MY EXPERIENCE WITH DEMAND PLANNING
August 18, 2026
5 min read
English
80 views
Professional space
#demandplanning
Less info
More info
Up until 2019, a significant portion of my professional career was spent working in Demand Planning departments. Across two distinct professional opportunities, the primary objective of my role was precisely the calculation of demand forecasts. After reviewing notes and reflections from that period, I realised how deeply curious I was about the state of the art in this field of knowledge. Companies frequently establish strict operational challenges and boundaries, which means that complementing professional duties with independent research and experimentation is the ideal approach for professional growth in a given domain.
One of the most engaging professional positions I held was as a demand planner in a Peruvian company with a significant global presence in the retail sector. Prior to my arrival, the enterprise typically downloaded sales reports, analysed them, and produced demand forecasts using qualitative models and commercial factors alone, which served to replenish merchandise across stores and outlets while calculating purchasing requirements. As soon as I joined, I began reviewing all available information, including sales reports, stock levels, objective tracking, market trends, and past demand projections. Within two weeks, I built an internal database and produced my initial demand forecast utilising three projection models: linear regression, exponential smoothing, and exponential smoothing with trend adjustment. By the end of the first month, I was actively controlling inventory and incorporating advanced commercial parameters into demand projections.
Recognising my rapid mastery of commercial information, management proposed working on more advanced models aimed at improving supply chain efficiency. I began by establishing an inventory model defining the target service level, thereby obtaining a safety stock to protect demand calculations against procurement lead times. To achieve this, I referenced the newsvendor model, which calculates the purchase quantity that maximises sales profits based on pricing and item costs, ensuring that products with higher demand variability faced a lower probability of stockouts. Furthermore, I formulated effective demand calculations by comparing sales projections against available merchandise, accounting for both physical warehouse stock and items in transit from origin. This required coordinating closely with procurement and product managers to execute actions ensuring the effective consumption of goods at retail points.
Throughout these experiences in demand planning, I learned that flawed demand forecasting can constrain sales potential or inflate logistics costs due to over-dimensioning, alongside generating various negative impacts on warehousing and distribution. Regardless of how sophisticated or suitable the algorithms and methods employed may be, a forecast error will always occur and must be continually monitored and minimised to guarantee an efficient and improving operation.
Up to 2019, the demand forecasts produced by the departments I worked with ranged from simple intuitive quantity estimations without calculations, or direct comparisons with previous forecasts, to the use of historical data and processing models featuring automatic parameter adjustment and advanced algorithms. In the later stages of these experiences, I generated demand projections segmented by business type, encompassing wholesale, department stores, retail shops, outlets, and online channels. I implemented customised product disaggregation tailored to the commercial requirements of each category and developed algorithms enabling forecast errors to be automatically analysed and minimised by the model, alongside an alarm system to project stockouts and anticipate inventory replenishment. I believe there is no inherently incorrect forecasting method. What is incorrect is failing to monitor the error and neglecting to refine forecast calculations. For example, with collectible products involving recurring deliveries over a specific timeframe, delivery data feeds back into the forecast, whereas for apparel products, seasonality and historical sales drive the forecasting process. In any scenario, emergency reprocessing in light of new evidence that challenges existing forecasts is common. This is precisely what makes demand planning fascinating, as the moment a forecast is finalised is precisely when it materialises in the real world and begins impacting actual resources.
Across both my demand planning roles, a recurring requirement was the automation of repetitive calculations. To achieve this, relevant data was extracted manually from the system, as the information technology department had not formally automated data extraction. Subsequently, data transformation was also performed manually, often without adhering to standard processing workflows or a common template to ensure consistent criteria. I consolidated my expertise in developing and managing Microsoft Access databases, with a strong emphasis on writing VBA algorithms and automated reporting. Consequently, a planner's ongoing project was invariably the automation of demand planning calculations to establish an automated replenishment proposal. As these tools improved and demonstrated the positive impact of the new forecasts, I noticed how interest multiplied. Suddenly, a larger number of professionals requested adjustments, calculations, and new approaches because minimising the computational barrier fostered experimentation and continuous improvement in a chain-reaction effect. In these roles, I also supported senior management in strategic decision-making regarding the opening and closing of stores, whilst analysing demand elasticity subject to commercial parameters such as pricing and stock levels. Although the involvement of other business areas strengthened the implementation and adoption of these automated solutions, the information technology department was rarely involved, as doing so risked slowing down the impact and reverting the automation back to the starting point.
Following these experiences, my conclusion is that while departments like Demand Planning are highly automatable and closely tied to a company's commercial strategy, once a measurable impact has been achieved, those solutions ought to be formalised through data governance practices led by the information technology department. Although this process may take time and prove complex, continuous experimentation and the operational reduction of manual effort remain crucial. Furthermore, automation tools are highly recommended as supportive aids rather than replacements; human judgment, intuition, and common sense remain indispensable for permanently evaluating the adoption of new techniques, assessing their impacts, and driving the constant minimisation of errors.