Data and Optimization are T.O by Lipton's Cup of Tea
How the loose-leaf tea machine brand uses consumer data and AI to drive optimization and conversion
In marketing it's good to have a strong intuition. But for Mathieu Bernard, intuition only goes so far. The head of e-commerce for Unilever brand T.O by Lipton says he backs up his intuition by looking at the data.
“We have a lot of data to process coming from thousands of consumers,” Bernard states in the video above. “So, the challenge is to put the focus on data we trust but also on data we can leverage and transform into opportunity.”
Bernard's main mission is to acquire new T.O customers and ultimately retain them. However, it can be difficult to decipher why one customer bought a tea machine or tea capsule and why one did not. He wanted to find a turnkey solution that would allow him to gather analytics quickly. So a few years ago, he and his team implemented UX analytics solution ContentSquare.
Diving into the data
T.O started working with ContentSquare in 2015 before the brand even launched. The company inserted a line of code on its website and then used cookies to record, compute, and analyze customers' data, like mouse movements and clicks. Before its official debut, the brand invited about 500 people to shop its website, Bernard says, and then used ContentSquare to gather insights and optimize its website before launch.
Since then, T.O has continued to use the solution for analytics and A/B testing. For instance, Bernard says that the company ran an A/B test on its tea machine's product page to optimize conversion. For the test, Bernard created about 15 different product page layouts and altered the placement of ratings and reviews in each one. He discovered that when the ratings and reviews were placed near the top of the page and were more visible they generated higher engagement and higher conversion rates.
Patricia Césaire, senior solution consultant for ContentSquare, says that the brand also used the vendor to examine consumers' web interactions and measure the efficiency of zones or elements on T.O's web pages. This helped T.O pinpoint which areas led to decreased engagement and performance. Based on this behavioral data, Césaire says, T.O identified two customer segments: returning customers and new visitors. The brand's digital team then isolated each segment's customer journey and saw that the returning customers were showing "back-and-forth browsing activity," she notes, which suggested confusion. After reviewing metrics like hesitation rate on its various page zones, she adds, T.O pinpointed which areas were causing this confusion and ended up streamlining the purchase process, such as by including related purchase items on its pages with distinct calls-to-actions to help consumers avoid cross-referencing multiple pages.
Learning more through machine learning
This year T.O started leveraging ContentSquare's latest development: an AI-powered bot named Arti. By taking ContentSquare's analyzed data, machine learning, and customer feedback, Arti can make recommendations to various members of the T.O team for how they can optimize the brand's customer journey or content. So, T.O's CRM manager might receive recommendations on funnel optimization while an acquisition manager might receive recommendations on how to drive Web conversion, Bernard explains. It can also notify different team members of changes in metric performance.
“This is really helping us prioritize our actions,” Bernard told DMN.
Since implementing ContentSquare, T.O has increased its conversion rate by “more than 100%,” Bernard says—105% by the vendor's count. ContentSquare also reports that Arti has helped the T.O team save time by up to 30%.
In addition to these measurable wins, Bernard says implementing the solution has helped encourage this test-and-learn mentality—even when the tests aren't always successful.
“Failure is part of the progress,” he says.
Still, Bernard isn't completely satisfied. When asked about where he'd like to improve, his reply was simple: “Being faster” and continuing to test and learn. As he put it, “It's very important when you do analytics on anything to work at a quick pace.”