I helped shoppers feel confident enough to act during Black Friday

I worked on Nike’s HO25 Black Friday promotion at INNOCEAN, from early interaction framing to post-launch analysis. Details are anonymized where necessary. The campaign ran in two phases: a 24-hour pre-heat, when shoppers could add sale items to their carts before the sale opened, and a five-day launch. My work included:

• 01
  DesignDesigned wireframes and localized assets within Nike’s global guidelines, defining user flows and key decision points across the purchase journey.
• 02
  CollaborationWorked with external development and design partners to carry UX intent into production while aligning multiple stakeholder requirements.
• 03
  AnalysisAnalyzed campaign performance data to evaluate behavior and turn results into post-campaign insights.

Shoppers wanted the sale but hesitated at the moment of commitment

Traffic and intent were high, but users dropped off at the moments that asked them to commit: adding to the cart and moving to checkout. The issue wasn’t motivation. It was decision friction.

How might we reduce cognitive load and build confidence during a high-pressure sale?

I looked at where shoppers hesitated, not just where they left

I compared campaign performance data with qualitative observation of user flows, from pre-heat entry pages to peak sale moments.

• Users explored early and committed lateThey browsed during pre-heat but waited for clearer context before acting.
• Exploratory language kept people browsingAmbiguous CTA copy led to prolonged browsing instead of decisive action.
• Early context sped up later decisionsUsers who saw the sale context sooner formed expectations and moved efficiently through later steps.

Then I changed the words, tested what each banner should say, and narrowed the placements

• Clearer CTA copyI moved the pre-heat copy from “Preview sale products” to “Add to cart in advance” so the action was clear before peak traffic.Why Exploratory wording kept people browsing, and users who saw the sale context early moved through later decisions faster.
  As is: Preview sale products

  To be: Add to cart in advance

  Shifting the CTA on the doorway banner (a secondary entry point on the app’s shop home screen) from browsing to early commitment moved users from exploration to action before peak decision moments.

• One message per goalFor launch I made three banner variants with different jobs: one led with the campaign name, one with the discount, and one with “last chance”, so I could see which words earn clicks and which earn purchases.Why A banner that earns clicks may not be the one that earns purchases, so each variant had one job.
  • Campaign nameClick rate 11%Purchase rate 2%
  • DiscountClick rate 6%Purchase rate 6%
  • Last chanceClick rate 8%Purchase rate 1%

  The three launch banner variants, with each one’s click rate and purchase rate.

• Fewer placementsApp Shop Home was narrowed to two placements, the top carousel and the doorway, so attention stayed in fewer places.Why Earlier seasons ran more sections and the doorway still underperformed, so fewer placements each got more room.

Shoppers who added to cart early bought faster when the sale launched

Pre-heat. Compared with the previous flash sale, whose pre-heat copy was “Preview sale products”, the share of pre-heat page visits that ended with an add to cart rose from 16% to 31% across .com and the app (+15 points). On .com alone it went from 9% to 33%.

Launch. Early carts paid off when the sale opened. Orders peaked at 10 AM on launch day, and about 88% of main-page orders on .com (91% in the app) came from carts.

The launch banner variants showed that clicks and purchases need different words: the campaign-name banner earned the most clicks, and the discount banner earned the most purchases.

In the app, Shop Home was narrowed to two placements, the top carousel and the doorway. Doorway clicks roughly doubled and total Shop Home clicks rose 16%.

Launch conversion was lower, even though the pre-heat worked.

What didn’t work. Demand concentrated early. Main-page conversion on .com was 6%, against 8% for the previous flash sale, and the post-campaign review pointed to a bottleneck between add-to-cart and checkout. Removing promo-code entry helped the later step from checkout to order.

What people said. After launch I went through social listening data on 383 public posts from blogs, communities, Instagram and X, from the first pre-heat day through the end of the sale (Nov 10 to 15). People shared the sale early and talked about adding items to their carts in advance, which fits the pre-heat result. Among those who knew about the sale but did not buy, a main reason was disappointment with the sale items and quantities. That points to the deals themselves, not only the flow, as a cause of drop-off. It is a small, self-selected sample, so I read it as a signal rather than proof.

Next time. I would build pre-heat around popular products and add stock and deadline messages to bring people back.

This campaign taught me to design for confidence, not persuasion

• Help people feel sure, don’t push harder.In high-pressure environments, good UX gives users enough confidence to act.
• Clicks and purchases need different words.The banner that earned the most clicks was not the one that earned the most purchases, so each message needs its own job.

