Enterprise Software Design Blog

Predictive Analytics – Predicting Future Sales

Posted on: January 1, 2010

Introduction

Predictive analytics is the latest buzzword in the enterprise software industry and it you go by predictions from industry watchers, every vendor will offer predictive analytics solutions. With most enterprise vendors acquiring predictive analysis technologies (i.e. Oracle acquired Hyperion, SAP acquired business objects and IBM acquired SPSS) we would definitely see some predictive analytics solution coming out from these companies. The customers are more than happy to give predictive analysis a try and why not in this economic malaise, if somebody is offering a sneak peek in to what’s in store in future, then why not.

Predictive Analytics

Predictive analytics is a technique by which the current and historical data are analyzed either thru sophisticated data mining tools or (and) applying statistical techniques to predict a future event. The future event could be predicting future sales, employee performance, budget requirements etc. The output could be either a detailed report (i.e. table, chart) or it could just be a single number. Predictive analytics is different from the traditional BI analytics in that predictive analytics looks at the present and historical data and predicts a future event where as tradition BI analytics looks same date and provides insights in to these data such sales by products, sales by territories etc.,

Traditionally application vendors provided predictive analytics tool as separate software that needs to be integrated with the existing database systems. Some tools will read the data directly from OLTP and data warehouse sources where as some other will define its own multi-dimension database (or cube) to load data in to it. These tools are capable for implementing sophisticated statistical techniques such as linear regression, time series forecasting, prediction model etc. The end users of these tools should be well versed with the business data, data warehouse concepts and statistical algorithms to effectively use these tools.

Pain Points

Though predictive analysis software offers real business benefit to enterprises, it is only used by very few with the enterprise. This is due the fact that majority of the enterprise users such sales managers, HR managers are not statistics expert and usually they don’t have enough bandwidth to engage statisticians and dive in to the maze of enterprise data. Senior executives such as the C’ level executives and VPs use these with the help of these experts. They basically use these to make important business decision such as budget allocation, high level territory performance and forecast etc. Most often these tools are not used regular day-to-day users.

To bring the benefits of predictive analysis to the end users, I suggest the following.

  1. Embed predictive analytics in to transaction pages
  2. Helps complete a business process
  3. Avoid technical and statistical terms in UI as much as possible.
  4. Do not over engineer

Design Solution

The example given below predicts the future sales of a particular product. It looks at the historical data ( i.e. from Jan – Sep ) and predicts the sales for the next quarter. This analytics will be used by a sales manager to view the sales target and the predicted sales through out a calendar year.The sales manager will use this tool to better plan and align the sales resources.

Month       Sales               No of sales rep
Jan             $100,000                 1
Feb            $120,000                  3
Mar            $123,000                  3
Apr            $110,000                  2
May           $112,000                  2
Jun             $119,000                 3
Jul             $115,000                   4
Aug           $110,000                  2
Sep             $100,000                2

Product Category         Customer Category        Sales
Desktop                                 2M-3M                                        $112,000
Desktop                                 3M-5M                                        $140,000
Printer                                   2M-3M                                       $120,000
Printer                                   3M-5M                                       $150,000
Laptop                                   2M-3M                                       $122,000
Laptop                                   3M-5M                                       $115,000
Services                                2M-3M                                      $120,000
Services                                3M-5M                                       $130,000

Predict Future Sales

The application uses time linear time series regression model to compute the future sales. The predictive model uses the following dependent variable to predict the future sales.

  1. Potential Revenue in pipeline
  2. No. of sales rep reporting to sales manager
  3. Product category sold
  4. Customer size

The sales manager can either use the standard forecasting method or use any of the standard averaging technique to predict the quota. Though this is little too technical for a enterprise user, having different technique till help the manager look at the other model and minimize errors and risk and it will give him confidence in the prediction.
The predicted sales and the red bar in the chart will change as and when any change is made. The model uses the liner regression with any of the chosen method and uses the time series forecasting to predict the future sales for the coming months and quarters. The user can adjust the dependent variables view real-time how the sales will be for next quarter and he can continue to do it until the target value is reached. At the end of the prediction, the end user will know what should the pipeline revenue, how many sales rep, what products should be sold to what kind of customers will be known and the he can strategize the sales activities

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Archives

Advertisements
%d bloggers like this: