Usually when you talk to people about data, first thing that comes up is the size, and not all the people that I talk to are males.
The amount of data is important, but the key for any data analytics project starts by asking the right questions, a stage that many are neglecting.
The definition stage can be very valuable. If you take the relevant stakeholders you will eventually create a uniform language in the organization.
A language that will define the business questions that your organization is facing with, in a way that can be answered using data.
Expect getting insights and better understanding of your business once this stage is completed. What you can and can’t do, and what is important to others in your organization.
No matter if you already established a solid BI (business intelligent) solution in your company, you still need to face the definition stage.
Having a reliable BI solution can help you speedup the 2nd phase of the predictive modeling process, the DATA and ETL (will not cover here).
Froude, Credit risk, Targeting, lifetime value, Churn and many other models are being used across different industries, but the definition of each one is different in various industries.
How can you build a Lifetime value model if your mobile app doesn’t generate any income?
What is the best “Churn” definition to your business?
How can you optimize your targeting efforts? What is it that you would like to optimize?
When you model your business questions, keep that in mind:
Business question modeling is not always an easy task. Adjusting and adopting a “predictive thinking” strategy will help you leverage your data and gain the competitive advantage that everybody is talking about.
In the past years, with the deeper penetration of the digital world, amounts of data collected have increased significantly.
The vast majority of today’s business and IT leaders agree there is huge value in putting Big Data to work. But how?
BI solutions are limited to explaining the past and therefor are considered as reactive solutions and not proactive.
Predictive Analytics is at the heart of proactive systems, creating a new type of “intelligent platforms” that collect data, analyze the data and react to events (like a user in a web site) automatically, or semi-automatically.
You are considering launching a campaign to acquire new customers through the web. How much are you willing to invest in acquiring a customer?
You suspect that a given customer is about to defect and switch to a competitor. Would you try to keep this customer or let her go?
You want to assess the value of your business. How would you evaluate the worth of your customers?
Well, all of these questions boil down to assessing the life time value of your customers, often abbreviated as LTV. Certainly one would be willing to acquire a new customer only if the cost of acquiring the customer is lower than the life time value of the customer. Not only companies are happy to get rid of unprofitable customers but in many cases companies may take active actions to make these customers defect to their competitors. And, finally, in the new world, and also in the more “conventional” industries, the value of a company is given by the aggregation of the life time value of their customers.