I looked at what keeps students practicing after the first week

Keens Academy is a global K-pop training platform where overseas learners take dance and vocal curricula from K-pop instructors. Students follow a weekly loop. I noticed a gap between how that loop was designed and how students actually experienced it, so I proposed research on their motivation, frustrations and how they read feedback.

The platform was built for a learning loop, but I didn’t know what made students keep going

Practicing a dance alone, online, is easy to drop. I wanted to understand three things: what motivates students, how they perceive their own growth, and how they take part in the community.

How might we create an environment where students share what they know and feel more motivated to keep improving?

I interviewed and watched students, because what they say and what they do can differ

Hypothesis. Students would be more motivated if they actively shared tips and know-how with each other.

Who. I grouped students by engagement style, such as those who ask many questions and those who give a lot of feedback, and also interviewed four students who were visiting Korea in person.

How. In-depth interviews, post-practice interviews, and observation of students learning choreography from the curriculum videos.

• Homeroom interaction boosts motivationPeers were the biggest source of motivation.
  • Moments of motivation. As the program went on, students practiced together with homeroom members, which boosted motivation a lot. When they saw others practicing, they joined in or used the community to find practice partners.
  • Past dance communities. Students had tried communities such as Amino, but a lack of active interaction led them to drop off. They wanted a place to share tips and insights, and an inactive community left them discouraged.
• Students prefer expert feedbackThey valued feedback from the dance TA and live lessons over peer feedback, which they found overly positive and sometimes lacking technical depth. Some went to close friends who would give constructive feedback instead.

  “Peer feedback tends to be too sugarcoated and overly positive. I don’t find it very helpful.”Student interview

  “I would rather reach out to my close friends, who give me more constructive feedback.”Student interview

• Dance questions are hard to put into wordsStudents preferred to show their entire dance and ask for overall feedback rather than asking about specific parts. Few used the “I have a question” button in the Practice Hall.Why If they ask “Is this part okay?”, the reviewer might focus only on that section’s flaws. Students wanted fresh perspectives by getting feedback on the dance as a whole.

So the hypothesis held in part: being together mattered, but more peer feedback alone would not help.

Then I proposed changes to the feedback, the button and the size of the room

• A structured feedback systemI proposed shifting from quantity to quality. Students record and review their own dance first and give themselves feedback. Reviewers watch the video at least three times and write specific comments with a minimum length.Why Students found peer feedback too positive and shallow, so the guidelines ask for more careful reviewing.
• A clearer question buttonI proposed renaming “I Have a Question” to “I Need Feedback” in the Virtual Practice Room, and a two-week A/B test with the third cohort.Why Students struggled to put dance questions into words, but they were comfortable asking for feedback on a whole dance.

  

  As isI Have a Question

  To beI Need Feedback
• Smaller homerooms and interactive eventsI proposed reducing homeroom size from 40 to 20 and adding icebreaker events (a GatherTown escape room, practice space tours, quiz-based scheduling), group practice rooms and peer feedback sessions after live lessons.Why Students were most motivated by peers they actually knew, and large rooms made that hard.

  

Students connected more, and feedback got more thorough

In the fourth cohort, icebreakers and group practice rooms strengthened connections between students, and practice time improved compared with earlier cohorts. Homework submission rose 15% compared with the previous cohort. The self and peer feedback guidelines led to more thorough, detailed and constructive feedback.

What’s unresolved. The button A/B test was not run because of time. There is also no long-term tracking of motivation, community engagement or achievement, so I can’t say how long the effect lasts or which change mattered most.

How I would test it. I would split the next cohort into two groups for two weeks, one seeing “I Have a Question” and the other “I Need Feedback”. I would compare button clicks and posted questions per student, and call the new label a success if it beats the current one on both. I would also check that the questions are useful, not just more numerous.

This project taught me to ask smaller questions and watch more

• Observe, don’t only ask.Watching students learn choreography gave richer and more accurate data than interviews alone.
• Ask about specific moments.“When do you feel most motivated?” was too broad. Situational questions for each participant would have produced better answers.
• Keep priorities realistic.Small operational changes can matter a lot with limited resources, as long as they are scoped so they can actually be validated.

Disposable umbrellas are cheap, so people throw them away

Disposable umbrellas are cheap and fragile. Most end up in landfills or incinerators, which produces greenhouse gases. About 50 million umbrellas are sold in South Korea each year, and in one Seoul district about 1 ton is thrown away every week.

