vendor software mainly due to the lack of interoperability and lack of customization capability inherent in closed-source tools. Moreover, some developers have a mindset focused on driving additional revenue with costly, copious periodic updates and develop their product to require such repeatable updates.
Creating a silo-resistant planning future
When deciding between in-house developed solutions or adopting a best-in-class vendor platform, companies will need to prepare for many potential questions and hurdles. To better prepare for a journey towards a silo proof future, here are areas to consider:
- Focus on deploying a solution that supports consensus or majority decision-making. This means having criteria that each stakeholder within the network has end-to-end visibility of the supply chain, the variables influencing the plan, and the impact those variables have on the end result.
- Emphasize the importance of interdepartmental collaboration and require teams to connect to understand the objectives and targets of each team. This will foster a greater sense of ownership and buy into the planning process, reduce friction, and ultimately drive better outcomes.
- Prioritize solutions that are user-friendly and built on open source technologies. By focusing on a best-in-class, built for purpose platform with a foundational strength to encourage innovation, collaboration and customization, organizations will drive significant ROI both financially as well as with in-house developer career satisfaction.
Complexity in supply chain planning is on an exponential trajectory, and regardless of where a company is on the journey, there are new challenges that must be addressed. Whether a mature organization that is ready to ingest new data streams and the nuances of how to harness those insights to a less mature organization that is just starting to think about moving past planning in Excel, the requirement to remove silos is constant.
At o9 we understand the intricacies of the problem and can bring world-class solutions, expert guidance, and proven industry results that will help you transform your organization and achieve the art of the possible. The o9 Digital Brain is such a solution that has the innate ability to evolve according to an evolving customer need and grow to meet organizational objectives.
Want to read more articles like this?
Join the aim10x Innovators Network and get access to all aim10x content and connect with your peers!Join the network
About Stefan IJfs
Stefan IJfs is a Global Business Strategy Manager at o9 Solutions. He has over 5 years of experience in the Supply Chain Planning field and supported digital transformations across several industries. He believes good planning means enterprises achieve their goals while using less of the Earth’s precious resources.