Your online store can bring in more profit. To do this, you don’t need to restructure business processes, change your approach to logistics, or spend millions on advertising. We’ll tell you how to approach audience analysis and hypothesis testing so that it brings measurable results: in visitors, baskets, orders, and the number of zeros in the average check.
Why analyze user behavior on a website
Several years ago, the international brand Marks & Spencer decided to update its DNA and carried out a large-scale redesign of the site. The update cost 150 million pounds. Naturally, all changes were made only after approval from the UX team and top managers of the company.
The website redesign was expected to increase the student data number of online orders by at least a third. But something went wrong. Users did not like the site: immediately after the launch of the updated site, the level of online sales dropped by 8%. The company was losing $10 billion of its standard revenue every month.
The reason is insufficient deep analysis of the audience. The brand's specialists, when developing the new version of the site, were guided by their vision and design trends, and not by user experience.
To avoid such losses, before implementing any changes to the site or application, you need to conduct an analysis of consumer behavior - collect and study data on customer actions, their preferences and purchases.
Audience analysis helps:
identify the most effective customer acquisition channels and optimize marketing costs;
find the site's weak points and improve user experience;
predict consumer behavior and identify bestsellers among products;
personalize communication with clients taking into account the characteristics of the target audience;
increase conversion into subscribers or buyers;
catch scammers or track competitors' attacks.
To achieve these goals, you need to understand your customers from all angles. Here are four essential tools for analyzing user behavior.
1. Cohort analysis
A cohort is a group of users united by some common characteristic.
In one cohort we will define everyone who made purchases on your site on 11.11 this year. In another cohort - those who also bought on 11.11, but last year.
Let's try to analyze these cohorts separately. For example, study what other common characteristics do those who bought goods on sale this year have. Maybe it turns out that most of them are people with similar demographic characteristics or come to the site from social networks?
Another way is to analyze two or more cohorts that are collected on a similar principle. Like ours: people in both groups buy goods on sale, but in different years. By comparing both user segments, we will understand:
whether the audience's interest in your sales grows or falls from year to year;
Do people who buy your products or services during the sale have anything in common?
new or already loyal users are more likely to buy on sale;
what is the average store receipt during sales;
what percentage of users make purchases on the site after a sale, and so on.
The conclusions from such research may be unexpected. For example, that a sale does not help attract a new audience, or that the average check during a sale is not higher, but even lower than usual. Or that users who were attracted to the site by discounts do not buy anything else from you.
In general, you are likely to find a problem. And that's good! There is a problem - there is a solution that will increase profits. This is where you will approach work with engagement, retention and other metrics that are important. The conclusions will appear by themselves.
It will turn out, for example, that it is better to show targeted advertising to a cohort of users who perform more targeted actions after switching from social networks, and to send email newsletters to those who respond well to letters. This is a reason to reconsider the structure of marketing expenses.