Mindful Mint
3 min read
Product & UX Designer, end-to-end.
Web & Mobile, Zero-to-one.
10 weeks, Fall 2024.
In a team of 3, I owned the web. Teammates owned the mobile & browser extension.
Figma, Whimsical, Proven By Users.
Created an app against dark-patterns in e-commerce
I built an app that counters impulse buys while preserving user agency. It gives online shoppers a cooling-off period they already wanted.
Impact
Task completion
50% → 85%
Hesitation at list creation
60% → 10%
Two rounds of tree testing (n=11) and one usability round (n=6).
These after-numbers are re-tested.
The problem: Lack of organized virtual window shopping
Competitors informed:
Budgeting apps tell you what you already spent.
Coupon extensions help you buy faster.
Wishlists just store things.
None of them show up at the moment the user is actually deciding.
Five user interviews informed:
5/5
Users shopped late at night with autofill or one-click checkout enabled.
4/5
Users traced their last impulse buy to a social post or a promo email, and felt guilt afterward, usually because the item was a duplicate.
4/5
Users had already invented their own cooling-off hack: leave it in the cart for a day, a week, a month.
Research conclusion:
People already had workarounds, just in their carts.
See the competitive Landscape —>
Affinity Board —>
The Job To Be Done:
When
I'm browsing late at night and something catches me,
I want to
park the item somewhere that holds it for me,
So I can
decide later whether I actually wanted it.
Decision made:
Don't block the purchase or stop their flow; add mindful friction and buy users time.
Save it with timer
Drop the item into a list that carries its own cooling-off period. Tags are one tap. No setup screen.
Answer a few questions
Want it, or need it, and why. One screen: I kept it off a form on purpose.
Decide when it comes back
The reminder returns the item when the window closes, purchase link still live.
VisualThree real screens
01 Lo-fi first
User-testing with screens on the table instead of questions.
02 Tree test
Somewhere to hold the thing that isn't the cart, and a reason to come back.
03 Usability
6 people, 1 round, 5 tasks. The same two labels broke again.
04 Fixes
Two renames and one collapsed entry point. See frame 05.
I brought wireframes to the first interviews so users got to engage with something real instead of having to answer loosely. Insight from that set up everything after it.
Turns out my architecture was fine. My labels weren't.
Participant said and I quote:
“The labels were mostly clear, except for the last one, which I was confused by greatly. I honestly just guessed.”
Tree test participant, round two
So what really happened?
87.5% took the wrong path to the reflection flow.
But once they got to the right section, they were fine, so the structure held.
People just couldn't read the doors.
“Reflection” was the word I used, and it proved ambiguous without enough context.
Then I renamed one label between rounds, and completion on that task dropped from 87.5% to 33.3%. One word.
Full research write-up —>
Save it, sit on it, decide later.
What features did I use to achieve that?
Search & Add
Find the thing, drop it in the review basket, one tap.
Lists
Everything you've saved and how long it has left, in one view.
Reflection prompts
Want or need, answered in seconds. Not a form.
Reminder
Pick your delay from presets built on how people already wait.
Item summary
Confirm, and it goes quiet until the timer's up.
Open the interactive prototype —>
Renaming & making items optinal - increased completion
List creation —> hesitation rate;
60% → <10%
Wishlist expiry date and notifications were required fields, so people stalled before they'd seen the point of the features.
Made both optional, moved them to toggles.
Pending product evaluation —> completion;
50% → 85%
Gave it a plain name "The Cooling-Off Cart".
Collapsed the redundant entry points into one.
MVP Scoping and User-flows —>
Iterations & Trade-offs —>
What I'd do with another month
Re-test
Test round two on high fidelity, and re-test the renamed labels.
Build
Build out the price calculator. “This costs you 4 hours of work” got the biggest unprompted reaction in testing, and it still only lives on one screen.
Render
Iterate an industry-standard UI.