No general advantage detected.
None of the five industries reached significance in the primary hotspot comparison.
SEOUL BUSINESS GUIDE / 2025
Find the fit.
Not just the fame.
A better location decision starts with your industry. Explore real district benchmarks and discover what Seoul’s hotspot reputation can — and cannot — tell you.
165 sampled districts5 industries2025 Q1–Q4
THE SEOUL STUDYGROUP 25
33Hotspot districts
132Non-hotspot districts
532District × industry units
A city’s reputation is only the beginning.
Look at the store-level evidence.
THE KEY DISTINCTION
Total district sales ≠ sales per store
A busy district can contain more businesses.
That does not guarantee more revenue for each store.
WHAT THE RESEARCH REVEALS
Three findings.
A better starting point.
One industry stands out.
Coffee & Beverage shows a positive signal in robustness tests. Its primary comparison remains uncertain.
Crowds are a weak predictor.
Total floating population explains very little of the variation in revenue per store.
EXPLORE THE STUDY AREAS
Start with a neighborhood.
Coffee & Beverage lens
Actual sample averages · million KRW / quarter
Six purposively selected hotspot zones. Figures reflect sampled districts, not every store or every district within each neighborhood.
FROM RESEARCH TO YOUR BUSINESS
Make your next move.
01 / EXPLORE
Find your district.
Choose an industry and open actual district-level observations.
District explorer ↗02 / UNDERSTANDQuestion the assumption.
Compare industries and see where the evidence is strong or uncertain.
Research insights ↗03 / PLANBring in your costs.
Explore break-even requirements using your own assumptions.
Business scenario ↗ SEOUL BUSINESS GUIDE 2025 Q1–Q4
SEOUL DISTRICT ATLAS
A quieter map. A clearer view.
Scroll to zoom. Drag to move. Click a district to explore.
Wheel to zoom · Drag to move
Major roadsRail / subway▤Station⌂Landmarks & parks
Zoom in for nearby stations and more road detail. Click a facility icon to identify it.
165 / 165 research boundaries linkedReport a map issue ↗
Map sources & dates
Graphic map with official Seoul borough outlines and OpenStreetMap rivers, major roads, railways and selected facilities. Minor streets and building footprints are omitted. Commercial boundaries: 23 Oct 2023; boroughs: 31 Oct 2023; orientation snapshot: 11 Oct 2026; research: 2025 Q1–Q4. All 165 sampled district codes and names match. Gray areas have no selected-industry data or are outside the sample. Rail lines show track alignments, including underground sections; this is not a transit route planner. All map layers are saved snapshots. Seoul Open Data Plaza ↗ · Borough source ↗ · Graphic map data ↗ · Orientation map data ↗ · Source details ↗
SALES PER STORE
Million KRW · average quarterMean store count
Quarterly floating population
GROUP BENCHMARKS
BROWSE THE OBSERVATIONS
Districts in your selection.
Historical estimates, not opportunity rankings. Click a district to update the overview and inspect its quarterly data.
THE QUESTION WE TESTED
Do hotspot stores earn more?
Million KRW / store / average quarter
HotspotNon-hotspot
View the selected industry's test details
Reported Group 25 results · α = 0.05 · all five arithmetic mean-difference 95% intervals include zero. Association does not establish causation.
THE COFFEE EXCEPTION
A signal.
Not a guarantee.
The raw-scale difference is uncertain. A log-scale comparison suggests a positive association, showing why the test and measurement scale matter.
ARITHMETIC MEAN DIFFERENCE+₩7.38M
95% CI: −₩2.37M to +₩17.13M
Primary p = 0.064
GEOMETRIC MEAN RATIO1.57×
95% CI: 1.105 to 2.217
Log-Welch p = 0.006 · Holm p = 0.031
THE FOOT TRAFFIC QUESTION
More people.
Very little explanation.
A group difference and a useful prediction
are two different questions.
COFFEE MODEL SANDBOX
Move the traffic.
Million KRW per store per average quarter
Fitted values only. Hotspot R² = 0.0195; non-hotspot R² ≈ 0. These weak models cannot reliably forecast a new store.
Model details & uncertainty
ŶH = 31.775 + 0.00238X; ŶN = 27.772 − 0.00014X. X is in 1,000 persons; output is M KRW per average quarter. Slider stays within the overlapping observed group range. At 1 million, reported 95% prediction intervals extend below zero (−16.94 to 85.26M and −20.76 to 76.01M), indicating an unsuitable individual-outcome model. Negative bounds are not realistic revenues.
Exploratory Screening of Foot-Traffic Predictors
25 predictors · 2 specifications · 250 models
Linear and log-log models; 50 candidate models per industry. The best screened R² is 7.28% (Japanese: people in their 40s). No slope remains significant after Bonferroni adjustment for 50 candidates per industry. These are exploratory associations, not validated forecasts.
| Business industry | Best screened R² | Raw p-value |
|---|
WHAT THIS MEANS FOR YOUR BUSINESS
Choose with evidence.
