How to Simulate Supply Chain Inventory Scenarios to Better Manage Cashflows for Retailers

Demand forecasting is a powerful and lucrative tool e-commerce businesses can employ to future-proof their inventory management. Below we discuss why demand forecasting is necessary and how to get started.

Inventory management has become a central challenge and often a recurring pain-point for modern business leaders. In a Forbes interview, Tim Cook said that he loathes inventory and finds it "fundamentally evil," revealing how extensive and complicated of a problem it can be even for the most high-tech and well-resourced companies.

For small and medium enterprises (SMEs), efficient and effective management of inventories is not just a matter of meticulously maximizing profit, but a question of survival. The boom of e-commerce services provided many opportunities to SMEs, but at the same time, intensified competition and raised consumer expectations. These days, consumers want everything "right now," and no one wants to wait.

E-commerce SMEs have multiple tools available to properly manage their inventory. To fully ensure customers can get their products whenever and wherever they need them, e-commerce SMEs can decide to purchase and keep a huge inventory in a warehouse–a logical response to the "always available" challenge. However, maintaining a fully stocked or overstocked inventory creates a potentially unnecessary freeze on a company’s working capital, and creates additional risks such as damage, loss, and stealing.

Given these risks and consequences, e-commerce businesses may decide to only keep a smaller amount of inventory on hand. Of course, that sounds great in theory, but how does one identify the optimal number of units to keep on hand? Today, many companies large and small are solving this challenge with demand forecasting, a method that employs statistical methods and qualitative insight to create simulations of customer demand to predict the most probable inventory needs.

Demand forecasting is a sophisticated field with a wide variety of forecasting methods, techniques and theories to choose from. In this article we’ll explain the basics. First, we need to understand that any forecast has assumptions, and reality will fluctuate from the estimates. So for any demand forecast, we need to estimate the standard deviation of potential demand. A better estimation of demand deviation will provide us with better insight into inventory planning.

We can all appreciate that the inputs to customer demand are full of randomness, and we have two possible ways to face this:

1) Leave it to chance, and learn from the results. This can actually be a great approach with a lot of insight to be gained,  but the problem is that it can be very costly and time-consuming.

2) Simulate different scenarios and select one with minimum risk. Of course, it does not guarantee that you will not fail, but it increases the probability of success. 

‍

To show the power of statistical simulations, let's consider a simple example:

We sell flowers. On average, we sell 25 bouquets every day. Still, one day we sell 23 and another 28, so the deviation of average sales is approximately 5 bouquets. Our price is $29 for one bouquet of flowers, and we purchase them for $9. We also have a cost of delivery of $2 per bouquet, storing and maintenance cost, which is approximately $1 per bouquet, and the labor cost of our florist roughly states $5 per sold bouquet. Calculating and allocating indirect costs is another important topic, which we’ll cover in another article. Finally, our profit for one unit sold is $12. Every 10 days our flower supplier delivers any quantity we want. The bouquet can be sold no later than 10 days after being delivered, after that, the flowers go to the trash.

With all the information we need to decide on the size of the replenishment order.

What is our first estimation without taking into account profit and simulation opportunities? One way of decision-making is: "We sell, on average, 25 bouquets every day, so let's just purchase 250 bouquets." Another approach is: "Well, we are a client-oriented company. Every guest should leave with flowers from our store, so 300 units will definitely serve every client." Once we order 250 bouquets, in the other 300, and after analyzing all the financials. Finally, we will find the best number, but will our business survive till that time, that is a big question.

The beauty of simulations is that we can do the same steps without spending significant money.

The simulation shows that the optimal order size is 225 bouquets, potentially providing us with $257 of profit. Surprisingly, purchasing 250 bouquets provides just $114, while 300 units, despite fulfilling all potential demand, will bring a loss of $636.

While this is a simple example, hopefully it’s a solid starting point towards understanding how simulating customer demand can be used to identify the optimal inventory on hand for e-commerce businesses.

Bucephalus is creating tools to efficiently forecast demand and much more. We’re a supply chain ops platform that empowers the millions of fast-growing e-commerce brands to move products faster, cheaper, and more sustainably using AI. Our product continuously processes customer data using our AI systems inspired by our team’s experience working in data science at Amazon.