Homebase

Making it easy for 3.5M+ hourly workers to share when they can’t work.

Role

Product Designer

Context

Internship

Timeline

May - Dec 2026

Tools

Problem

Employees couldn't enter their true availability using just the app.

Due to limitations in our current availability model, employee availability requests repeated weekly with no end date.

“Once my track season starts in January, I won’t be able to work on Fridays or Saturday until March or April.”

Example of an employee using the “Availability Notes” field to describe a seasonal schedule the app had no way to enter.

“Once my track season starts in January, I won’t be able to work on Fridays or Saturday until March or April.”

Example of an employee using the “Availability Notes” field to describe a seasonal schedule the app had no way to enter.

“Once my track season starts in January, I won’t be able to work on Fridays or Saturday until March or April.”

Example of an employee using the “Availability Notes” field to describe a seasonal schedule the app had no way to enter.

Class schedules, other jobs, and one-off changes were more easily communicated outside of Homebase.

The majority of availability updates typically happened over text, email, or in-person conversations, meaning the availability saved in the app was often out of date.

My Role

I owned the availability redesign end to end, from initial research to eng handoff.

Research
Prototype
Test
4 iterations

Background Research

Every scheduling app I studied asked for days and times first, which isn’t how employees think about availability.

These days-and-times forms work, but our testing showed they didn’t match how employees generally think about when they can and can’t work.

7shifts

Deputy

When I Work

Gusto

Initial Insight

One availability request could be hiding several reasons, each with its own cadence and end date.

Employees had to put every commitment into one availability request, but each reason they couldn’t work had its own end date or recurrence, which our existing system couldn’t capture.

A single weekly schedule could hide several commitments (old UI).

Iterations

The availability form went through four major revisions, with each round of changes informed by user research.

I tested every version before starting the next one, with teammates, customers or employees depending on the round.

To start, I rebuilt the availability request to mirror our current model: one weekly schedule, Monday to Sunday, with new inputs for custom start and end dates.
One weekly schedule could mix a few commitments, such as practice on Sunday and class on Monday. Each reason gets its own “time block” that can change or end independently.
On calls, small business owners talked in dates, such as “I’m not available Mondays, starting on the 12th.” In V1, the days and the start date were two separate questions. Here, I made a calendar the form’s primary input, so tapping a date sets both the weekday the entry repeats on and the date it starts.
V3 cleans up the V2 form into the final version, validated through usability testing and multiple rounds of design crit.
To start, I rebuilt the availability request to mirror our current model: one weekly schedule, Monday to Sunday, with new inputs for custom start and end dates.
One weekly schedule could mix a few commitments, such as practice on Sunday and class on Monday. Each reason gets its own “time block” that can change or end independently.
On calls, small business owners talked in dates, such as “I’m not available Mondays, starting on the 12th.” In V1, the days and the start date were two separate questions. Here, I made a calendar the form’s primary input, so tapping a date sets both the weekday the entry repeats on and the date it starts.
V3 cleans up the V2 form into the final version, validated through usability testing and multiple rounds of design crit.

Final Design

The form starts by asking why you can’t work, then builds the request around your answer.

Splitting requests into one entry per reason is a new way of thinking about availability in Homebase, so the form has to work for active employees who are used to putting every reason they can’t work into a single request. To teach one request per reason, the form opens on a single question, “Why can’t you work?”, and builds the request around that one reason.

Why can’t you work?
Select a reason
Autofill the form
Send for approval

Validation

A final usability test settled the last few open decisions about the form.

In addition to presenting prototypes at design crits, I ran a usability test with several members of Homebase’s internal CX and Growth teams (who were great proxies for customers), and built the final form around those findings.

I tested two versions of the form for adding a time you can't work. One put the reason, days, times and repeats on a single screen you scroll. The other moved days and times to a second screen you tap into, which looked cleaner but added a step.

5 of 7 participants preferred the single screen. Navigating between two sheets bothered participants, whereas nobody had an issue with seeing every field on one sheet.

The first field is where an employee says what the time is for, like Classes. One version called it "Label", and the other called it "Reason".

All participants resonated more with "Reason". To them, a label is a note to yourself, whereas a reason feels like something you give your manager.

The form already asks for a reason, days, times and locations, so an optional note to the manager could be one more field that repeats what the form already says.

Many participants found the note useful, each for something the reason didn’t already cover: how urgent the request was (for example, a son in the ER), context that changes the manager’s decision (classes that end this semester), or flagging a mistake. I decided to keep the optional field.

“I like seeing everything on the screen and going down, going section to section and selecting everything.”
P5, CX Team
“I think with showing all the options instead of having to click into a lot of loops, the simplicity of it makes it easier to use for everybody.”
P6, CX Team

Success Metrics

More employees with availability in the app, and updating their availability there instead of over text.

Of the millions of employees with a shift in the last 90 days, about 1 in 4 had set their availability, and about 1 in 7 edited their availability in that time. Twelve months after launch, we’re aiming for:

27% → 35%

of scheduled employees with availability on file

14% → 20%

of scheduled employees with updated or confirmed availability in the last 90 days

Reliable availability is also a prerequisite for unlocking Homebase Assistant!

To support future agentic scheduling tooling, we’ll need reliable employee availability. The new availability design empowers employees to say something like “My class on Tuesday got canceled, let my manager know I can work,” and the assistant could find the Classes entry, mark Tuesday available and tell their manager, all in one go.

Reflection

Availability is a surprisingly complex surface.

Exploring ideas in Figma and Claude Code made changes cheap to test, which left more time for choosing the “right” option and making the inputs clear enough that a 16-year-old working at an ice cream shop could understand them.

Next project

University of MichiganBuilding a Google Slides extension to cut accessibility remediation time by ~80%.Jan - Apr 2026