Sephora is one of the earliest retailers to offer a chatbot experience consistent with its chic and modern brand values. The Sephora chatbot is available on Kik, a messaging platform popular with millennials. The team at 1080bots is continually reviewing chatbots to figure out how to create brilliant high-touch ecommerce experiences that represent brands and deliver on business objectives.
Hi, Welcome to Sephora
The conversation begins with the chatbot introducing itself and what it’s able to do for the user. In this case, the chatbot provides makeup tips and reviews. The quiz is great because it draws the customer into the Sephora’s products and discovers what they might be interested in. The use of the available buttons and emoticons in Kik makes the communication interesting and the quiz ends with a short product video.
Looking For That Red Lipstick
The conversation context is lipstick and the video was ‘infrared rouge’. Therefore we wanted to find out about the range of red lipsticks in particular. We interrupted the button-driven dialog by typing the term ‘red lips’. However, the chatbot admitted to not understanding the term and offered up 3 button options. We selected the ‘Product Search’ button, but this seems like an extra and unnecessary step to get us to a search function.
We used the identical search term as we’d used previously: ‘Red Lips’ which produced a set of results. Clicking on one of the results provided a link to the product page on the mobile site. this was a success, however, it took longer to arrive at the result than it needed to.
Dusky Green and Dirty Blonde
Having had some success with a color, we figured that Sephora as a cosmetics retailer should be really good at colors. We entered the term “Dusky Green” (again interrupting the button dialog) but this caused the chatbot to fail. It’s best practice if a chatbot provides a text input opportunity that the bot should try to recognize it. Buttons should only exist to help a customer, not restrict them.
Once again, clicking the Product Search button, we continue the adjective-color combination and try ‘Dirty Blonde’. This failed as well, which is disappointing as ‘blonde’ should be a term that an in-store associate would understand. Next, we try ‘Dirty Blonde Hair’ which delivers a result – and provides a clue that the Sephora chatbot is not using much Natural Language technology but matching on keywords.
Learnings from Sephora
Bindu Shah, Vice President of digital marketing at Sephora, explains: Think Tank: What Sephora Learned From Building a Chatbot With Kik
- Bots are not that complicated (at least, they shouldn’t be)
- They’re a different kind of social
- They thrive on experimentation
“Clearly, there’s no “playbook” for [chat]bots just yet, but we’re hopeful that by experimenting, learning and iterating, we will continue to uncover the potential in chat and inspire the rest of the industry while doing so.”
Interested in how we can help you build a world-class chatbot? Contact us at firstname.lastname@example.org, or fill out the form below and we’ll get right back to you.