When users sign up for Le Tote's fashion rental subscription service, they are asked to answer a series of questions related to their fit and style preferences. These answers inform the algorithm that populates the shipment of items, a.k.a. "tote" users receive after signing up. The algorithm also uses this data to surface more product recommendations post-checkout.
PLATFORMS: MOBILE WEB, Desktop
In 2016 I led the redesign of the onboarding for Le Tote's rental subscription service on mobile web and desktop. My responsibilities included user research, concept ideation, aligning key stakeholders on product goals, designing user flows, visual design, prototyping, user testing, incorporating user feedback into design iterations, and quality assurance.
Defining The Problem
We conducted interviews with 16 women who had never used Le Tote before, and found that most participants were skeptical about personalization. The value proposition of a styling algorithm remained unclear to them when they walked through the onboarding flow. The current way style info is asked for in onboarding is too subjective to provide results that actually reflect a user's preferences. In order for the algorithm to populate a user's tote with items she actually likes, we need a less subjective way to ascertain an individual's style.
1. Increase customer acquisition by converting more visitors into subscribers
2. Collect better data in order to provide users with better personalization
2. Clarify value propositions and how the service works
3. Create a framework that can support future A/B testing and optimization
1. The design should be modular and flexible - this includes messaging, imagery, and overall flow.
2. Balance educating the user, the number of steps in the flow, and the amount of data collected with the ease of getting through the flow quickly.
Playing A Guessing Game
The current style quiz engages users in a Tinder-like game of swiping through put together outfits. While fun to interact with, the quiz itself shows products (such as shoes) that aren't actually available through Le Tote's subscription service, which is misleading. Additionally, from a data perspective it's difficult to parse what a user is responding to when they like or dislike an outfit.
Easier To Interpret Answers
The updated version of the style quiz focuses on building a personalized style profile for each user based on product attributes and subcategories she selects. These attributes and subcategories reflect what is actually available in Le Tote's inventory and can be updated seasonally. By asking her to define her style with these answers, we can also start to collect data about what appeals to both our existing users and potential subscribers who go through the onboarding flow, and inform our merchandise purchasing strategy.
When rental inventory circulates between all users in the Le Tote ecosystem, availability of a particular item is not always guaranteed. This poses a unique challenge when presenting the inventory to users and asking them to respond to it. The updated style quiz also aims to give users a sense of what is generally available in Le Tote's inventory, while minimizing fluctuating impressions.
AN Interesting INSIGHT
We found through testing different types of product shots (professionally styled outfits, full body photos, lifestyle photos, and product laydowns) that users responded the most positively to cropped photos where the model's face was not shown.
We saw a 61% increase in conversion when we released this update.
A Whole New World
Collecting style preferences based on attributes and subcategories provides us with the ability to surface recommendations to users in new and interesting ways. I am currently exploring different touchpoints where we can make smart recommendations and provide increasing value to Le Tote rental subscribers.