What if people could borrow and share umbrellas instead of buying new ones?

Instead of buying a new disposable umbrella each time the weather changes, people could borrow one nearby. I worked with two personas.

• The Forgetful Umbrella UserOften gets caught in the rain after forgetting an umbrella, and is frustrated by how many disposable ones pile up.
• The Small Business OwnerCustomers often leave umbrellas behind. The owner can’t use them, but can’t simply throw them away either.

How might we turn forgotten umbrellas into opportunities for users and local shops?

Shops and users both gain when umbrellas are shared

Parasol manages umbrella rentals and gives local stores a marketing platform. A shop offers a little space for umbrellas, earns rental income and attracts customers. A user rents nearby, saves money and takes part in eco-friendly sharing. The design challenge was making this effortless for both sides, especially for busy shop owners.

I found a problem in the first flow: shop owners had to scan every rental

We improved the UX through continuous user research and usability testing. The original design split Parasol into a User mode and a Store Owner mode. Owners had to scan each user’s QR code by hand for every rental and return. We questioned whether that was too much for busy owners and looked for a simpler way.

I asked 70 potential renters how they feel about it

I ran an anonymous Google Forms survey of 70 potential renters. Past rental services were inconvenient to return, had return spots that were too far, and had too few umbrellas. For renting from local stores, their worries were having to return the umbrella, waiting when the shop is busy, and feeling self-conscious about renting only an umbrella.

I interviewed shop owners around Sinchon, because the service needs them too

These were qualitative interviews with small business owners in the Sinchon area about their pain points, daily operations and expectations.

Shops worry about effort and complaints, and users about feeling awkward

Then I changed the flow to work for both owners and users

Rentals get smoother for both sides, and shops get more exposure.

I tested the revised flow with three target users

We evaluated and refined the experience through usability testing. Three target users completed three tasks: renting an umbrella, returning it and finding a store. They gave positive feedback that the revised flow was more efficient. The test also revealed new pain points, which led to two further improvements.

This was a qualitative test with a small group, so I did not measure success rates or completion times.

The final screens bring these changes together

Map tab

Umbrella rentals nearby: a quick map for sudden rain

The map shows partner stores for quick location. Users can see store details and umbrella availability in a bottom sheet, and navigate, rent and return directly from the map.

9:41

Search

MapMy umbrella

9:41

Search

Bellapraha Bakery77 Ewhayeodae-gil, Seodaemun-gu, SeoulOpen · Closes at 10 PM

Umbrellas available3

RentReturn

• The QR code scanner sits at the center of the bottom bar, so users can act quickly.
• The “Directions” button lets users navigate to a store without leaving the app.

Store details screen

Maximizing local store visibility

I revised the original design by enlarging images and adding more of them, to maximize store visibility. After an umbrella is returned, the app recommends nearby stores to help users explore.

9:41

Bellapraha77 Ewhayeodae-gil, Seodaemun-gu, SeoulOpen daily 10 AM – 10 PM

Best dessert spot in Sinchon
Fresh-baked financiers for rainy days

Umbrellas available: 3

RentReturn

9:41

Bellapraha

77 Ewhayeodae-gil, Seodaemun-gu, SeoulOpen Daily 10 AM – 10 PM

Best dessert spot in SinchonFresh-baked financiers for rainy days

Umbrellas available3

RentReturn

9:41

Umbrella returned

9:41

My Umbrella

🌈After the rain,
there’s a rainbow.

June, how about these shops?

BellaprahaSeodaemun-gu

HyeonokSeodaemun-gu

Local SikdangSeodaemun-gu

• The store detail screen has larger images and additional photos.
• After a return is complete, the app recommends nearby stores.

Rental and return screen

Contactless umbrella rental and return

Users rent and return umbrellas directly with the QR code on the umbrella locker. They can act without worrying about how busy the store owner is, and owners carry less of the burden of managing umbrellas.

9:41

To rent, scan the QR code
on the umbrella stand

Scan QR code

9:41

Umbrella rented

User

“I like that I can rent an umbrella without worrying about the store owner.”

Store owner

“I don’t have to worry about umbrella rentals when the store is busy!”

My umbrella tab

Easy umbrella management

Usage and overtime are shown in a simple graph for quick insight, and umbrellas are sorted into rental and sale to keep management simple.

