Research on website traffic and conversion optimisation in e-commerce

Abstract

This report provides an analysis and evaluation of AXA Belgium’s house and apartment insurance e-commerce channel with the purpose of identifying changes that could be done in order to optimize the conversion rates of visitors who go through the website. As AXA Belgium has set out to perform A/B tests, a method that allows the company to experiment with changes in the website and measure their performance with a low cost and risk, the project aims at systematically identifying which elements and pages present the best testing opportunities in terms of potential gains, as well as possible issues users may be facing. The approach to identifying hypotheses, which were then suggested as A/B tests, was divided into three segments: website and conversion data analysis, industry benchmarks and usability analysis. The first refers to studies of the data gathered from the website in order to identify differences between cohorts, bottlenecks in the conversion funnel and the possible reasons for these bottlenecks. The benchmarks were investigated by looking at the websites of competitors and AXA France in order to identify points of parity that AXA Belgium did not have, as well as looking at the flow and questions asked. The usability analysis looked at video session recordings and user interviews to identify points of friction in the website. Results of the analyses indicate that the way the price is displayed on AXA’s website, as well as the fact that some elements that affect the price are not evidenced, are a major point of frustration for users causing many to abandon the website without a purchase. It is also clear that there are too many steps and questions in the website, which is discouraging for many potential clients. Finally, there were significant differences in terms of conversion of certain cohorts based on nationality, language spoken and type of insurance they were seeking. The report finds that the flow must undergo some structural changes if the company wishes to see significant increases in return on investment. The main recommendations made are: to simplify the flow in terms of length and number of questions; to display the pricing in a pick-and-choose model; and to further investigate the reasons why cohorts converted more (foreign nationals) or less (Dutch speakers and house insurance-seekers) to address these targets strategically.