Distribution Management Application provides distributors with real-time information to manage their business with the CPG Organization. It assists in fetching real time information from ERP and Distribution Management Softwares.

Information Exchange

• Orders, Deliveries and Shipments
• Sales and Promotions
• Invoices, Credit/Debit, Payments
• Damages and Claims

Trade Promotion Optimization attempts to discover multiple linear and non-linear relations between features of trade promotion and historical demand in order to simulate accurate demand forecasting for future campaigns. Uplift Modeling is a predictive modeling technique which should be applied in order to distinguish baseline demand from the effect of the promotion.

Trade Promotion Management contributes to 15-25% of Top-Line Revenues of almost all the Consumer Packaged Goods Companies. Optimizing this would have a direct impact to the bottom-line of the company. The nuances of Trade Promotion are highly complex due to multitude of factors affecting it and these factors reside in the mind of Sales Executive Managers who deal with it on a day to day basis, making it difficult to scale these insights to at the whole organization level. It is therefore important to develop a progressive analytics approach that is scalable, modular and responsive to the current and evolving business needs.

Why Traditional Software are not able to solve this problem?

A one size fits all approach to solve Trade Promotion Optimization has not resulted in expected benefits primarily because of singular, finite mathematical approach undertaken in most Trade Promotion Optimization Software available in the market. A short term analytical project might solve the problem in the short term but is unable to scale the effectiveness of the insights hidden in the data in terms of continuous improvement.

What Data is required for Effective Analysis?

• Secondary Sales Data & POS Data at SKU Level
• Shipment Data in both promoted and non-promoted weeks
• Historical Promotion Data

Harmonizing and Sanitizing this data is the first challenge to be overcome in order to build Machine Learning Models.

How is Machine Learning Model superior to existing Software or one of analytical projects in CPG Organizations?

The linear approach to solve Trade Promotion Optimization fails because of multitude of factors affecting the desired business outcome, massive increase in growth of data and because additional sources of data might become available in the future which are not currently getting accounted.

Advanced Analytical Model built on underlying methods like Neural Networks are able to identify non-linear relationships between factors which might otherwise remain unidentified by human brain. The models built are scalable and modular as per the necessary business outcomes. This makes it possible capable of using the right models and techniques, capable of incorporating both existing and new data sources. The models are self-learning in nature which improve over a period of time and are capable of providing insights to the Sales Executive Managers, taking the role of an Artificial Intelligent Decision Support System.

Trade Promotion Optimization is possible but requires an analytics partnership with an organization which not only understands data science but has immense domain experience. It is also important to make the sales team understand the model and capable of tweaking it on a regular basis so that the forecasts and insights given do not become a black box but a Decision Support System that is capable of providing better and informed insights.

Marketing Mix optimization can help CPG companies to optimize their spend plans across different media channels - TV, Print, Digital, etc. Our solutions can help CPG companies improve execution of their advertisements and promotions, generating higher ROI from their campaigns.

By merging the sales and marketing data, NeenOpal can help organizations start making accurate marketing predictions, by finding answers to some of the questions listed below:

• Which product should the company promote this month
• What type of campaign will be most profitable for this product
• What consumer segment should the company target
• How can the company get value from its social media data and use current consumer sentiment to create timely marketing campaigns

FMCG products usually have a short shelf life and companies incur significant costs from an over supply or over manufacture condition. NeenOpal's demand management solutions help CPG companies prevent such conditions by accurately understanding and predicting demand at different levels - by SKUs, category, brand, by different retail partners, by region and at the country. CPG companies also stand to benefit by optimizing their distribution and transportation fleet system - while maintaining various demand fulfillment constraints.

Our models can help CPG companies find answers to some their most critical questions:

• How can we guarantee on time delivery
• How can we shorten the time to manufacture a product
• How can we minimise product returns / complaints

One of the areas where Machine Learning has made a significant impact is Intelligent Forecasting as compared to traditional forecasting largely derived from historical trends. With the availability of Cloud Computing and evolution in Data Science, CPG Organizations are able to leverage actionable insights that are predictive in nature. The term 'Growth' is married to CPG Organizations, be it portfolio growth due to new product introductions, growth in terms of geographical introduction of products and growth in terms of multiplying sales channels.

But this growth creates a challenge of how to accurately forecast for new products, new channels and continuously evolving competition and landscape. Due to these factors, traditional forecasting techniques have a low level of accuracy. We often come across our clients that suggest that inaccurate forecasts is an impediment to improving their service levels. The problem is even more acute for certain exceptional items for which the difference between the actual and forecast is often more than twice which leads to a lot of organizations in a situation in which they don't know what to do to improve the forecasts of these items.

How Machine Learning Forecasting is different from Traditional Forecasting?

1. The data is split into test data and training data e.g. 30% Test Data and 70% Training Data. Various Forecasting Algorithms are tested at the SKU level and the "Best-Fit Forecasting" is automatically selected for current and future scenarios as well.

2. Traditional Forecasting is only able to check for linear relationships whereas Machine Learning Forecasting using methods such as Neural Networks can detect non linear relationships and incorporate them into forecasts.

3. Traditional Forecasting is characterized by manual manipulation and cleansing of data whereas Machine Learning Algorithms are scalable to multiple data streams without cleansing. e.g. Price, Promotions, Advertisement, Weather, Temperature, Features all in one place.

Application for a CPG Organization

• Self Learning Forecasting Algorithms continuously improving in performance
• Incoporating Price Elasticity and Promotion Data into Forecasting
• Incorporating Complex Seasonality Traits
• Incorporating Climate and Weather Data
• New Product Introduction

Power BI Implementation

Coming soon!