Parasolnow

Your umbrella is 3 hours overdue and the late fee is ₩300. Return it now?

9:41

My Umbrella

BellaprahaRented Jul 2, 2023 · 4:00 PM

Overdue

Time used24 / 100 hrs

Overtime3 hrs

Late fee₩300

Late fees are capped at ₩10,000.

Return umbrella

SellHistory

My points5,000 P

Carbon saved3,049 g

Sell an unused umbrella

MapMy umbrella

9:41

Rental history

Aug 5, 2023Bellapraha

In use

Jul 9, 2023Hyeonok

48 hrs₩2,400

Jul 5, 2023Local Sikdang

10 hrs₩0

Jun 3, 2023Hongik Bookstore

100 hrs₩10,000

MapMy umbrella

9:41

Umbrellas I sold

Aug 5, 2023Sold

+1,000 P

Jul 15, 2023Sold

+1,000 P

May 5, 2023Sold

+1,000 P

Mar 25, 2023Sold

+1,000 P

MapMy umbrella

• Users get an alert when the return time is exceeded.
• Users can sell their disposable umbrellas and receive points that can be redeemed as gift vouchers.
• Users can view their rental history and the umbrellas they have sold.

Testing early helped, but I missed what shop owners expected

• Test early, test often.Testing early and often improved the product. Changing an idea early is easy, but changing a finished plan is hard, so testing right before launch does less. Prototype tests validated the core concept before design and development, and we ran them throughout planning.
• Business context matters.Shop owners decide whether the service exists at all. Their effort, risk and gain shaped adoption as much as user needs did, so I should have tested a concrete incentive for them earlier.
• Finding problems is not enough.After testing the improved MVP, users were not interested in the ads, and the promotion did not satisfy shop owners. I need to understand root causes, not only symptoms.
• UX alone has limits.User research validated our hypotheses and confirmed the problems. But we did not accurately identify what shop owners expected, so UX improvements alone fell short.

Parasol never launched, so here is the hypothesis I would test next

We built the app, but it never launched. The last round of testing showed that better visibility alone gave neither users nor owners a reason to act. So the question I would test next is about rewards.

More people live alone and cook for themselves, but cooking is hard

The number of single-person households in Korea keeps rising. Most of these households cook for themselves, yet many struggle with cooking knowledge, limited ingredients or lack of time. Students living alone told us about the same daily struggles.

Sources: OpenSurvey, 2022 Single-Person Household Trend Report; Statistics Korea.

I asked 75 people in their 20s and 30s about cooking and short videos

The survey covered cooking pain points and how people feel about short-form recipe videos.

Survey of 75 participants in their 20s and 30s.

People like short videos, but can’t cook along with them

• Searching for recipes.People switch platforms depending on the need. Short videos suit simple recipes, while blogs and text recipes work better for baking or precise measurements.
• Using short-form videos.The pace is too fast and pausing is awkward. Some people screenshot ingredient lists because they are hard to remember.
• Cooking alone.Bad past experiences with difficult recipes or unfamiliar ingredients make people avoid some recipes. They are more willing to try recipes that use ingredients they already have.

Friction shows up at every step, from deciding what to eat to saving the recipe

Four insights shaped the product

Existing options are complex, scattered or hard to save

The gaps point to an opening: easy recipes, curated collections and meal plans, and one place to save them.

Recipeasy: one-minute recipes for the home cook

Recipeasy solves everyday cooking dilemmas with one-minute short-form video recipes. The recipes are easy, entertaining and varied, with themes that are beginner-friendly and ultra-simple, made for people living alone.

Make cooking easy, entertaining and diverse

Three goals guided every screen: easy, entertaining and diverse. These are the screens that put the strategy into practice.

UX Strategy

Make Cooking

EasyEntertainingDiverse

Quick and simple to followBeginner-friendly Recipes

Engaging and entertaining1-minute
Short-form Video Recipes

Diverse Themed RecipesSolving Everyday
Cooking Dilemmas

Rethinking the business model for perceived value

We launched the MVP for free to lower the barrier to entry. Interviews and surveys then kept showing the same thing: most users hesitated to pay because they did not yet see the service’s value.

I led the call that this was a business-model problem, not only a UX one. I shifted the question from how to charge to how users first experience and appreciate the value. We explored advertising, partnerships and premium features as ways to balance user satisfaction with sustainability, but our work stopped at exploration, so none of these was validated.