Algorithm for Forecasting Sales at New Product Launch in IT Retail

In the domain of sales and procurement, new product forecasting (NPF) is still risky adventure. The main reason is dearth of time series data for a new product, that can be utilized to predict customer demand, especially when the merchandise is utterly groundbreaking. Thereby, an inappropriate forecast conventionally causes a scarcity of products. On the other hand, the warehouse may be overloaded, due to poor inventory. Both situations can erode the firm’s profit. Moreover, the fierce struggle between companies for market share and technology growth has led to a shrinking of product life cycle (PLC) and thus, makes prediction even more sophisticated. Therefore, we problematize the NPF as a research objective. The attention is here paid to developing the new forecasting method stems from the PLC by analogy due to absence of historical data on the merely introduced product. This method ought to employ primary product characteristics, to cluster PLC curves by comparable products. We also explore, how changes of influence parameters may reflect customer preferences and consequently project demand. Finally, to meliorate model performance, it will be extended by a trend indicator. We prove that the diffusion model underpinned by auxiliary parameters, illustrates high accuracy result on the contrary to the ordinary model. The proper employ of a combination of judgmental and statistical techniques will improve operational cost and firm’s profitability. Keywords: New product forecasting methods, Diffusion model, PLC curve patterns, Google trends analysis