System and method for analytics-driven sales revenue prediction
Publication Date: 2014-Nov-07
The IP.com Prior Art Database
Sales revenue prediction is a challenging task due to subjective judgment involved and complex sales process. This invention is aimed to tackle these two technical problems: subjectivity and complexity, the goal is to improve prediction accuracy in a systematic and effective way. Our basic approach is to use balanced sales data and adopt multi-stage combination method driven by analytics. A computer-implemented software system is proposed to collect and make good use of various sales-related data in existing CRM, ERP, and SCM system. At the same time the system incorporates analytics into all the processes of data processing, revenue prediction and executive reporting. Our invented system can provide sales executives with an accurate and rolling quarterly revenue prediction on a weekly basis before the quarter ends.
Page 01 of 16
System and method for analytics
System and method for analytics-
--driven sales revenue prediction
driven sales revenue prediction
Sales revenue prediction/forecasting is a challenging task due to subjective judgment involved and complex sales process. About subjectivity, sales raw data are collected and estimated by sales departments and salespersons, these data may be biased even manipulated. Many existing prediction methods incorporate judgment, i.e. they are judgment-based forecasting. Sales process and sales revenue are affected by many factors, including internal business process and external economic environment, and these influential factors change over time.
This invention is aimed to tackle these two technical problems: subjectivity and complexity, the goal is to improve prediction accuracy in a systematic and effective way. Our basic approach is to use balanced sales data and adopt multi-stage combination method driven by analytics. Specifically, we are designing a better approach by taking advantage of objective sales data to offset or minimize subjective effect in data preparation and to adopt combination method in data preparation, modeling, and final result attainment. To implement a sales revenue prediction system with satisfying prediction accuracy, we leverage analytics in each stage of the whole sales revenue prediction cycle.
Various sales revenue prediction algorithms, such as linear model, non-linear model, time-series analysis, Bayesian model etc., have been developed and implemented in computer system. Due to the complexity of sales process, single prediction algorithm can't achieve satisfying prediction accuracy. Combining forecasts is widely accepted as an effective way to improve accuracy [Paper1]. However, no effective and practical combination method is established and therefore satisfying prediction accuracy is not achieved.
[Patent1] tried to overcome subjectivity by using Bayesian principle and including sales associated conditions. However, Bayesian principle still needs user's judgment as prior probability, and sales associated conditions are not defined in that invention. [Paper2] proposed a system and method for sales volume prediction by products, not sales revenue prediction by sales sub-organizations. Meanwhile, just single model (dynamic linear model) is used, not combined models, and the system is customized to a specific system and hard to extend to other companies
[Paper3] proposed a Bayesian modeling framework for unit sales and price prediction by taking external sales drivers into consideration. It omits internal sales drivers and does not model the relationship between sales revenue and sales drivers within a sales sub-organization.
Sales revenue prediction involves complex situations. Companies can have different modes of selling and disparate products. In some situations, sales revenue prediction is referred to as annual sales revenue projection or revenue budget. This invention...