A few months at Shipt allowed me to conduct a few user research projects, help build our DLS, & more!
Shipt is a same-day delivery service - focusing on groceries and home products.
The problems I worked on necessitated spending much of my time conducting user research and build consistency across our app / website.
— designer asking for a
larger research budget
Survey says ____, but what do they really think?”
Don't mind me I'm just breaking up all this text with another empty state :)
Worked on a team of four to conduct user research on our product re-ordering feature. Buying the same product over-and-over again is common; making being able to easily repurchase items a necessity for Shipt.
We used usage analytics, screening surveys, and interviews like the previous project - albeit in different ways.
By targeting mostly highly active users we learned the various ways people re-order items and their pros/pains. Live tests were conducted during interviews to more accurately paint the paths taken when shopping with Shipt.
Talking with a top cohort of users ended up shedding light on the reasons people loved our product and chose us for delivery. This led us to allocating some time for gaining these insights as we carefully iterated on our interview process.
Worked on a team of two to set goals for onboarding research based on funnel analytics. We aimed to uncover different issues people faced preventing successful onboarding and highlight the main reasons successful onboarding occured. This would allow our design team to fix the issues while building on what was already working.
We used analytics to design survey and interview questions. Sent out surveys to screen for potential interviewees — using the results to improve our interview questions. Conducted interviews and synthesized findings / actionable insights.
These steps involved general and cohort specific questions for three groups: people who stopped partway through the sign-up process, those who signed-up but did not place (m)any orders, people who were successfully onboarded and consistently used Shipt.
End-to-end research on our user onboarding and product reordering experiences.
A nice amount of my time was spent working on improving our product's consistency by building design language system components and editing / making icons.
Perfect pixels make perfect
Head of Design
I also worked on a few smaller projects involving competitive analysis, flows, benchmarking, and empty states.
During interviews I had a design challenge to redesign Shipt's checkout flow to reduce friction and cart abandonment.
This was a 30+ minute presentation, but I will provide some highlights of my process. Leaving the designs as they were presented so I can point out what I would do differently now.
I have a background in data-science which lends me to emphasizing quantitative research when possible. Combining this with benchmarking of successful flows and secondary research led to creating 5 points to emphasize for the redesign.
Baymard Institute found the pre-2019 average documented online shopping cart abandonment rate to be 69.57%
The Virgin America booking, B&H Photo and Video checkout, and Amazon checkout processes were among those benchmarked due to high success, third-party ratings, and scale respectively.Points of Emphasis:
emphasize closeness to free delivery
feature necessity & staying power
short & uncomplicated
keep total cost visible
one task at a time
Current checkout process.
Feature Necessity & Staying Power:
Delivery Address and Payment Method are pre-filled. Though they should be editable. Having them at the top wastes a scroll and your first look—increasing friction and the chances of you leaving checkout.
Costco's Membership Info does not need to stay in the checkout flow (though maybe business deals require it). If you don't have a Costco cart you need not see an input field every time you checkout. It can be better placed during on-boarding as payment info is added and in the payment section of your profile.
One Task at a Time:
Redesign surfaces one task at a time. After you select your Delivery Time, etc. the next option becomes the most important element. The bottom (eventually place order) CTA also hints at what to do before clicking it to place your order.
Emphasize Closeness to Free Delivery:
$35 orders give you free shipping. Your proximity to this is highlighted.
Keep Total Cost Visible:
By emphasizing one task at a time there is enough room to keep the cost visible.
Feature Necessity & Staying Power:
Waiting until the end to bring up Address/Payment allows you to focus on what truly needs your attention first. Then usually jump over these two options, leaving them unchanged, and placing your order.
Short & Uncomplicated:
The hierarchy changes limit the need for scrolling and give you a clearly defined path to checking out. The option buttons resemble product category buttons in the app; enabling component consistency to ease your checkout. Lowering the chances of friction causing you to leave the screen and not get your groceries.
Looking back with more experience, there are a few decisions I would change. The major changes are described below.
I overuse Shipt's brand and CTA color. This draws too much attention to the page's title and already completed tasks. Instead it could be more helpful to de-emphasize already selected options, bringing more attention to the next task and free shipping copy.
These weren't designed on smaller screens making the Order Total potentially hidden throughout the process. This presents a placement challenge of deciding between placing it in the bottom CTA bar, decreasing the above copy/spacing (it is pretty tight), etc. I'm curious how a Typeform-esque checkout approach could keep costs always showing but push Delivery Time, etc. above the viewport after an option is selected; forcing a scroll up to change previously selected options.
Small print is more hidden than I feel appropriate—it concerns item pricing based on options above. Bringing attention to it during your first checkout could allow for it to be more hidden, as it is, later on. As it is unnecessary to read every order.