Calm in the complexity
Spark in every insight

Welcome to Luna's Notes

Keep your center. Bring your own point of view.

I'm a strawberry matcha girl. The colors of Luna's Notes come from a strawberry matcha: two flavors that stay distinct in one glass. They are also how I work.

Matcha

My grounded side

Patience with messy information, careful analysis, and the steadiness to work through uncertainty.

Strawberry

My spark

Curiosity about people, a distinct point of view, and the confidence to question the first explanation.

The blend

My working style

Thoughtful analysis with enough imagination and empathy to make the result useful.

A calm, curious marketing analyst who finds overlooked signals and turns them into thoughtful action.

Let's have a Strawberry-Matcha Chat With Me

Hi, I am Chanbyeol Luna Choi

I am Marketing Analyst and Content Coordinator in Vancouver.
I find the signal in the messy parts of marketing: the customer feedback, campaign results, and overlooked details that reveal what people need and what could work better.

As a Marketing Analyst & Content Coordinator, I bring those insights into social media content, email marketing, branded design, SEO, segmentation, campaign strategy, and go-to-market execution. I use performance data to understand what’s working and why, then turn those learnings into more relevant campaigns and measurable customer acquisition.

View résumé choichanbyeol@gmail.com LinkedIn

How I work

Find the signal

I start with the parts other people skip: the setting no one checked, the data that only exists in a handwritten log.
Know the audience first

I segment the audience before writing a single line of copy, so the creative starts from a real situation.
Make it land

Good analysis is half the job. The other half is catching a risk early and turning a number into something a decision maker can act on.

Toolkit

Marketing: Canva, CapCut, A/B testing, SEO, Google Ads, Google Analytics
Data: SQL, Python, Excel, Power BI, Tableau
AI: Claude Code, Codex, ChatGPT, Gemini

Education

Data Analytics Co-op Diploma: Greystone College, Vancouver
Bachelor of Business Administration: Inha University, Incheon

Where I've built career

Roles, the problems I owned, and what changed.

View résumé

Sep 2026 – Present Vancouver, BCHybrid
Marketing Specialist / ProductBC
- Runs email campaigns for BC's product management community
- Applies audience segmentation and data analysis to inform targeting and improve conversion
- Designs on brand visual assets in Canva for email and social
- Creates and publishes LinkedIn content to drive event awareness and registrations
Selected work · ProductBC 2026 Conference campaign

Email header, Canva

Speaker announcement, Canva

Read the email LinkedIn post 1 LinkedIn post 2
Jun 2026 – Aug 2026 Vancouver, BCFull time
Marketing Analyst / PhoneBox
- 265 organic clicks and 6,300 impressions within 3 weeks, after building SEO from zero and tracing lost visibility to missing canonical tags across 6 pages
- 8 Meta Ads creative concepts, 33% beyond the brief and all approved for launch, grounded in situational empathy instead of volume
- 50% less designer review time and about 33% less production time, by sharing AI generated visual samples inside a restructured creative brief
- 2 of 8 candidate countries kept for a Google Ads campaign, after testing nationality assumptions against university specific data
Apr 2026 – May 2026 Vancouver, BCFreelance
Business Analyst / Platform 7 Coffee
- 5 months of handwritten POS, inventory and shift logs digitized into SQLite where no data infrastructure existed
- 40% more espresso bean orders against 20% fewer espresso sales, surfaced by SQL KPIs that turned the next version toward the swing shift cut
- ~$570 a month in saved wages set against a ~$5,171 a month lunch revenue decline, with 4 recommendations focused on demand
See for more →
Oct 2025 – Dec 2025Palo Alto, CARemote
Data Annotator / Turing
- Selected by team leadership as QA reviewer for evaluation accuracy on AI generated voice content
- 3 annotators recalibrated through example based training materials and a live session I led, after tracing a recurring misreading of rating guidelines to a team wide alignment gap
- 37 to 45 daily tasks per annotator and an estimated 22% less rework time after the calibration, with the recurring error patterns gone from later QA reviews
Nov 2022 – Aug 2023 Seoul, South KoreaInternship
PTKOREA
NPS Data Analysis Assistant

Mar 2023 – Aug 2023
- 3 point NPS improvement during the reporting period, supported by reclassifying unstructured Other responses from Samsung UK and Japan into actionable subcategories
- URL based satisfaction monitoring proposed after I found untapped tracking data in the NPS dataset, adopted by the client as an ongoing reporting service
Samsung Unpacked Global Website Communication Assistant

