PETROANALYTICS PROBLEM STATEMENT “Data Analytics is the process of discovering meaningful new relationships, patterns and trends by shifting through data using pattern recognition technologies as well as statistical and mathematical techniques.” A Data Analytic Tool uses the combination of powerful Statistical Analysis and Data Mining equations to predict the useful information for a Petroleum retail outlet manager. As the Retail Outlets (RO’s) are the prime areas for obtaining retail related data of any Oil Marketing Company, development of data analytics for monitoring various patterns in the Retail Outlet is an essential factor. The Inputs to the system will be selection of various attributes by the User and the Output would be the computed results of the data analytics that would enhance the User’s (Retail Outlet Manager, Regional Office Manager, etc.,) decision-making capability, which would be backed by concrete facts. Since Retail Outlets (RO) are a major point of interaction with the customers and their goldmine of data, the Oil Marketing Company’s treat the Retail Outlet as their premier face-outs. This generates the need for in depth analysis of data for higher operational capability of the Retail Outlet and also the study of customer pattern at this level".
Current Business background & introduction Oil Retailers have to do many tasks. Retailers have to analyze their customers, develop strategies, choose markets and channel to compete, make Location decisions. Retailers do design, purchase, price and promote merchandise and services. They organize their operations and manage their employee and stores. Retailers create an atmosphere that is inviting to customer and conductive for buying. It is very important to understand the customer behavior at retail outlet level. The development of Retail services in Oil & Gas Company arises the need for powerful analytic tools in order to analyze and interpret the huge amount of data for fact based Decision-making. Developing powerful Data Analytics Tools at the Retail Outlet Level could satisfy the business demands of an oil marketing company. The main purpose is to integrate the Retail Management in Oil and gas Industry and to develop and customize the various ‘Data Analytic Tools’ so that they could be utilized to enhance the Operational Capability and Profitability at the Retail Outlet Level. Data analytics can also identify opportunities to change fueling patterns, leading to campaigns that can increase fuel sales during non peak hours of the day. For most of the Retail players today, Coverage is the key driver. The more cities and outlets they launch their program in, the larger their member base naturally grows. After coverage begins to hit a plateau, data and analytics will take center stage as companies plunge into the deep end of data mining for elusive insight to drive their campaigns. In addition, analytics driven campaigns is the benefit of member database present. A company could launch a new product, a new fuel blend, a new facility at its outlets, a new partnership and reach out to its best customers using the results of data driven data analytics tool.
Scenarios and problems areas in retail Analytics
o o o o o o
To To To To To To
get the idea of profitable customers / Profitable segments get the idea of sequence of service to be provided help in taking decisions for higher operational capability help in achieving higher sales at the retail outlet level know the no. Of Products / Services per sale. have plan on what type of merchandize/product mix to be kept in the Customer-Store
o o o o
To get the clarity in what new services to meet the demands of the various customer segments To identify the implications of loyalty programs / promotion programs on product sales. Retail managers are not in the position to correlates effect of another variable (such as environmental etc) on the sales The retail manager will provide the details of period, on which the system should retrieve data, use Analysis techniques .Attributes are retrieved from databases for analysis purposes using Database Connectivity establishments according to the statistical algorithms of Predictability, flexibility, Trends analysis.
The key areas of analytics also include o o o o
Peak hours of operations Correlation analysis between Volumetric sales of various grades to Product grade price Impact of promotional schemes on sales volume. Sales analysis (Trend Analysis) - In the sales analysis context, the real time analysis of the data should be
enabled, which at present is available only for past data. o
Sales Forecasting Analysis Using equations (both short term and long term forecast) - The forecasting requires forecasting analysis to be done for weekly, daily basis and yearly
basis. o o
High value customer identification Potential customer identification
o
Gap Analysis: This functionality must cover the performance analysis between any two retail outlets in terms of fuel sales and revenues generated. This gap analysis determines the performance levels between two retail outlets. Multiple correlations Analysis: The existing system possess the correlation analysis between two product grades. This should be enhanced with multiple correlation of product grades sales .There by the relationship between effect of a group of grade sales over a particular grade can be undermined. Forecasting analysis using mathematical Model: Model is to be used to predict the short term sales using three parameters in this model namely pricing factors, promotional cost incurred and past period sales .This model must have advanced mathematical equations to be solved by the system. Profitability Analysis: This functionality examines the profitability of the retail outlet on any particular day or any interval based on the various parameters such as total sales on product grades and other expenditures. sales order analysis: This analysis lists out the list of sales order generated by the retail outlets at any dates and the list of products delivered on time .Here the analysis can again be done on product receipt times and average fuel quantity ordered between any intervals of time. New product grade forecasting: when a new product grade gets launched by the company the sales data available for the product will be ranging between one month intervals to less than two years. In such cases retail outlet existing analytics doesn’t have the functionality of forecasting .So, the concept of single exponential smoothing analysis and decomposition algorithms have to be developed. Top list of product Analysis: This functionality is required to get information about top selling products which are either volume wise (or) revenue wise.
