Predicting Customer Behaviors through Data and Analytics

Authors: Bryan Copeland

Most organizations are on the lookout for ways to operate more efficiently and improve their customer insights through data. While the phrase predictive analytics is popular to throw around in meetings, we have found few companies have a process to implement this popular concept. By leveraging data and effectively applying predictive analytics, companies can use statistical analysis to proactively make data-backed decisions that drive future business initiatives, reduce customer churn and improve customer value.

Start by Understanding the Problem

Framing the business problem is the first step for a successful project. We’ve all witnessed the deployment of automated models that fail to impress because they are not aligned with business goals. Stephen Hawking once said “A super intelligent AI will be extremely good at accomplishing its goals, and if those goals aren’t aligned with ours, we’re in trouble.” While trouble in this case would refer to a waste of staffing resources, time, and capital investment, project alignment to business goals is still mission critical. By framing the design of the model around a business’s goal, the result of the predictive model can effectively drive top and bottom line growth.

Crawl, Walk, Run 

If predictive analytics is the goal, effectively sourcing and preparing your data should be the foundation. The internal and external data will be prepared, organized and cleansed in a way that will be able to feed your data model. The data foundation will determine the outcome of your analysis.

We know there are different levels of data maturity for implementation, so let’s explore what some of these phases could look like.

1. Crawl: In the crawl phase, your customer data is just getting off the ground. You likely have a Customer Relationship Management (“CRM”) system in place and have out of the box reports to start forming a view of your customers. Most of the data you have is historical or reactive with events and behaviors that have happened in the past. The good news is you have started to know your customers better through data. There is a foundation that is beginning to form and the data is usable. You are now in a position to start moving faster.
2. Walk: You’re building up your data strength now. Now that a foundation has been set and you know what it feels like to move, your data is “walking” to become insights. There is a single view of the customer that has been developed. In this phase, perhaps you’re understanding customer behavior and linking it to key operational metrics. Data relationships, customer preferences and product views all link nicely to a single customer. It feels like the data is ready for a race, maybe even at a place where it can be in front of your customers.
3. Run: Once the business objective of the campaign is clear and there is a “running” understanding of customer data, your predictive analytics campaign take off. We start by merging data sources so it can act as an analytical record to perform data science on. By standardizing the data set, the data can easily be consumed by the model. After viewing the standardized data, we then pick which statistical algorithm to implement.

There are many different statistical options to choose from when performing predictive analytics and the best option will depend on the shape of your data. Once the model is built, it can be further refined by performing feature engineering. Finally, the output can be used to create business value and redefine what’s possible.

The Benefits of Predictive Analytics

You can leverage predictive customer analytics to drive value in many areas of your business:

• Reduce customer turnover by identifying ‘at risk’ customers
• Accurately predict the value of customer orders to lead to more accurate quarterly forecasts
• Improve efficiency with targeted and effective customer interactions

A successful predictive analytics campaign can have a lasting and sustainable impact on your business. It’s always fun to run with data, but make sure you understand your business problem first. Alignment is key. Especially when data is helping you solve the problem.

Bryan Copeland is a senior consultants at RevGen Partners. He is passionate about helping clients connect the dots through data and building customer-centric cultures.