5 Keys to Creating an A1 Analytics Culture

The Big Idea

The Admiral Group had grown to become one of Britain’s biggest insurance companies with its successful direct-to-consumer business model that served customers online and via phone. In 2009, Admiral entered the U.S. with its Elephant Insurance brand. 

The U.S. market showed huge growth potential, as consumers could only comparison shop by researching multiple insurers, filling out multiple forms and enduring long wait times for quotes. The inefficient process meant only 20 percent of U.S. drivers were comparison-shopping for auto insurance, compared with 90 percent of U.S. consumers who compared prices for airline tickets.1  

In 2013, Admiral launched its U.S. insurance comparison site, Compare (then called comparenow.com), with the mission to make finding the best price easier and quicker. 

The Scenario

Because Compare was new to advertising and did not know which messages would best resonate with U.S. consumers, it tested 12 different TV campaigns across six Californian markets in 2014, keeping ad buys consistent. Embracing Admiral’s “deep test-and-learn” culture, Compare analyzed web traffic before and after ads ran to identify the most effective campaigns. It found that the best performing ads improved sales in even the worst performing markets. By 2016, Compare had grown to 50 employees, offered policies in 48 states and partnered with nearly 100 U.S. insurance brands. 

However, TV advertising was costly, so Compare also used Google AdWords and broker websites to acquire potential customers. Visitors to broker websites typically found them through search terms, and once they clicked on the broker’s link and answered basic demographic questions, those with the same characteristics of Compare’s highest converting consumer groups were connected with the company. 

Though the broker channel was inexpensive, completion rates were low, likely because users had to provide the same personal information twice — on the broker’s website and Compare’s — in order to receive quotes. There was also a lot of traffic coming to Compare’s website from mobile devices. Compare’s completion rate sank from 18 percent to 12 percent between February and March 2016. 

The Resolution

In its quest to optimize the customer’s purchase funnel, Compare stayed true to its testing and learning culture and conducted field experiments. 

Starting early in the funnel, Compare looked at its email marketing, designed to drive consumers to the site. The company used “before and after” and A/B testing to optimize the language, subject lines and cadence of its emails. The next step was to change and test the site itself, with the goal that visitors would complete a questionnaire to receive an insurance quote; completion rates were directly correlated to higher purchase rates, which then would be reflected in Compare’s total revenue from its insurance provider partners. 

On the site, the company chose to test a banner ad with an estimate of the premium that drivers could receive, in order to motivate them to submit the questionnaire. The banners were tested using the average vs. lowest potential premium; using the lowest might motivate form completion, but what if drivers ended up with a higher than estimated premium — would they be too disappointed to purchase? 

The banner was placed on the page where most consumers were dropping off: the “vehicle page,” which was the first step of the online form, as customers preferred to provide vehicle information before personal information. This was one of the strategies Compare was already using to improve completion rates, in addition to prefilling information for consumers who had already offered it on broker websites and offering to email customers the questionnaire. These processes were tracked in the organization’s analytics canvas, as the test would need to improve completion rates beyond the efforts already in place. 

The experiment increased quote completion and click-through rates by 4 percent and 6 percent, respectively, when the banner ads showed consumers the lowest quote, rather than the average. 

The Lesson

Compare’s adaptability was enabled in large part by an agile development process, in which regular test sprints were conducted and team leaders met briefly every day. Employees at all levels of the business were encouraged to suggest ideas that were tested as the website was tweaked. The atmosphere of experimentation accepted failure, which is critical for firms to implement field experiments — and learn from them. 

While an exceptional customer analytics operation is built on data, firms that thrive share five culture enablers:

1. Truth Seeking: These companies value truth and ruthlessly seek it out. Their recommendations are supported with facts. Leaders are not afraid to tell a junior employee that they win because their facts and analysis prove it.
2. Analytical Embeddedness: Analytical talent is placed at every stratum of the business. Doing so prioritizes analysis and spreads belief in its importance throughout the organization. With experts on every team, immediate transformation occurs. 
3. Methodological Consistency: Infusing an organization with analytical talent is a challenge, and so too is ensuring a consistent approach in methodology. The best firms create panels of business leaders who review methodologies and ensure consistent application across the organization, which also helps nonanalytical employees learn quickly from the system.
4. Training Through Merchandising: The best firms use presentations and reports to both share project progress and drive alignment of teams. One large retailer, for example, has an “open house” every two weeks, at which teams present their objectives, measurable progress and future plans. The sessions create a culture centered on using analytics to move projects forward.
5. Hiring Consistency: Companies with excellent analytical cultures usually ensure all employees meet a minimum threshold for understanding and leveraging data to ensure analytical decision-making. Procter & Gamble, for example, delivers a test to prospective employees to test their ability to use data to reach reasonable conclusions. And Compare makes employees pass a portion of the GMAT. 

Rajkumar Venkatesan authored the case series Tackling Low Completion Rates — A Compare.com Conundrum (Darden Business Publishing) with Senior Case Writer Jenny Craddock and alumnus Kyle Brodie (MBA ’17). 

This article is based on Rajkumar Venkatesan and Kimberly A. Whitler’s “Developing a Customer Analytics Culture to Drive Causal Inferences: Insights From Field Experiments at Compare.com.” 
 

• 1Joe Mahoney, “Comparenow Trying to Change the Way Consumers Buy Auto Insurance,” Richmond Times-Dispatch, 12 May 2014, https://www.richmond.com/article_55420f9a-d7e8-11e3-b2f8-10604b9f6eda.html