Test Quickly, Fail Fast: Utilizing Data to Retain Customers

In order to grow and maintain a business, regardless of the industry, it is important to be able to properly identify two types of customers –those who are happy and those who you are at risk of losing. Analytics can play a key role in helping identify and segment these customers to initiate retention programs proactively. Here at Digital Insight, we’ve found that one of the most effective ways of using analytics involves brainstorming and testing hypotheses to build solutions backed by data.

Map out your hypotheses

As analytics professionals, we sometimes jump to the “solve” before visualizing and mapping out the analysis. However, it’s important to think through several hypotheses and then determine which data sets are needed to answer each. To come up with these hypotheses, reach far and wide across the organization to brainstorm different ideas, variables and scenarios that could be contributing to the problem. Then, rank the hypotheses using a 2×2 across appropriate dimensions like difficulty to answer or analyze, and revenue impact if the hypothesis is found to be true. Doing so will help you focus your analysis by tackling the hypothesis with the greatest impact first and shortening the overall time to solve.

Put your hypotheses to the test

After your team has brainstormed several hypotheses, it’s important to find ways to identify the appropriate analysis for each. Think about what data you already have and what data points need to be mined or collected. Some may be answered through existing reporting or dashboards. Others will require deeper analytics with end user data. Use data sampling for analysis vs. large datasets.  Sampling is an option that can dramatically speed analysis time with reliable results and allows you to test quickly, fail fast, and pivot to the next hypothesis.

Analyze your results and put them to use

After spending time putting your hypotheses to the test, be open to the discoveries you may find during the analysis phase. Many times there is much to learn from the surprises uncovered during analysis that take you down a path you hadn’t considered. In addition, what you learn may impact tangential problems yet to be solved and can be leveraged later if well documented. When effectively analyzing customer data, businesses are able to identify, test, and validate leading indicators of clients at risk, predict high-risk clients before it’s too late and develop “treatments” to test for the most successful way to re-engage with the customer.

At the end of the day, acquiring a new customer can be 4-10 times the price of retaining an existing customer. Building a solid analytic approach to identifying and predicting at-risk customers will result in direct impact to the company’s bottom line.


Brenda Shimmons holds an MBA in Marketing and is a Digital Insight Innovation Catalyst. She has over 15 years of analytics leadership experience within financial services working with such industry leaders as GE Capital, Wachovia Bank (Wells Fargo), SunTrust, Scotiabank, and a portfolio of over 1,200 community banks and credit unions. At Digital Insight she leads the Analytics Team supporting multiple functions of the business. The Digital Insight Solutions organization leverages the team during product development where they drill deep into customer behavior to inform future design and improve user experience. She consults with Digital Insight Marketing, developing strategy and defining segmentation and hypothesis testing. She constructs deep dive analysis linking product adoption with retention and improved profitability for our clients growing their most profitable segments. 

Written by BrendaShimmons