About
Top

Chewy
Autoship

UX Strategy, Service Design

Two phones showing Autoship setup and item-level management
Problem Data showed Autoship churn. Customers were canceling and reordering through the cart.
Insight When a repeat order came up, customers often weren’t out of pet food after all. But the system locked every item to one schedule, so they canceled instead of skipping.
What I did Led the service model redesign to item-level subscriptions, and reworked the order timeline with payments, fulfillment, and stock teams.
Result Order notifications with twice the time to edit, price and stock held at the reminder, and 27.75% fewer negative contacts.

Autoship subscriptions are Chewy’s largest revenue driver: a $7B+ program facilitating ~73% of net sales. Yet, a critical flaw in its design left customers frustrated and frequently canceling their subscriptions.

As it turns out, customers really do want to just set it and forget it—Autoship promises regular recurring delivery of pet medication, food, supplies, and treats on a schedule of the user’s choice. For Chewy pet parents, that generally means getting food once a month, flea meds quarterly, and the rest sporadically throughout. It’s easy to opt into Autoship during checkout, and the service is supported by a management portal and email-based notifications.

However, customers struggled to rely on the system, often finding themselves spending more time coordinating restocks than focusing on their pets. The legacy basket-enrollment model simplified initial setup by allowing customers to apply a recurring order agreement to an entire order, but it posed significant challenges for managing individual items. Customers frequently found that their needs for specific items, such as pet food, did not align with the cadence of other items, like medications—and the management interface didn’t allow individual items to be rescheduled.

Legacy Autoship order page surrounded by customer quotes about managing items and dates

Serving as Staff Designer for Autoship Management, alongside my partner dedicated to Acquisition, we formed a working group with PMs and Engineering leads to define the new service architecture and operating principles for a subscription service driven by individual item scheduling. At the core of our negotiations was convenience versus control, for both the customer and the business. Consolidating shipments served the business, while simpler rescheduling flows gave customers back control.

In the legacy interface, many customers resorted to canceling Autoship orders and using the cart when they couldn’t adjust timing to their needs. While this appeared as a high rate of churn quantitatively, UX research provided the full picture and helped define the customer mental model. Our redesign empowered customers to manage Autoship the way they think about grocery shopping—stocking up when needed, skipping when they don’t.

Before and after: basket-based Autoship screens beside the new item-level screens

With the shift to item-level Autoship, the order construct became an event rather than the pattern. Items could ship together one month but recur only when needed. I expanded flows for customers looking to expedite their orders and simplified the process to postpone. Most critically, we ensured that customers just starting their Autoship subscriptions weren’t locked into receiving the same basket mix every month.

New Autoship pages across breakpoints, with order rescheduling and expedited order flows Project scope: 540+ hi-fi screens, 30+ user flows, a component-defined design system, and a rolling release schedule Item-level details: schedule options, contextual review, expanded management, and item status

Though the updated interface aligned more closely with the customer's need to manage item frequency independently, we soon discovered other frustrations with the service—many tied to order processing mechanisms and backend timing.

Customer quotes about reminder email timing, charges, and stock

Patterns emerged regarding service areas that primarily relied on email as the customer-facing communication device. Autoship emails had no single owner, and they had been designed by several stakeholders, often without support from the CX or design team.

Timeline of the existing Autoship service and emails, showing friction points across a two-week order cycle

Creating a comprehensive timeline of our processes and email triggers helped frame customer frustrations—and pinpoint specific backend inefficiencies. Reminder email timing was tight, giving customers only 48 hours to make amendments. Many Autoship customers reached out for support upon seeing the authorization charge, which wasn’t mentioned in the interface or emails. Most surprisingly, while we promoted Autoship as a way to prioritize stock, we weren’t actually reserving inventory for Autoship customers.

Timeline after the service restructure, with stock reserved and more time to edit

Improving the Autoship process required negotiations with payments, fulfillment, and stock teams. We introduced additional look-ahead services, allowing us to lock in customer prices and reserve stock the moment the reminder email was sent. This revision doubled the time customers had to edit their order, providing a clear picture of what was in stock before being charged. Authorization amounts better matched the final order total, and we ensured order processing worked ahead, in sync with each customer’s Autoship settings.

Three redesigned Autoship emails: review your order, preparing your order, and on the way

As new services rolled out, emails were updated incrementally, but our first win was limiting messaging during order processing to clearly communicate our authorization hold procedure. This alone reduced negative customer contacts and email replies by 27.75%.

Getting payments, fulfillment, and stock teams to change how orders were processed took more than a good design. I paired churn data with what customers told support and what research uncovered, so each team could see its piece of the problem. Framing trade-offs plainly let us ship quick wins, like clarifying the authorization hold, while bigger backend changes lined up.

One of the toughest trade-offs in this project came early: rather than iterating on the existing architecture, we were tasked with building an entirely new system and gradually migrating customers over time. While this approach allowed us to rethink the service from the ground up, it also meant that we couldn’t test individual features within the legacy interface—such as an item list view or single-item rescheduling. In hindsight, this monolithic approach ultimately cost the project its original launch date. The decision was later revised to introduce the experience in phases, but had we engaged engineering partners sooner in questioning the all-or-nothing model, we might have found a more flexible path earlier. Given the chance to revisit this decision, I would have focused on the iterative rollout strategy's role in ensuring we could validate improvements without delaying customer impact.

Colleague feedback about Eva’s work on Autoship

Moving from a rigid order structure to a flexible, item-driven system future-proofed Autoship for new subscription options. The team’s item-level enrollment launched first and added $1M+ a month in profit within two quarters. In management, customers now handle each item on its own schedule, with twice as long to edit orders, prices and stock held at the reminder, and 27.75% fewer negative contacts. Autoship finally delivers on its promise, letting customers spend time with their pets rather than their Autoship account.

Go back