BudgetCart
Testing showed me which details shoppers needed to choose.
BudgetCart looked at grocery shopping when several constraints pile up at once: a weekly budget, SNAP/WIC eligibility and dietary needs.
The first browse direction stripped away brand, price and store details for simplicity. During prototype tasks, people struggled to find items and felt less sure of their choices.
Design decisions.
Planning around a budget, benefits and food needs.
A weekly budget, SNAP/WIC eligibility and dietary needs can all matter in the same decision.
We hid too much.
The first browse direction stripped away brand, price and store details for simplicity. During prototype tasks, people struggled to find items and felt less sure of their choices.
Bringing back brands, stores and prices.
The revised browse and store screens showed details at the moment of choice. We did not retest those revisions.
Giving shoppers a place to start their cart.
I developed starter cues for the AI Cart Builder interaction instead of an empty prompt. The Budget Calendar was another feature in the team concept.
Revising the browse screens after testing.
We revised the prototype based on task feedback. The project didn’t launch, and no grocery spending changes were measured.
Build a cart from a budget.
The cart builder starts with a budget and food needs. Shoppers can edit the suggestions.
Show the details at the moment of choice.
After the first prototype made browsing harder, the team restored brand, store and price information. The revised flow was not retested.
Compare stores before checkout.
Compare the cart total and SNAP/WIC labels before choosing a store.
More process notes
BudgetCart looked at grocery shopping when several constraints pile up at once: a weekly budget, SNAP/WIC eligibility and dietary needs.
I started with interviews and paper prototypes before opening Figma. It was my first Figma project.
Show the cues people choose with
We chose: Item-first browsing with price, brand and store visible
Over: Hiding brand, store and price to make the screen look cleaner
When those details were hidden in testing, people had a harder time finding items and trusted their picks less.
Prototype task testing. We didn't retest the revised screens.
Give the AI a starting point
We chose: Upload cues and a starter prompt in the cart builder
Over: An empty chat input
An empty prompt doesn't tell a shopper what the assistant can do. The cart it drafts is a starting point, and they edit from there.
My design rationale. Not retested.
Keep the budget in view while shopping
We chose: Budget and spending days in a calendar the shopper can check any time
Over: Showing the budget only at checkout
The budget matters while you're choosing, not after you've checked out.
Team concept rationale. We didn't measure spending.
Prototype task testing with the high-fidelity Figma prototype.
Not tested: The revised screens
Not tested: Long-term spending or checkout behavior
Tested with prototype tasks, then revised the browse and store screens based on what we saw.
Not a live service. We didn't measure spending, and the revisions weren't retested.
People needed the brand, store and price to choose. I’ll be more careful about removing useful details just to make a screen look cleaner.
In a retest, do shoppers find items faster and put fewer back at checkout?