Yelp Nearby Search (Location-Based Service)¶

Quick Reference Guide for System Design Interviews

Problem Statement¶

Design a location-based service like Yelp that allows users to search for nearby businesses (restaurants, shops, etc.), view details, and read/write reviews.

Requirements¶

Functional Requirements¶

Search businesses by location and category
View business details and reviews
Add/edit business listings
Write and read reviews
Search with filters (rating, price, hours)

Non-Functional Requirements¶

Latency: < 200ms for search
Availability: 99.99%
Scale: 500M users, 200M businesses
Accuracy: Results within specified radius

Back of Envelope Estimation¶

High-Level Architecture¶

Geospatial Indexing¶

Approach 1: Geohash¶

Approach 2: QuadTree¶

Approach 3: PostGIS / Spatial Database¶

Search Flow¶

Data Models¶

-- Businesses
CREATE TABLE businesses (
    business_id     UUID PRIMARY KEY,
    name            VARCHAR(255) NOT NULL,
    address         TEXT,
    city            VARCHAR(100),
    state           VARCHAR(50),
    zip_code        VARCHAR(20),
    country         VARCHAR(50),
    latitude        DECIMAL(10, 7),
    longitude       DECIMAL(10, 7),
    geohash         VARCHAR(12),  -- For index
    category        VARCHAR(100),
    price_range     INT,  -- 1-4 ($-$$$$)
    rating          DECIMAL(2, 1),
    review_count    INT DEFAULT 0,
    is_open         BOOLEAN DEFAULT TRUE,
    created_at      TIMESTAMP,
    updated_at      TIMESTAMP,

    INDEX idx_geohash (geohash),
    INDEX idx_category (category),
    INDEX idx_rating (rating)
);

-- Business hours
CREATE TABLE business_hours (
    business_id     UUID REFERENCES businesses(business_id),
    day_of_week     INT,  -- 0-6
    open_time       TIME,
    close_time      TIME,

    PRIMARY KEY (business_id, day_of_week)
);

-- Reviews
CREATE TABLE reviews (
    review_id       UUID PRIMARY KEY,
    business_id     UUID REFERENCES businesses(business_id),
    user_id         UUID NOT NULL,
    rating          INT CHECK (rating BETWEEN 1 AND 5),
    content         TEXT,
    photos          TEXT[],
    useful_count    INT DEFAULT 0,
    created_at      TIMESTAMP,

    INDEX idx_business (business_id, created_at DESC),
    INDEX idx_user (user_id)
);

Caching Strategy¶

Read/Write Paths¶

Interview Discussion Points¶

How do you index 200M businesses for geo search?
Geohash for simple queries
Redis GEORADIUS for in-memory speed
PostGIS for complex spatial queries
How do you combine location + text search?
Two-phase: geo filter → text filter
Or Elasticsearch with geo queries
How do you rank search results?
Distance, rating, review count
Relevance to search terms
Paid placement for businesses
How do you handle updates?
Write to primary DB
Async update search indexes
Invalidate caches
How do you scale read traffic?
Geospatial index in memory (Redis)
Read replicas for DB
Aggressive caching
How do you handle user location changes?
Client-side debounce
Search result caching by geohash
Smooth transitions in UI