The Virtual Interior Decorator
About the company
Hutch was founded on a simple idea: allow anyone to discover their inner interior designer. Whether you’re looking to design your entire home or select finishing touches, we put all of the tools in your hands to try out new styles, customize looks and visualize everything in your own space before you buy.
Roles & Responsibilities
I joined Hutch as a co-founder before raising our first seed round in 2015. Since then, the company has rasied over $17M from the most notable investors and companies including Founders Fund and Zillow.
At Hutch, I am the lead product designer and I'm responsible for taking every feature from concept to reality. Usually this starts with tons of research that mainly consists of conducting user tests/interviews and looking at product data. After some conclusions are made, we prioritize a feature list into a design sprint. I then take that list and rapidly generate mockups/prototypes with different options for each feature to show the product team. It's important to note that because Hutch is supported by three platforms: iOS, Web, and Android, each view needs to be designed differently given the context and screen real estate for each platform. As screens get approved over multiple revisions, I'm in charge of making sure each final design is pixel-perfect and ready to be handed-off to the engineering team. This means all assets are properly formated and each screen is uploaded to Invision.
Sketch, Principle, Photoshop, Mixpanel, Invision, Trello, Lookback, Gliffy, Excel
Democratizing Interior Design
The first version for Hutch was based on a very simple idea, send in a photo of your space and style it with "Instagram-like" filters. As shown below, I first prototyped a vision of what this might look like given no technological barriers.
However, snapping a photo and instantly gathering data about a room using computer vision seemed to be too hard of a problem to solve before shipping the first version of the product. I believe it's important to ship things fast to get some user feedback on the core product offering before spending years of research, time, and money on something that users may not even find valuable.
I needed to design a flow where the user had to send in their photo and then wait until we turned their room into a 3D environment. In addition, the engineering team estimated it would take longer than expected to be able to "swap" each filter for another instantly. To account for this, we needed to make sure the user could only select one style filter beforehand and when it was ready, they could view that one style in their space and the associated products to purchase.
What we learned from V 1.0 was that there was a huge drop-off from signing up to submitting a photo of their room. After all, the barrier to take a photo of your space is high given you might not be home when you download the app. This means that a large percentage of our user base wasn't experiencing anything.
“There's a very short amount of time to get a user to the aha moment before they bounce”
In order to get users to experience the product before they submit a photo of their own space, we decided to allow users to design with template spaces. This solves a few issues:
After the user completes their first design, they are immediately prompted to take a photo, increasing the likelihood of submitting photos of their own space. When the user's photo is ready, the design experience from using a template room is identitcal to designing their own room.
Another optimization we made was to change the "edit" screen to "products" and have a grid of products. During user tests, we noticed that not every user was figuring out how to select items by tapping on the render. This provided another option for users to select items:
Optimizing for Swaps
By version 3, we learned that users loved to "swap" items to compare how items looked together. Swaps were by far the most sticky aspect of the product, so we decided to optimize for that. We had a few ideas but the one that we thought would really move the needle was to have users design their room from scatch. This meant instead of entering into an already fully designed room, users would add each product one by one until the room was filled. The hypothesis was that users would be more invested in swapping each item until they found one they liked, making the experience more personal. Additionally, because previously the room was already filled, there wasn't much incentive for users to design on their own. Designing from scratch would motivate users to complete the entire room on their own, thus increasing engagement.
Looking at the Metrics
“Great design is designing with data-driven results”
By looking at the data, we can compare how starting from a full design vs. starting from scratch effects user behavior. The week of July 10 - 16 (2017), users were starting from a full design. In this week, the total number of swaps were 355,677 and the average swaps per user was 72.68. One week later we released starting from scratch. In this week, total swaps increased to 1,000,817 and the average swap per user jumped to 172.26. By the week of Sep 11, total swaps (per week) soared to 1,706,687 (4,682% increase) and the average swap per user (per week) was over 333 (359% increase) since the feature's release.
Below are the graphs tracked from Mixpanel:
Average Swaps Per User
Improving the Shopping Experience
Although users loved the product and thought it was fun, we found there was a huge gap in the shopping experience. Because there were no shopping filters, search bar, or sorting features, users couldn't easily find what they were looking for. The app experience was becoming more of a design game than a furniture retailer. One of our main business goals was to sell more furniture on the Hutch platform so we decided to prioritize this to the top of the list.
Give the users the ability to filter down and sort through what they're looking for by adding shopping filters. We started off with price, style, brand, and color.
Although using numericle data to design feature sets is one of the most important things you can do as a designer, sometimes just looking at the numbers isn't enough to truly understand user behavior. Conducting user interviews and/or user tests provide another level of insight that is extremely helpful when designing new features. To do this, we use Lookback.io, a service that records the screen of a subject and captures an audio/visual of their face at the same time. We conduct these tests every 2-3 weeks and the information we recieve is often massively insightlful.
When certain features get too complex, I usually turn to a flowchart to map out exactly how everything is supposed to work. Doing so not only weeds out any design flaws, but is useful for engineers when they are building any given feature to scope.