Sephora VUI

Virtual Shopping Assistant Concept Design

Individual Project
Time: April 2018 - July 2018

What Problems Did I Try To Solve At Sephora?

Context Definition

(Online Image)

– “The aisles between the product shelves are too narrow. Especially when several salespersons and customers are walking around, it can get very crowded in the Sephora stores and difficult for customers to try the products. Sometimes the customers have to go back and forth to let other people walk through and our shopping experience are interrupted.”

– “I am not the chatty and social type of customer. I are not really looking for anything specific and just want to enjoy my little adventure in Sephora. But these enthusiastic salespersons would always “kindly” offer help in the wrong moment. I wish it could be a private shopping experience for me after work and save me the unnecessary social time during my shop.”

– “I just turned 16 and not good at doing makeup, I feel too shy to walk in Sephora and try the products in front of everybody.”

(data extracted from Google reviews and direct quotes from interviewees)

The Challenge

- How to provide a more private shopping space and a more comfortable shopping experience for the Sephora customers?

Taking the weight management app as an example, people tend to trust a virtual chatbot more than a real person to expose the data of their body and admit the mistakes they make while trying to lose weight.

If during the shop at Sephora, all the customers need to do is to interact with one chatbot and accomplish all the tasks at once, they could avoid the unnecessary social moments with the real sales people, feel more confident to walk in the Sephora stores and finish the shop much quicker and easier.

Keywords: private space, quick and easy.

My Design Process

What Have I Discovered From Research?

research

Choice-Based Conjoint Analysis

conjoint
conjoint

According to this conjoint analysis, the most important attributes are makeup preview feature, followed by Voice and then popular product feed. The speaking tone and the past shopping reminder don’t make a huge difference in customers preferences.

By reading the utility level among each attribute, I first learned that the Sephora’s customers strongly prefer the makeup preview feature. Among the four levels in voice, I found that customers are generally more happy with a young voice, surprisingly! I personally assumed that customers would prefer a female voice regardless the age. But my research shows that the age factor outweigh the gender factor. Among the different types of popular product feed, the customers care the most about popular products within the same age group and then the same price range from their past shopping experience. They are not interested in the popular products in the same location they are in (perhaps people don’t like to look the same as the others around them :)) Besides, among the speaking tone, they have a slight preference on the casual speaking over the professional speaking. Lastly, they slightly prefer not knowing their past purchased items during their shop.

Therefore, among the 126 profiles of the virtual shopping assistant in Sephora. The best virtual shopping assistant would look like this: a voice from a 18 – 28 years old female with a casual speaking tone. She will introduce you popular products based on your age group and send you makeup previews to your phone for you to choose your favorite products.

What Type of Bot Is It?

According to the Designing Bots: Creating Conversational Experiences by Shevat, Amir, there are different types of bots for different purposes. I defined the type of this virtual shopping assistant, Sela, based on the 2nd chapter of this book.

bot

Each customer uses Sela and interacts with it one by one

Sela serves each customer as a personal shopping assistant.

Sela provides a single service and represents Sephora only

Customers can connect to the Sela voice bot and view makeup preview features via the Sephora app

What Do The Customers Look Like?

persona

What Are The Use Cases?

Storyboard

I first created a storyboard to help me understand the shopping process and many different possible scenarios.

storyboard

The Online Reviews

In order to understandthe customers’ pain points and emotions during each touch point of the shopping journey, I crawled the 157 reviews on consumeraffairs.com and selected Google reviews. By analyzing the most frequent words and reading the contexts, I found that some of the big issues during the customers’ shopping experience are about the in-store service, the refund process, the samples and the shipping service.

GitHub link: https://github.com/MelanieZeng/Sephora-review-analysis

The Shopping Journey

After analyzing the issues, I created this shopping journey map to help me better understand what exactly Sela should do to help improve the good shopping experience which is currently not delivered by the real sales assistants.

Shopping journey map

How Would The Conversations With Sela Look Like?

Sela would be a chatbot that connects with the Sehora app. Her voice is from a 24-25 years old woman and her tone is mature, gentle and knowledgeable. She would provide you recommendations when needed, such as popular cosmetics within your age group,  products within your budget and tutorials on how to use the products. If you worry about it would take too much time to find the perfect cosmetic for you, she would kindly send you previews of many different products on you. Lastly she will help you to pay for it or order products online easily.

Features: product navigation, product recommendations, product preview, payment processor, online order

Technologies: GPS, Natural Language Processing, Voice Generation, Face Recognition, Data Analytics.

According to the shopping journey map above, I selected three scenarios to design conversations for Sela. My goal was to improve the customers’ shopping experience by using Sela instead of the real shopping assistants. I used Google Dialogflow to design these conversation scripts and train the chatbot.

Chabot VS Real Sales People

- How would the chatbot actually improve the shopping experience?

1.Pain point: try makeup sample

It’s a pain point for both the customers and the sales people. The customers don’t always enjoy the experience of trying the makeup because they may not find the best fit the first time and will have to keep trying many times. The sales people may not always maintain patient and provide satisfying service during these many attempts. On the other hand, sometimes the sales people simply can’t allocate enough time on serving only one customer. They may end up spending too much time on one customer but still don’t accomplish the sales and lose other potential customers.

Prisoner’s dilemma

However, the makeup preview feature broke the Prisoner’s dilemma. It can not only save the sales people’s time but also provide unlimited times of different previews for the customers in an easy and private way.

2. Rich, precise and customized information

Based on the data Sephora can collect from its large base of customers, Sela can always provide customized recommendations that fits the customers’ demographics, budget and personal condition (skin type, skin tone, eye color, hair color, etc.).

3. Express Order

Similar with the express order on Amazon, the customers can simply verify information with Sela and order products online. The tracking information will be placed in the customers’ sephora account. It can help avoid the human mistakes of making the wrong order. The customers can also avoid hopping around between the sales people and the cashier. They can enjoy a private shopping experience through Sela and avoid unnecessary social moments if they wish.

All in all, the key is not to guide the customers do the way Sephora does and follow the procedure sephora built for every customer, it is to understand every customer’s different intent and help them accomplish what they want in the easiest and most comfortable way.