Executive Summary

The Idea

A lot of people want to open a restaurant. They cook well, their friends tell them “you should open a place.” But almost nobody does, because it’s terrifying. The rent, the fit-out, the licenses, the competition. Half a million dollars on the line, and all you have is a gut feeling and a broker who wants his commission.

On the other end, there are people who are completely delusional. They’ll sign a lease on a street that already has twelve noodle shops and open a thirteenth, convinced theirs will be different. It won’t.

Both problems are the same problem: nobody quantifies the risk before taking it.

Large restaurant chains do this. They have location intelligence teams, spatial analysts, demographic models. They know exactly what foot traffic looks like at 2pm on a Tuesday before they sign anything. Independent operators don’t have access to any of that.

Until now. The Hong Kong government publishes 3,712 datasets for free. Population, income, transport, restaurant licenses, rental indices, crime stats, live parking data. Everything you need to build those same models is sitting on data.gov.hk, waiting.

Four Layers

1

Open Data

3,712 government datasets: who lives here, who works here, who passes through, what’s already open

→

2

Your Concept

What you’re opening, at what price, for whom. Same address, different concept = completely different math

→

3

Math Models

10 models from academic literature: Huff, Gravity, Catchment, Regression. Textbook formulas, not black boxes

→

4

AI Agents

Synthetic personas that simulate real customer decisions: would they walk in, or keep walking?

The key insight: your concept changes everything. A ramen bar and a cha chaan teng on the same block attract completely different people at different times. The open data tells us who’s there. The models tell us how they behave. The AI agents tell us whether they’d choose you.

Tip

We can approximate what percentage of office workers, tourists, residents, and commuters pass through any location at any given hour. That’s the foundation: simulating real foot traffic with real demographic data, not guesswork.

Two Directions

Direction 1: “I have a concept. Where should I open?” You know you want to open a pasta place. We run the models against addresses you’re considering and tell you which one gives you the best shot.

Direction 2: “I have an address. What should I open?” You found a great space with good rent. We run the models in reverse: given the demographics, foot traffic, and competition around this address, what type of restaurant has the highest probability of surviving?

Both directions use the same data, the same models, the same agents. Just different starting points.

How to Read This

Each page covers one concept, cross-linked to everything else. Start anywhere:

• Open Data — What free government data we use and how
• Your Concept — How your business type changes every formula
• Models — The mathematical models, one per page, with full formulas
• AI Agents — How synthetic personas stress-test the location
• Case Studies — How the full pipeline works end to end