Sales & Volume Planning with Promax PX

21 Jun 09

The forecasting capabilities of Promax have been a core part of the application’s functionality for many years. Since the inception of the system it has addressed the need to provide a forecast of financial data to be the basis of accruals for deferred trade promotions expenditure. This soon involved the transformation of those forecasts into revenue and volume estimates for each promotion. Those estimates required the development of algorithms to calculate baseline sales and seasonal effects to further improve forecasting accuracy.

The innovation and research at Promax over the last couple of years has taken the modelling and sales/volume planning capability of Promax PX to a new level, placing it at the forefront in Trade Promotions Optimisation and Sales & Volume Planning. This paper describes best practice for the combination of activity based forecasting, statistical modelling and market intelligence to give a holistic approach to sales/volume planning for CPG (Consumer Packaged Goods) companies.

WHY ARE CPG COMPANIES DIFFERENT?

Most computer based demand planning systems were derived to solve the problems of manufacturing and purchasing organisations with large SKU (stock keeping Unit) counts and limited amount of “market intelligence” on consumer demand signals. They are typically based around statistical models derived from sales (ex-factory/warehouse) in monthly buckets. The fundamentals are there, in that they all produce a statistical forecast that reflects the seasonality of the sales and its growth or decline from the current level for each SKU. The frustration for CPG businesses is these tools do not deal effectively with those products that are highly promoted. Promotions cause large spikes in the demand, way above the baseline sales rate. The demand signal often appears confused as the spike hits the supplying warehouse well before the promotion is launched to the consumer. The incorporation of this information into traditional demand planning systems consumes endless hours of debate and consultation in the Sales & Operations Planning forum often resulting in poor forecasting accuracy, expensive and unrewarding business processes and inevitably the consequential out-of-stocks and expedition costs that are all too familiar.

The renowned forecaster R G Brown, who lays claim to have invented the statistical modelling technique known as exponential smoothing, has been asked many times by those trying to solve the forecasting dilemmas “What is the best source of data for forecasting?”. His reply, with quintessential ambiguity, is consistent “You should try and use the most reliable source of data that most accurately records consumption”. That used to be ex-factory sales from invoices/shipments but now for CPG companies there are reliable sources of POS (Point Of Sale) data that are clean and verifiable. The granularity of this data is improving with weekly sales by product/customer readily available from syndicated data providers. A number of companies are working with their retail customers to obtain data on a daily basis for each store.

The real point of difference is that CPG companies need to understand the forecast from the perspective of baseline sales and promotional uplift. This requires a multi-causal approach to modelling data that concurrently models baseline, trend, seasonality and the effect of price and other promotional factors. Furthermore the character of the customer’s buying behaviour has to be understood and modelled as the generation of a useful forecast for the supply chain requires a transformation of what the consumer will buy from the check-out to replenishment orders on the manufacturer from the retailer for re-supply. These models need to operate effectively without the need for inventory data, as this is rarely available in a timely and convenient form.

DEVELOPING AN “ACTIVITY BASED” FORECAST

An “activity based” forecast is one where events such as promotions are individually defined and modelled. To derive an activity based forecast we need to decompose the historical demand data into identifiable elements that can be used to establish the forecast.

Read more about “Sales & Volume Planning with Promax PX” by downloading the complete whitepaper