May-Foods (Novo Forecast Enterprise)

2021

The company "May-Foods" is engaged in the production and distribution of tea and coffee, has 6 main brands. It produces more than 25,000 tons of products per year and occupies 22% of the domestic market share.

Each of the departments in the company had its own demand forecast and its own vision of the process. Often these data did not intersect and there was no single algorithm based on which it was possible to plan future sales. The accuracy of the demand forecast was extremely low (from 15 to 50%) and did not take into account many important factors.

Economic losses due to the inaccuracy of the demand forecast can be colossal. Based on this, the company came to the conclusion that it is necessary to work together so that each department works with unified data in a single planning space.

The company is large, it works not only in the Russian Federation, the CIS countries, but also in many others. Most of them do not grow tea and coffee, which means a sufficiently large logistical shoulder. In addition, the use of all possible sales channels: network retail, national and local networks, traditional trading through distributors, significantly increases the amount of data that must be taken into account.

The company came to planning in Excel, the data was collected once a month, but another problem arose: it is impossible to take into account everything. For a correct forecast, it is necessary to count not only the volume and weight, but also the cost and quantity in pieces.

There are factors, you can analyze immediately on the promo calendar, and there are those that depend on the level of the baseline, sometimes there is only promo, but there are no regular sales, there is an investment in the client.

When the company realized that Excel was not ideal, did not cover all the necessary needs, they began to look for other planning options. The solution was a digital system, where artificial intelligence, using algorithmic machine learning, analyzes the entire amount of data and considers the forecast, taking into account the factors of structural change in demand: listings, delisting, various locks, promos and about a hundred different subtleties of setting up forecast calculation.

The system adapts quickly as soon as new information appears, it recalculates the data and outputs the result. You can choose any network, fall to a specific product in it and see how many days it has on the roadlines, how much time is planned for the promo, what deviations, how many SKUs in the matrix, and so on.

Results of the Planning System

• The next step is to create an agreement loop for trade-marketing actions directly in a single digital platform.
• The next step is to predict commercial profit. There is already a revenue forecast, several factors need to be added to it: cost, commercial conditions, promo costs and get a quick calculation of the profit forecast.
• Today, the company has moved away from calculating the baseline, based on distribution and invented offtakes, now all data are based on sales statistics, clearing promo, different depths are used, depending on the characteristics of the client or territory.
• If earlier the accuracy of the forecast was quite turbulent, now it has been possible to stabilize it to 68% (an increase in accuracy by 20%, thanks to the digitalization of forecasting and planning processes) and make it continue to constantly increase.

The project was implemented jointly with partner Raitek DTG.