Nov 2022 – Feb 2023
- 0 reported issues on the Galaxy S23 Series website launch day, down from more than 10 at the previous launch, working with APAC region managers
- 15% less testing time with a QA issue tracking and reporting system I designed in Excel and PowerPoint, adopted across regional teams including APAC
Mar 2021 – Feb 2022 Incheon, South KoreaInternship
Marketing Intern / Emons
- PEST and SWOT analyses of market trends that fed into the 2022 New Year plan
- The company's first live commerce event in February 2022, after I researched competitor promotions and identified live commerce as the opening

How the results got made

The problem, the analysis, and the decision behind each piece of work.

Data strategyData lead

VANORI

AI-Based Local Information Curation Mobile App

An AI app that curates Korean community information in Vancouver, rebuilt on data it could legally use.

Data strategy redesigned 22 days before launch

See for more →

Go to marketMarketer

TooGoodForFun

Last-Minute Activity Deals Marketplace

Same day deals on empty off peak activity slots, built in 6 weeks at the ProductBC Build-A-Thon.

Working prototype demoed live on Demo Day

See for more →

Operational analyticsSolo analyst

Platform 7

Café Shift Analysis

A data backed counter argument to a café staffing decision, built from 5 months of handwritten logs.

A demand problem, not a cost problem

See for more →

Scroll over the cards, swipe, or use the arrow keys

VANORI

An AI powered mobile app that brings Vancouver's scattered Korean community information into one place. I was the team's data lead.

PM & Data Lead 30 day AI app bootcamp2026

← All projects

My role

Data lead on team DooriDoori, 4 members. I owned API research, the database schema and the curation strategy.

Problem

Korean community information in Vancouver is split across KakaoTalk open chats, Instagram, Naver Cafe and Google Maps. Students, working holiday visa holders and new immigrants struggle to tell what is relevant. The original data plan also assumed sources the app wasn't allowed to use.

The audit

Source	Decision	Why
FSQ OS Places	Base layer	Apache 2.0: permanent storage and AI use allowed
Google Places	Display only	Terms prohibit AI training and long term storage
Meetup API	Events	Only free API with city wide event discovery
City of Vancouver Open Data	New openings	Licence dates surface new spots before Google does
Threads oEmbed	Community voice	The only legal channel into Korean community social content
Yelp, TripAdvisor, Reddit	Excluded	Analysis bans, approval timelines or cost

The design

Universal coreName, address, coordinates, category, hours, price range and source for every row.
Persona matching tagsVibe, neighborhood, Korean signal score, budget tier and time of day, the inputs for AI recommendations.
Category extensionsOptional fields for food, events and lifestyle places.

Results

App demo · 72 seconds, no sound

A redesigned data strategy and a curation workflow for the MVP :
a bulk import of 5,000 to 15,000 Vancouver places, 100 newly opened businesses, 50 community events, and 200 priority places tagged by hand.

The deck

Scroll sideways · 14 slides

View the project deck · PDF

What I learned

Checking technical assumptions against legal limits early saves the build. The real work was turning compliance findings into a decision framework a non technical team could act on.

TooGoodForFun

A mobile marketplace where local businesses sell empty off peak activity slots at a discount, and people nearby book them the same day. Built in 6 weeks as Team MVP.

MarketerProductBC Build-A-ThonSep – Oct 2026

← All projects

My role

Marketer on a team of 4 with 2 PMs and 1 Developer.
I ran the customer and business feedback survey, owned the go to market strategy and wrote the pitch content.

Problem

Spas, bike rentals, paddle clubs and yoga studios lose revenue every week to empty off peak slots, with no fast way to reach people who are free right now. Those people don't know what is open near them, so they stay home.

What the survey said

Customers · 25

Price is a barrier72%

Already look for discounts65%

Would use the app for deals75%

Businesses · 18

Worry about off peak demand78%

Would offer off peak discounts60%

Would use a platform to fill slots70%

Two customer problems were stacked. Discovery came first: people didn't know what was open nearby. Price is what turned interest into action. One feed solves the first, and the discount solves the second.

The pivot

Our Week 2 brief named business owners as the primary user.
Before Demo Day we changed it: a business will post deals all day,
but if no one books, the loop never starts.