o
o
o
o
o
o
Develop Analytics and model problematic areas for Oil retailers
to
solve
the
above
More information on Retail Business Environment In any retail outlet the transactions are basically made of primary and secondary activities, which may be monetary or non monetary. The Primary business activity includes sales of Petroleum Products. The secondary activities are focused on the extended services to the customer, which are responsible for increase the ROI. Secondary activities catch the attention of different customers in order to boost the product sale. The secondary activities also play a major role in the revenue generation of the retail outlet. The primary activity comprises of selling products like Gasoline, Diesel, LPG, CNG, Lube oils etc. The secondary activity serves the arriving customer using a variety of services which includes Convenience- store, Vehicle Service, Air station, Bill payment, Rest room, Restaurant, Parking. Retail Managers are interested in the sales improvement of petroleum products through the value added services of the secondary activities. These secondary activities incur a variable cost on the retail outlet performance. The major players in oil retailing across the world have customer loyalty programs. The Oil companies have introduced their individual promotional schemes in the form of various customer loyalty programs like petrocards etc. These companies are focused on their marketing efforts on building their retail outlet brand and services. Analytic Constraints Data analytics tool requires large amount of data to show accurate results in form of graph. Hence the availability of such data which is obtained as part of the real time business environment should be a validated data on which various checks are performed. This information may have to be obtained from the dealer of retail outlet [a case where Retail Outlet is not operated by Oil Company]. In case where the retail analytics works on incorrect and invalidated data the results of the systems would be a divergent from the normal. Analytics of the data can be of two lines namely industry specific metrics and company specific metrics. The data analytic tool focuses on the general industry metrics which can be applicable across all the oil retailers irrespective of the company specific metrics, though these requirements can be incorporated as per retail outlet requirements. The constraints that could influence the objectives are unknown profitable segments, unknown idea on the sequence of services to be provided, lack of clear plan on product mix to be kept in the Convenience store and ineffective implementation of loyalty programs / promotional schemes. In US market quality and price are considered as an important factor in oil retailing. Hence importance will be given while designing the Data Analytic tool. The geographical location of the retail outlet plays a significant role in product sales. Revenue generation will be higher, if retail outlet is present at busy lanes (i.e. free way or highway). Promotional schemes also play a significance role in product sales at different location. The systems available in the market are less powerful, non cost effective and cannot be easily integrated in the disparate computing environment. Data analytics require interfaces to the data warehouses of the retail outlet. The existing system though capable of producing accurate results needs the above mentioned performance criteria. Generation of business analysis reports (in form of graphs) are shown to the retail manager .Here the relationship between the dependent and independent variables should be shown with their legends. The graph shows the effect of other variables (random) on business sales. A detailed analysis of the sales data can reveal many action areas for different variables. When there are multiple decision variables, it is recommended to take a systematic look at all the variables. The
systematic look should be seen as an art of seeing various relationships, their interrelationships in the whole so that those structures are visible. In retailing of oil industry, the margins are very thin. Also the retailers face intense competition from their competitors. Any retailer would like to know about the predicted sales for the products of various grades for the coming period of months. The demand number for each product for the forward month or a week is one of the key inputs which are also being called as PDP (Pre-dispatch planning) schedule. This input actually gives the rate at which the refinery operates for the next month. The following inputs give demand numbers i.e. forecasted data of the retail outlet. o o o o
o o
Historical data for the same period and the latest sales Marketing inputs based on marketing plans to introduce any promotional schemes Forwarded product prices in the coming months In mathematical terms, time series analysis is the effective technique used for forecasting of sales. Time series Analysis accounts for the fact that data points taken over time may have an internal structure such as Auto correlation, Trend or seasonal variation .The usage of time series model is of two fold To Obtain an understanding of the underlying forces and structure that produces the observed data To fit a model and proceed to forecasting, monitoring or even feedback and feed forward control.
Hints and points to consider in your solution: (1) In United States there could be 6 grades of gasoline, diesel etc. so companies would like to correlate the sales of grade fuel. This would give them an idea to introduce new grade fuels or abolish if any. (2) Sales of some retail outlets products might be high in some localities during school timings etc. (3) Sales volume might change during peak summer times and would be low during winter seasons. It can go high during festival seasons. It can follow cyclic and acyclic patterns (4) To promote sales , a RO manager could reduce the price of fuels on display outside from his nearest RO competitor (5) Part of promotional campaign would involve inviting customers for free car wash, granting fuel points, gifts etc. (6) An addition of a new fuel station (retail outlet) could impact the sales of the existing retail outlets. (7) Customers are identified using unique customer numbers (fleet cards) , credit cards or vehicle number (8) Oil companies in United States tend to increase their sales by imparting promotional schemes. Very often sales Maximization is taken as the single largest target of retailing. Here sales volumes are increased through greater promotional effort. The increased sales promotion has some additional cost to it which has nullified the effect of improved realization on account of more number of units sold. Analytics must help the company to know the extent of success of its promotional schemes, there by giving them scope to plan their promotional activities. Since promotional periods generally ranges between 1-20 days. (9) Fuel Prices can change over week, months etc. In United States fuel prices could change even 6 times a day.
