Forecasting Trade Promotion (TPF) is a process that tries to find some correlation between trade promotion characteristics and historical demand to provide accurate demand forecasts for future campaigns. The ability to differentiate increases or requests due to the impact of trade promotions as opposed to basic demand is crucial to modeling promotional behavior. Model determination allows what-if analysis to evaluate different campaign scenarios with the goal of improving promotion effectiveness and ROI at the channel-product level by choosing the best scenario.
Video Trade Promotion Forecasting
Challenge of trade promotion forecast
Shopping Promotion Trade is one of the largest expenditures of the consumer goods industry at a cost to large producers ranging from 10 percent to 20 percent of gross sales. Understandably, 67 percent of survey respondents recently said they were concerned about return on investment (ROI) earned from the expenditure. Measuring ROI relies heavily on the ability to accurately identify "baseline" requests (demand that will exist without the impact of trade promotion) and improvement.
In fact, the accuracy of forecasts plays an important role in the success of consumer goods companies. The Aberdeen Group study found that the best-forecasting firm in its class (with an average estimated accuracy of 72 percent) had an average gross margin increase of 28 percent, while the slow-paced firm (with an average forecast accuracy of only 42 percent) had an increase in margins gross less than 7 percent.
Estimated bottom-up sales at SKU-account/POS levels require consideration of product attributes, historical sales levels, and store specifications. The many different variables that describe product, store and promotional attributes, both quantitative and qualitative, potentially have many different values. Selecting the most important variable and putting it into the prediction model is a challenging task.
Apart from these challenges, two thirds of companies in the consumer supply chain consider the approximate accuracy as a high business priority. 74 percent said it would be helpful to develop bottom-up estimates based on inventory storage units (Sku) by key customers.
Maps Trade Promotion Forecasting
Traditional trading promotion approximation method
Many companies estimate the impact of trade promotion primarily through the approach of human experts. The human expert can not take into account all the variables involved and also can not provide analytic predictions about the behavior and trends of the campaign. A recent survey by Aberdeen Group shows that 78 percent of companies use Microsoft Excel spreadsheets as a forecast tool for their major trade promotions. Limitations of spreadsheets for planning and forecasting of trade promotions include lack of visibility, ineffectiveness and difficulty in tracking cuts.
Special trade promotion forecasting applications have been developed and become more common. 35 percent of companies now use legacy systems, 30 percent use Sales and Operations Planning (S & amp; OP) applications, 26 percent use integrated Enterprise Resource Planning (ERP) modules and 17 percent use home commerce promotion solutions developed. This application supports the planning process, while still relying on human knowledge and intuition to estimate. One problem with this approach is that humans tend to make optimistic assumptions when predicting and planning. The result is the most commonly wrong estimate on the optimistic side and that human forecasters also tend to underestimate the amount of uncertainty in their estimates.
The next problem is that the legacy trading promotion system contributes to the internal fragmentation of trade marketing data. Many companies that use these tools currently produce assumption-based estimates with limited accuracy.
Analytic approach to traffic promotion forecasts
TPF is complicated by the fact that campaigns are represented by both quantitative variables (such as prices and discounts) and qualitative (such as showrooms and support by sales representatives). New approaches are being developed to address this and other challenges. Most of these approaches attempt to include large amounts of heterogeneous data in the forecasting process. One researcher validates the multivariate regression model's ability to forecast impacts on product sales of many variables including price, discount, visual merchandise, etc.
The term Big Data describes the increase in the volume and speed of heterogeneous data coming to the company. Data can be used to improve the accuracy of trade promotion forecasts as they usually contain real relationships and causalities that can help to better understand what customers are buying, where they buy them, why they buy and how they buy. Often, the challenge is to combine this data across all the silos in the organization for a single look.
Traditional methods are not enough to assimilate and process large volumes of data. Therefore more sophisticated modeling and algorithms have been developed to address the problem. Some companies have begun using machine learning methods to capitalize on the large volumes of unstructured and structured data they have held in order to better understand these relationships and causality.
Machine learning can make it possible to recognize the joint characteristics of promotional events and identify their effects on normal sales. The learning machine uses a simplified version of a nonlinear function to model complex nonlinear phenomena. Learn the process of machines processing input and output data and develop their relationship model. Based on this model, the learning machine predicts the output associated with the new input data set.
Intelligible Machine Learning (IML) is an implementation of Switching Neural Networks that has been applied to TPF. Starting from a collection of promotional characteristics, IML is able to identify and present in clear form correlations that exist between relevant attributes and improvements. This approach is designed to automatically select the most appropriate lifting model to illustrate the future impact of a planned promotion. In addition, new promotions are automatically classified using pre-trained models, providing an easy way to learn about different what-if scenarios.
The TPF system should be able to connect and analyze large amounts of raw data in various formats such as company sales history and online data from social media. The analysis should be done very quickly so that the planner can respond quickly to the demand signal.
Danone Groupe uses machine-learning technology to forecast trade promotions of fresh products characterized by dynamic demand and short shelf life. The project improved the approximate accuracy by up to 92 percent resulting in an increase in service levels of up to 98.6 percent, a 30 percent reduction in lost sales and a 30 percent drop in product obsolescence.
References
Source of the article : Wikipedia
0 Comments