How Data Driven Approach Turns Prescriptive Analytics Into Predictive Analytics Roadmap
How data Driven Approach turns Prescriptive analytics into Predictive analytics road map.
Current market tools & most widely industry accepted analytics are focused on data capturing. For Eg., Google Analytics, Adobe Analytics (Omniture) etc., are good at viewing the basic reports based on the device, OS, location, custom tagged report. Any action or event triggers based on the sequence of events is going to be a cumbersome job. At the end, marketing team needs close the loop of sending customized email/sms/push notification to engage the end customer. These descriptive analytics will always be irrelevant within few months and there is a need to upgrade.
When to be upgraded? Do we need to monitor descriptive analytics daily? How can i do it in a more tactical way?
Strategically, assign simple decision making points to your analytics data. This will have direct impact on your descriptive / prescriptive reports and it is time to upgrade the strategy. Let’s take an example, as a product owner you are receiving daily reports of how many online shoppers are visiting a product at what interval of time , from which devices , which browser versions etc., . This will be a plain table report. Add some decision points, percentage of visitors visiting a product, percentage of your visitors checking out the product. This will help us to take immediate action to address the dropouts by responding with strategic offers/discount. More the marketing team has bandwidth , the more the networking & potential partners can be reached.
Getting in Deeper:
Let’s not deviate our point on getting meaningful information & meaningful decision points. Anything we want to predict for near feature needs detailed data and start with simple decision making ruleset. Whenever, we start addressing the predictive analytics, it goes with business priorities rather than the technology or process. Let’s start with an example, if e-business has 1000 products identify which products are having maximum profit margins & products that are close to zero return shipment. Remember, return policy of products is associated with the operating cost for the business.
Start Scoring for data decision making:
As a product owner, we would like all customers not to return the shipped products. This is not an ideal statement for calculating the forecasts & profit margins. However, let’s divide the products including the physical products into categories based on their expiry date like short term expiry, long term expiry and no expiry (eg.,electronic gift cards).
Scientifically divide the products into a simple score card based on the returns.
Physical products with short expiry date (say Type PS) - Food, mobile phones (frequent model upgrades), medicines etc., . Needs extensive marketing , high margins & lower satisfaction ratio, price warrior, return shipments, challenge to have shorter shipment duration . I would like to score it as 40 out of 100
Physical products with longer duration & High shipment Returns (say Type PLH) - example - electronic appliance, House hold item’s etc., High Margins , price warrior , return shipments , satisfaction ratio . I would like to score it as 65 out of 100
Physical products with longer duration & Less shipment Returns (say Type PLL) - example - Diapers, Soaps, detergents, cleaning liquids, Inner wears, packaging needs . Very high expiry date , less impact on satisfaction ratio, Return shipments is negligible . Greater the marketing & greater the advantage. I would like to score it as 85 out of 100
Non Physical products (say Type NP) - example - Cash Gift card, App store purchase, Reward point Gift card… percentage of returning products will be less & it can be handled through electronically. This makes us to stick with the store brand, though margins are less. I would like to score it as 95 out of 100.
All the above scores are related to the probability of success. In other sense, how much confident on your forecasting of the business based on the percentage of sales happening on the product type.
Now it’s time for the adding the score for all the products as probability or type of the product (Type PS, Type PLL, Type PLS, Type NP) .
Now divide your sales report not only as product, but also as a type of product. This is the basic start for doing predictive forecasting and you know your confidence on the numbers.
Sample Calculation for finding the product confidence:
Maximize the revenue opportunites using data driven confidence factors
Above table shows us, confidence on each product type based on the return policy, better margins. Now it’s time to focus on increasing the investment on specific type of products rather than products which are on high traction & associated with high product returns. Now, to remarket your products, you don’t have to consider all the products, but can prioritize based on the identified product type for that customer. Success in remarketing and digital marketing can happen only when you have full control of your analytics data. Choose the options which are supporting private analytics tool/product and vendor free lock in .
Based on the previous architectural decisions on a complete marketing automation , we are able to leverage CustomerEngagePro product hosted in our private cloud. One good thing with CustomerEngagePro is, No-lock between the hosting companies and able to generate customized marketing events.