Then check the costs.
A famous location does not guarantee stronger store performance.
ONE PRACTICAL QUESTION
What would you need to break even?
Enter your monthly costs, or try an illustrative demo.
Enter two assumptions to calculate.
User assumptions · no measured rent data. This is a break-even calculation, not a sales or profit forecast.
BEFORE YOU COMMITVisit the location.
Visit the location.
Test the business model.
Our data help you compare revenue benchmarks. Rent, customer conversion, store size, brand positioning and opening hours still need local investigation. Use the evidence to ask better questions, not to choose a location automatically.
Choose your district & industry.
Open District explorer. Select a study zone and industry, then drag or zoom the map and click an official district boundary. Choose “All districts” to browse the full sample. Search by English or Korean name; open quarterly detail from the map or any district card.
On the map: scroll to zoom around your pointer, drag to pan, and double-click to zoom in. Use +/− buttons or pinch on touch devices. Gold lines are major roads; dashed blue-gray lines are railways and subway tracks. Facility icons mark stations, parks, universities and landmarks. Zoom in for more detail; click an icon for its name and source. Turn “Roads, rail & landmarks” off for a simpler view.
Read the right metric.
Open district explorer →Sales per store is estimated industry sales divided by store count. The headline number is an average quarter, in million KRW.
Compare the evidence.
Open Research insights. Change the industry lens to compare hotspot and non-hotspot means. Expand test details for p-values and confidence intervals; try the coffee traffic slider to see the fitted values.
A higher mean is not a guarantee.
Open research insights →All five primary tests remain non-significant at α = 0.05. Coffee has mixed evidence across tests, and traffic models explain little variation.
Check the costs.
Open Business scenario. Enter total monthly fixed costs in million KRW and variable costs as a percentage of revenue. The calculator shows the monthly revenue needed to break even.
Try a clear example.
Open business scenario →Click “Try demo costs”: fixed costs of ₩5M and a 40% variable cost ratio require about ₩8.33M monthly revenue. These costs are illustrative assumptions.
PRESENT WITH CONFIDENCE
A few useful controls.
Guided demo
“Start demo” walks through three focused screens. Use Back / Next or the ← / → keys. Exit demo returns to the full website.
English / 한국어
Use the language buttons in the header. Your selections and inputs stay in place. Korean labels retain English research terms, and your language preference is remembered in this browser.
Full screen & sources
Use ⛶ to request full screen where supported. Open Sources & study notes for the sample, units, original dataset and research limits.
KNOW THE TERMS
The language of the study.
- District
- A sampled commercial area. A study zone such as Seongsu contains several districts.
- Hotspot / Non-hotspot
- The study’s district groups: 33 hotspot districts and 132 stratified non-hotspot districts. These are sample classifications.
- Sales per store
- Estimated industry sales ÷ store count, averaged across available 2025 quarters. Revenue is different from profit.
- Floating population
- Quarterly foot traffic in the district. Visitors are not the same as paying customers.
- Confidence interval (CI) & p-value
- The interval shows uncertainty in an estimated difference. A p-value below 0.05 meets this study’s significance threshold; it does not prove causation.
- R² & Break-even revenue
- R² measures explained variation. Break-even revenue = fixed costs ÷ (1 − variable cost ratio); its unit here is monthly revenue.
Before you interpret a number.
Does this site recommend the best district?
It provides historical comparisons to support your judgment. It does not rank opportunities or guarantee future results. Investigate rent, customer fit and competition locally.
Why do quarterly sales and monthly costs use different periods?
The research dataset records quarters. The calculator uses your monthly assumptions independently; it does not convert historical quarterly sales into a monthly forecast.
Why are some districts missing quarters?
Eighteen district–industry observations have fewer than four available quarters. Their averages use available records only. Open the quarterly detail to check coverage; missing quarters are not zero.
Sources & study notes
The sample
165 districts: 33 hotspot and 132 stratified non-hotspot. 532 district × industry averages from 2,092 quarterly records. Eighteen observations use fewer than four available quarters. Six purposively selected zones: Seongsu, Hongdae, Itaewon, Jamsil, Myeongdong and Sinchon–Ewha.
What the numbers mean
Sales per store = estimated industry sales ÷ store count, averaged over available 2025 quarters. Group bars are historical averages across district–industry observations. They are not forecasts for a new business or observed monthly revenue.
Sources
Group 25 — Prediction Model (Simple Linear Regression), supplied project document.
Seoul official commercial district boundaries ↗
Map: © OpenStreetMap contributors · boundaries: Seoul Metropolitan Government, KOGL Type 1.
G25_merged original dataset ↗
ACDT Week 4 course material ↗
Limits
Revenue is not profit. Data are aggregated and observational, with omitted rent, size, brand and operating-hour factors. Non-significance does not prove equality. Reported directional ANOVA p-values and robustness results are preserved. English district labels are reading aids; original Korean names remain available.