Differences between business intelligence and predictive analysis

Recently, I was discussing the difference between predictive analysis and business intelligence with a colleague. To simplify it all, I used an example taken from daily life, involving household technologies that we use every day.

Last summer, as the temperature hovered around 32 degrees Celsius, I decided to go for a 10km run. Very few people would have been able to predict that I would go jogging under this blazing sun and at close to 100% humidity (not the best idea in the world). The results of my run were rather disappointing. The application I use to monitor my runs – Runkeeper – indicated that my performance was among the worst of the last few months.

I explained to my colleague that Runkeeper offers me a descriptive analysis of my performance, a sort of business intelligence that allowed me to watch my progression without giving me any indications with regards to my next training sessions.

I continued by explaining that a friend uses his much more sophisticated Garmin 920XT watch instead, which provides him not only with basic information, like my app, but also offers tailored programs and sessions taking into consideration factors such as weather, recent performances, goals, etc. …

In this sense, his watch offers more to him than my application: a descriptive analysis and a prescriptive analysis, allowing him to plan his training sessions, to follow up and adapt his program and sessions according to different variables.

I am certain that his watch would have advised me to wait until the sun set before going for a run!

CRM solution to increase sales productivity for manufacturing companies

How can predictive analysis complement a business intelligence strategy?

Like Internet fanatics and runners, it is important to note that in 2016, with advances in technology, the concepts of “machine learning” and democratization of software tools, many businesses plan to develop a strategy aiming to predict broad client trends and behaviour, so that they can adapt their messaging, offerings and tactics.

For many of us, business intelligence (BI) refers to dashboards, complex calculations, reports and key performance indicators. But what about prescriptive analysis (PA)? Besides the huge potential that the concept inspires, very few people can actually draw the line between a BI report and a PA report.

Business intelligence lingers on broad tendencies at an aggregated (macro) level and allows for a focus on the dimensions that interest us (for example, geography, products, clients, branches, campaigns, services, etc. …). “BI” offers a descriptive analysis of the past, of what we have already done.

Instead, predictive analysis pays attention to the lowest level of the detailed information and attempts to detect tendencies allowing for model creation with the goal of predicting future behaviours. “PA” offers a prescriptive analysis about what could happen, of what lies ahead.

The following chart provides an example of questions that can be answered with each strategy:

Questions BI claims to answer Questions PA claims to answer
How many prospects did the last campaign generate? How many prospects could be generated by the next campaign?
What was the purchase trend for product X over the last 12 months? What will the purchase trend be for product X over the next 12 months?
How much revenue did we generate for this product line in the last trimester? How much revenue will be generated for this product line during the next trimester?
What client segment uses the majority of the services we offer? Which client segment will most use the majority of our services in the future?
What was the last date Mr. X performed a transaction on our website? What is the probability that Mr. X will perform a transaction during the next month?

Having a good idea of the next product or service that clients will potentially use enables us to define, operationalize and measure our strategies and initiatives. Certain CRM solutions allow for the detection of tendencies and suggest offerings with a high likelihood of attracting our clients. These client trends can be easily identified and a specific offer could be sent their way, or a list of potential clients can be generated for a sales rep.

More sophisticated tools allow for the calculation of a propensity score of a client or client segment when faced with a particular offer. This score can then trigger mechanisms or business rules that automatically instigate a series of actions for this client or group of clients via a CRM solution (campaign, email, experience with a portal, social media messages, etc. …).

Leave a Reply