Increase your demand planning accuracy with AI

For retailers, the basis of their entire business is buying and selling products. In the past, if you had one store and small stock, most of the work could be done manually. However, once the volume starts to increase and more locations or an online store are added, planning becomes increasingly more complex.



Demand planning and replenishment: complexity demands more certainty

A typical retail chain might have 20 stores, an online store and a distribution center. This is still quite manageable with a small product assortment. However, demand and replenishment for these stores and the distribution center become more complicated and essentially multiply when considering more product types and the number of products the retailer carries. Suddenly, things are no longer manually feasible. Additionally, retailers deal with certain financial restraints. For instance, they must pay their vendors before they can sell the products. The accuracy of forecasting and replenishment is therefore very important from a financial perspective.


For example, imagine a bakery...


What is the amount of stock that should be left in the bakery by the end of the day?


The correct answer is that one of every single product should be left. Why? In this way, the bakery knows they’ve sold the maximum. If an item sells out, that may seem like a good thing at first glance. However, that also means a retailer doesn’t know what potential sales they missed! If the bakery runs out of a product, like brown bread, everyone will begin buying white bread out of necessity. Yet, if the bakery starts optimizing the white bread stock, the business would be mistaken in its priorities.

Translating the bakery example to a business, retailers come to see that in today’s world businesses need more technology to forecast these developments as accurately as possible. There is a vast amount of variation: the weather influences purchasing, new product launches, peaks in the summer days, Easter falls on a different date every year and the list goes on. If a retailer relies on like-for-like forecasting and does not take these particularities into account, they run the risk of getting the wrong forecast for the wrong week of the year.


As the retail sector becomes more interconnected and the global economy more unpredictable, every business should harness its prediction power.



Speeding past spreadsheets

Besides forecasting, replenishment is extremely important. Not every establishment sells the same items across stores. In the same total sales volume, the number of products and their volume sold can be different. Retailers must be able to forecast this accurately, accounting for the fact that not every store gets replenished in the same time intervals. Some may get replenished every week, others once a month or three times a week. Therefore, retailers must consider different volumes. These individual needs and fluctuations quickly become overwhelming to manage in a spreadsheet. This is the starting point for automation.


Case in point: HEMA

IG&H has worked with the Dutch variety-store chain HEMA and focused on their beloved Easter eggs. As with any seasonal product, HEMA often had either too many or too few eggs. Selling off the complete stock after the holidays is near impossible. For one specific product, retailers can be dealing with numerous variables. This quickly turns into too much work to be manually. Here, planning must be extraordinarily precise.


Predictions in the retail sector: History is no longer repeating itself

Planning has become more complex for many reasons. A major contributor is the fact that retailers cannot rely heavily on the past anymore. The COVID-19 pandemic is a clear illustration that affected almost all businesses in one way or another. Normally, if a business grew over the years and had a stable growth of 10% p.a. you could forecast it. Due to COVID, the prediction accuracy fell significantly. While some businesses went through the roof, others collapsed completely.


Software can filter and normalize these factors; it helps calculate the answer to the question ‘What should the behavior be now?’ For instance, if a retailer's current model works on a 10% increase every year, then all forecasting will be off.


Regardless of whether a company recently experienced a negative or a positive turn, the market has completely changed over the last two years. The moral of the story is that retailers cannot rely solely on history anymore.


Artificial intelligence and automation to support supply chain professionals

When planning gets convoluted, humans can turn to technology. Artificial intelligence and machine learning require businesses to set up less rules at the beginning. These options recognize similarities, make connections and foresee patterns that weren't initially programmed.


While they do consider historical data, a vast number of other factors come into play as well. This technology isn't rule-based. Instead, the machine essentially says ‘Alright, if A looks like B, then C might look like B as well, or not at all, or it behaves the same as C, etc.’ These options are automating but also making things smarter, providing supply chain professionals with a bird’s eye view on their stocks and all the underlying relationships.


Considering the financial importance of forecasting, what retailers lose in missed sales or overstock adds up fast. Overstock, understock and, perhaps worst of all, out-of-stock situations can be avoided with the help of advanced solutions.


Embracing machine-based learning and AI in demand planning is a convenient way for retailers to harness the power of their data, while still staying in control.


Supply chain professionals are ultimately the ones in charge and can choose to accept suggestions, introduce variables they know from their own experience or other particularities of their business. Going back to HEMA as an example, we saw that this data-driven way of working led to:


Whether it's Easter eggs, an e-commerce platform or a massive retail chain, planning is more challenging than ever before. Retailers can navigate increasingly complex demand planning with the support of AI and machine learning algorithms. IG&H has over three decades of retail experience and advanced data and analytics capabilities. We have combined this expertise with our tech capabilities and created a demand planning tool.


Are you looking to optimize your demand prediction?




Recent Posts

See All