Get started with Business Analytics: A brief guide to setting things in motion

Orientation for beginners

If you're new to the world of business analytics, we'll you're not alone.  With data driven decisions, business analytics and data analytics, have converged to some extent. There are specific roles like analysts and data scientists, but hybrid roles are becoming the norm. This implies the ability to frame questions, interpret outputs, and drive decisions is increasingly distributed, rather than confined to a single specialist role.

A practical takeaway is that the entry point is not a job title. It is the ability to ask the right analytics question and connect that question to the business decision that needs to be made.

A customer support example

Here's an example of how each can be applied in a Customer Support business function.

Customer Support is a useful place to apply the model because it naturally includes customer feedback, business representative data, time of call-in, issue resolution and time to resolution. Those elements make it easier to see how different analytics styles can use different inputs and produce different kinds of outputs, while staying anchored to one function.

The four analytics styles

There are 4 types of analytics. The two "D's - Descriptive and Diagnostic", are more backward looking, while the two "P's - Predictive and Prescriptive are forward looking".

This split provides a simple way to set expectations: backward looking analytics is focused on learning from what already happened, while forward looking analytics is focused on anticipating what may happen and determining what should be improved.

The four questions

Define one analytics question corresponding to each of the four styles of analytics: descriptive, diagnostic, predictive, and prescriptive. 

• Descriptive: How can historical data be used to glean insights about type of customers who connect with the firm?
• Diagnostic: How can customer feedback and business representative data be combined to create relationships between time of call-in, issue resolution and time to resolution?
• Predictive: How to take external data factors such as education, sex, salary, geographical…etc, and create models that can provide insights into product quality and logistical bottle necks?
• Prescriptive: How to feed self-learning algorithms, that can ingest multiple data variables and suggest product and service improvements?

These four questions can be used as a repeatable template. The purpose is to keep analytics tied to decision-making by forcing clarity about what is being asked, which style the question belongs to, and what kind of output is expected.

A step-by-step way to apply the model

Feel Free to apply this as a model to any business function. A consistent way to do that is to keep the structure stable and only swap the business function context.

Step 1: Start backward looking with descriptive analytics

Descriptive: How can historical data be used to glean insights about type of customers who connect with the firm?

This establishes the baseline by using historical data to glean insights about the type of customers who connect with the firm. It keeps the effort focused on understanding who is interacting with the business, based on what has already occurred.

Step 2: Move to diagnostic analytics to understand relationships

Diagnostic: How can customer feedback and business representative data be combined to create relationships between time of call-in, issue resolution and time to resolution?

This moves from “what happened” to relationships that can be examined by combining customer feedback and business representative data. The structure of the question also keeps attention on time of call-in, issue resolution and time to resolution, which creates a clear frame for what relationships the analysis is trying to create.

Step 3: Use predictive analytics to look forward with models

Predictive: How to take external data factors such as education, sex, salary, geographical…etc, and create models that can provide insights into product quality and logistical bottle necks?

This introduces external data factors and calls for creating models, with the goal of providing insights into product quality and logistical bottle necks. In this model, predictive analytics is forward looking, but still positioned as an insights-producing step that informs what comes next.

Step 4: Use prescriptive analytics to suggest improvements

Prescriptive: How to feed self-learning algorithms, that can ingest multiple data variables and suggest product and service improvements?

This frames prescriptive analytics as forward looking and recommendation-oriented, because it aims to suggest product and service improvements. It also implies pulling multiple data variables together into one system that can ingest them and produce suggested improvements.

Choosing what matters most

Which of these questions you conceived do you think is the most important to answer, and why? Why is this question more important than the other three questions? How would answering it help drive business value?

I believe that Prescriptive analytics is the most important, as it can leverage excerpts of relevant data from the other three styles of analytics and suggest insights/recommendations for future improvements.

This position makes prescriptive analytics the culmination point: descriptive, diagnostic, and predictive outputs become inputs that prescriptive analytics can leverage. In this model, the prescriptive layer is where insights are turned into recommendations for future improvements.

Business value and time to insights

Time to Insights about consumer behavior has become a key differentiator for a firm, as the consumer pallets are constantly changing.  Prescriptive analytics can take data from sales, operations, marketing…etc and create a persona based approach.  These personas can allow for interactions where consumers feel a personalized approach.  The personas can be constantly updated through continuous analytics, thereby ensuring consistency and alignment with consumer expectations.

This sequence ties differentiation to time to insights, then ties time to insights to prescriptive analytics. It then connects prescriptive analytics to a persona based approach that can be continuously updated through continuous analytics, with the stated goal of consistency and alignment with consumer expectations.

The connected consumer and narrative control

As a connected consumer, the power has shifted from producers to consumers.  The Prosumers, through their direct and indirect channels can impact the business and its brand.  Controlling the narrative therefore has become extremely integral in the age of consumer influencers.  By leveraging a persona-based approach, which predictive analytics can help create and constantly update, businesses can control and influence the narrative.  Consumers can therefore become brand ambassadors, driving value and growth for the business. 

This logic connects a changing power dynamic to the importance of controlling the narrative. It also links a persona-based approach to predictive analytics by stating that predictive analytics can help create and constantly update personas, which then supports the goal of controlling and influencing the narrative.

Apply, review, refine, and repeat.