Beauty Marketers Are Getting Personal with AI

Product recommendations are becoming smarter

Author: Krista Garcia

July 3, 2018

Beauty marketers have embraced an array of artificial intelligence (AI) applications like using augmented reality (AR) to virtually try on lipstick or relying on chatbots for guided selling and to power personalized product recommendations.  

With so many brands, products, formulations and shades, it's not surprising that 70% of US female beauty buyers said they were overwhelmed by product choices in a May 2018 survey by conversational marketing firm Automat.  

One answer to the overwhelming sea of choices could be virtual beauty advisors like Sephora’s Virtual Artist, which allows users to upload photos in Messenger to "try on" bot-recommended user-selected shades of lipstick, and Madison Reed's bot Madi, which suggests hair colors based on user uploaded pics. 

Nearly half of beauty buyers in the study said they would be likely or very likely to use a virtual beauty advisor when shopping for products online or offline.

Roughly one-third were unsure, which isn’t unreasonable since the concept isn’t widespread. When asked, though, 32% had heard of virtual beauty advisors, a higher recognition than other beauty tech like virtual try-ons (28%) and in-store skin scanners (22%).

Just a few years ago, there was excitement around using AI-powered chatbots in ecommerce. That hype has died down a bit, but the beauty industry is still experimenting with the technology. According to an April 2018 Celebrity Intelligence survey, a majority of beauty marketers worldwide believe technology like AI-powered personalization across email and social (65%) and AR for makeup discovery (58%) will be adopted at scale in the next two years

Malik Abu-Ghazaleh, vice president of digital marketing and ecommerce at Lancôme, said he was exploring conversational commerce, but it didn't seem right as a near-term engagement solution for the brand. "It's more about things working in the background and suddenly transforming your experience on the site as a result of the AI that we're able to infuse," he said.

More specifically, Lancôme is testing AI to deepen its site personalization beyond the typical, "customers that bought A, also bought B" approach. The company launched a beta version in October 2017 that uses a combination of shopper behavior and user-provided details to customize the experience. 

For instance, Lancôme discovered that skincare shoppers and makeup shoppers are two different segments: It can hypothetically show a set of products, videos or articles for eye creams but not mascaras. The brand is also gathering data from user profiles, so it can display foundation in a range of shades close to a shopper's skin tone instead of showing the full range of light to dark. Lancôme advertises 10 foundations in 185 shades, so this is one solution to the overwhelmed-with-choice conundrum. 

Now if a shopper picks a lipstick shade, they will get automated recommendations for an entire look instead of suggestions based on human-dictated rules. "I was a little afraid of what combinations of colors were going to be put on people in the beginning," Abu-Ghazaleh said.  "But it all made sense. There was nothing strange about it."

This feature hasn't been publicized because it's still in testing, but so far around 20,000 profiles have been filled out and users who have completed the profile have converted at 3 times the rate of those who didn't.

It's also still a work in progress. The current profile is long and asks too many questions, so the goal is to make it more fun and snackable by letting users answer a few questions at a time. Ideally, users will spend more time on the site and ultimately buy there. "The dollars might not be the first thing that comes, but if  you're engaging with the site more and going deeper, that's a win," Abu-Ghazaleh said. Perhaps not surprisingly, "well over 50%" of Lancôme's sales come from loyalty members, so engagement can pay off.

According to Automat, the shoppers that are most interested in virtual beauty advisors are also a brand's best customers. Half purchase beauty products two or more times per month vs. 40% overall, and roughly one-third purchase beauty products two or more times per week vs. 22% overall.