Aspect	Week 2 brief	Demo Day
Primary user	Business owners filling quiet windows	Customers looking for something to do now
Core action	Business posts a time limited deal	Customer discovers and books a nearby deal
Payment	Pay at venue	Pay in app, free cancellation until booking closes

Go to market

Start with businessesVisit and DM local businesses. Posting is always free.
Bring customers on InstagramDaily Reels like "3 deals tonight", aiming for 100 sign ups in 2 weeks
on a $30 budget.
Earn on every booking10% of each booking, so businesses pay nothing until a customer shows up.

Results

Prototype demo · 46 seconds, no sound

A working prototype with browse, filter, book, pay and QR code flow, demoed live on Demo Day. Deals were seeded for Vancouver and payment was simulated; real listings and Stripe are the next steps.

The deck

Scroll sideways · 28 slides

View the pitch deck · PDF

What I'd do differently

Validate the primary user before writing the design brief, not after. Next time I would run the two sided survey first and let it pick the side that has to move first.

Operational Waste Analysis

Is the cost cut actually saving the money?

Business AnalystPlatform 7 CaféVer 1 and Ver 2

← All projects

−64%

swing shift hours, 199 to 71.5

−24.1%

revenue in the 12 to 1 PM slot

−$4,601

estimated net per month

Overview

The café cut swing shift hours to reduce labor cost.
I wanted to know whether the cut really saved money.

· Role Solo analyst, data collector and presenter
· Tools SQLite, Excel
· Data Oct 2025 to Feb 2026

The data problem

Detailed transactions and supplier invoices weren't available, a POS migration had erased everything before October 2025, and shift and waste logs existed only on paper. Sales were monthly while inventory and shifts were daily, so measuring waste day by day was impossible.

Process

Rebuild the dataDigitized handwritten POS, inventory and shift logs into SQLite, with no imputation for missing points.
Pivot to proxy KPIsVer 1 used milk per bill, revenue per labor hour and coffee inventory intensity to triangulate efficiency from monthly totals.
Widen the windowVer 2 added 5 months of data, 16 sales categories and hourly sales slots.
Compare the decisionsSet the swing cut that was made against an alternative that never was: calibrating espresso bean orders.

What the data said

Revenue fell because fewer people came in, not because they spent less. Average bill held steady.
Espresso sales fell 20% while orders of the main espresso bean rose about 40%.
Lunch revenue fell 13 to 24%, two to three times the morning rate, in exactly the window swing coverage served.

Results

Revenue impactWage saving Swing cut, actual
net −$4,601 a month

−$5,171

+$570

Bean order calibration
net +$270 a month

$0

+$270

Monthly figures from the Ver 2 analysis, bars on one scale.

Option	Wage saving	Revenue impact	Net per month
Swing cut, actual	+$570	−$5,171 lunch decline	−$4,601
Bean order calibration	+$270	$0	+$270

The saved wages are dwarfed by the revenue decline that came with them. Monthly data can't prove causation, but the direction is clear: this is a demand problem, not a cost problem.

Restore swing at lunch, the highest revenue window
Tie bean orders to the prior week's espresso volume
Focus on bringing customers back instead of further labor cuts
Build basic data habits: monthly POS exports, weekly inventory, shift logs

What I learned

Acknowledging the limits of your data is not a weakness. It is the foundation of credible analysis.

Calm in the complexitySpark in every insight

Keep your center. Bring your own point of view.

My grounded side

My spark

My working style

Hi, I am Chanbyeol Luna Choi

How I work

Find the signal

Know the audience first

Make it land

Toolkit

Education

Where I've built career

Marketing Specialist / ProductBC

Marketing Analyst / PhoneBox

Business Analyst / Platform 7 Coffee

Data Annotator / Turing

PTKOREA

NPS Data Analysis Assistant

Samsung Unpacked Global Website Communication Assistant

Marketing Intern / Emons

How the results got made

VANORI

TooGoodForFun

Platform 7

VANORI

My role

Problem

The audit

The design

Results

The deck

What I learned

TooGoodForFun

My role

Problem

What the survey said

Customers · 25

Businesses · 18

The pivot

Go to market

Results

The deck

What I'd do differently

Operational Waste Analysis

Overview

The data problem

Process

What the data said

Results

What I learned

Calm in the complexity
Spark in every insight