Integrating analytics and research
When I Googled Customer Insights Analyst, LinkedIn presented me with 3,000+ Canadian job openings across a multitude of industries. Out of curiosity, I randomly picked five job descriptions to read. Unsurprisingly, the hiring managers turned out to be looking for several different skillsets, from data management to reporting to analytical model development to survey questionnaire design. What a one-size-fits-all job title!
One commonality across all these postings is, apparently, that these analysts are expected to bring businesses closer to what their customers are thinking and how they will be behaving. Behind the broad diversity of talent in need lies one trendy fact: data analytics and customer research need to work together to reveal the true insights and to connect the dots between what customers say and what they do.
Machine learning, for example, significantly magnifies a company’s capability to predict customer churn and allows early intervention to reverse it. However, without the deep knowledge derived from research, particularly of a qualitative nature, data itself is not going to tell you about the root causes that drive customer disengagement.
That is why some companies have conducted early experiments to bring their data scientists and analysts under the same roof as market researchers and insights specialists. By doing that, their expectation is to get the best out of the two worlds. Our experiences, however, tell a slightly different story. Employees of these two groups think differently and they work at different paces. Many organizations are not set up to accommodate the processes and structures of these two groups within one team.
More importantly, not every analytical solution needs research insights, and vice versa. A more effective way to integrate the two is to let a third stakeholder lead the business initiative. Only when the business lead clearly defines a problem, could the most powerful insights be pulled out of both sides to address the challenge. Simply put, it is better to merge on specific projects than to merge the teams.
One of the best practices is direct data integration. When a firm has abundant first party customer data, it is most valuable to perform analytics as well as generate research insights from the same customer dataset. In one example we saw, the customer satisfaction research indicated that the majority of complaints were around pricing transparency. Before making changes to pricing, data analysts raised a different point. They used the first party customer data to assess the lifetime value of those less satisfied customers and realized that very few high value customers gave the company a suboptimal rating because of price. Instead, their top irritant was the lack of knowledgeable servicing staff. Seeing both aspects helped the company prioritize their customer experience initiatives.
Direct data integration of this kind is also changing the way the traditional market research firms recruit research participants. Sampling based on targeting demographics is still practical, but could now be enhanced by look-alike modeling based on first party data.
Financial services and telecom are among the early adopters of this integrated insights generation approach because they have farmed much richer first party data than other industries. To make it work, a number of retail and CPG companies developed their in-house loyalty programs in the past few years. In exchange for points and rewards, customers have voluntarily shared personal data and allowed loyalty program operators to collect their transaction history. Sooner than anyone could predict, enough such first party data will be accumulated to enable integrated insights.
In other industries where direct data integration may be out of reach, cross referencing analytics and research outcomes is still feasible. By bringing them together into a common storyline, data points from both sources could be validated against each other, and insights could be delivered with a more holistic and wholesome account of facts.
When it comes to putting insights to work, real-time pilots and bite-size enhancements are always better than boiling the ocean. Insights, from either analytics or research or both, need to be tested in real life before being fully baked into your long-term strategy.
Authors
Edward Li
Director, Business Banking, CIBC
Richard Nestor
AVP, Marketing Analytics and Insights, TD Bank Group
