How to correctly forecast the demand for a new product and not miss the profit
Does your retail chain spend a lot of resources forecasting the demand for a new product, then come up with a plan for its distribution so that the “new product” appears on the shelves of the store? There are ways to solve this problem, which we will talk about today in our article.
Why should a retailer predict demand for a new product?
Inventory management (purchasing) specialists always face the question: how to predict and order a "novelty" in stores, or how to calculate the need for new opening stores? But there is always a risk of not covering the needs of buyers and "freezing" working capital in commodity balances.
You can leave everything as it is. But in this case, you take risks that you will not be able to meet the needs of customers and that the company will lose part of the profits or customer loyalty. But if the company wants to manage risk, it is necessary to improve the quality of calculations.
There are several approaches to predicting demand and ordering "novelties". Here are 5 of the most popular ones:
- Expert evaluation of an analyst or procurement specialist. The peculiarity of this approach is that there is a dependence on the quality of the analyst's work and the human factor.
- Fixing customer demand when customers leave a request for a new product, for example, if it is not on the site. This method is complex and opaque, since the client can duplicate orders, and this must be taken into account in further calculations.
- Calculation of insurance reserves rounded up to logistic units. This method is often used to order a one-time "novelty" and then track the sales history.
- Development of a sales history, for which forecasts of a new product or the required quantity are simply entered into the order during the trial period until the data is sufficient to evaluate the effectiveness of the “new product”.
- Using the approach "goods-analogs". The method allows you to calculate the demand forecast for a new product based on historical data from previous periods in conjunction with the SKU already presented in the assortment.
The "goods-analogs" approach has proven to be practical, as the chain sells similar products, and this experience helps predict future demand.
Analyze your business processes with our demand forecasting and ordering checklist and see what can be improved and automated!
The essence of the "products-analog" approach
Forecasting demand and ordering a new product must be started with the assumption that the store will sell the new product similar to the sales of the analog. If the products differ in volume, weight, type of packaging, and other characteristics, a coefficient of uniformity (homogeneity) is applied, which allows you to convert the sales history of the analog and the new product into comparable logistic units of measurement.
The coefficient allows you to link two similar products both at the SKU level and at the LU (logistics units) level. In this case, the connection consists of two components: the type of connection and the coefficient of uniformity.
In sales forecasting solutions, relationships are used to trigger a forecast and order a new item and allow users (purchasing and inventory management) to determine how the new item can be matched to the SKU in stock.
Options for applied linking types:
- a "substitution" - this link type is when a "new" item permanently (forever) replaces an existing SKU. The history of both products will be combined to calculate the demand forecast;
- the "variant" link is means, that the history of both SKUs will also be combined to calculate demand forecasting, only a new product temporarily replaces an existing one, and there is also the possibility of joint accounting of balances;
- the "reference" link is when the sales of a "new product" are similar to those of another existing product. For weeks when no new product is purchased, reference product data is used. As soon as a new product has its own sales data, the history of the reference SKU is no longer included in the forecast.
After the binding type is determined, the coefficient of uniformity between the new product and the existing SKU, you can begin to predict the "new". How this works in practice, we will analyze in the next paragraph.
Have questions about demand forecasting or auto-order systems? Ask them to our specialists: write to [email protected] or fill out the form to order a callback.
Examples of demand calculations for a new product
As we have already said, the best way to determine the right amount of a new product is through calculations, not predictions based on speculation. Let's look at how product links are used to calculate demand, using business cases as an example.
The "substitution" link
"The Drink 1 liter Cherry" will permanently replace "Drink 1 liter Lime". Therefore, a "replacement" relationship is created between the two products. In this case, the forecast for "the Drink 1 liter Cherry" will be calculated based on the sales history of "Drink 1 liter Lime", as well as the first sales of the product "Drink 1 liter Cherry".
Let S be "the Drink 1 liter Cherry", which will replace "Drink 1 liter Lime" (A). k is the homogeneity coefficient of units, the meaning of which was described in detail in the previous paragraph. In this case, the calculation formula will look like this:
Forecast(S) = History(A)*k + History(S)
The quantity of a new product for an order will take into account the sales history of the analog adjusted for the uniformity factor, and subsequently, the sales history of the new product will be added.
The "variant" link
The manufacturer of the product "Drink 1 liter Cherry" has created a new special package (label) for it, for example, during the holidays. Because it is a variant of a product that affects the sales and forecasts of the main product, a "version" binding is created between the two products.
Let's denote V - SKU with special packaging (label) for "Drink 1 liter Cherry". In turn, it affects A - the demand for "the Drink 1 liter Cherry", with which it is associated. k is the homogeneity coefficient of units. Then the calculation will be as follows:
Forecast(A) = History(A) + History(V)*k
The sales history of the product with "temporary" packaging will be added to the quantity forecast for the main SKU, taking into account the uniformity factor.
The "reference" link
"Drink 1 liter Lemon" is a new taste that will inherit the sales history of "Drink 1 liter Mandarin". Since drinks share a common characteristic "taste", we can use the history of a tangerine-flavored drink as a basis for the forecast. The "reference" binding is created between the two products. "The Drink 1 liter Lemon" will first use the sales history "Drink 1 liter Mandarin" and then its sales.
Let N be "The Drink 1 liter Lemon", which will inherit the history of "Drink 1 liter Mandarin" (R). k is the homogeneity coefficient of units. Accordingly, the calculation formula will be as follows:
Forecast(N) = History(R)*k + History(N)
The forecast calculation is performed for two products, and the new product does not replace any existing one. The connection itself is created so that the "new" product can inherit the history of the reference product, and does not affect its forecast in any way.
In conclusion, it is worth noting that to improve the forecasting and order of “new products” in your company, you can use any convenient approach: rely on the opinion of experts or create a minimum requirement to fill the shelves. All of these exist approaches. The “analogous products” approach we have considered is a tool that allows you to model forecasts, order new products and make management decisions based on statistical data. The main thing is to start moving towards solving this issue, which will improve the quality of the forecast and calculation of orders.
The C4R team has over 15 years of experience in business process automation, delivering proven solutions and systems for demand forecasting and auto-order. Want to learn more about solutions for your business? Email or fill out the form below!
The author of the material is Anton Tsemerov,
ERP Consultant Consulting for Retail