Real Estate Production System

Coldwell Banker Indonesia

Unified CRM & Listing Management Platform for Real Estate Operations

15+

Active Agents

99.5%

System Uptime

3→1

Systems Consolidated

3 months

Timeline

Executive Summary

We built a unified CRM and property listing platform that consolidated operations for 15+ agents into a single role-based system. The platform replaced disconnected spreadsheets, CMS, and CRM tools with a centralized solution — separating write-heavy admin workflows from read-heavy public traffic, and achieving 99.5% production uptime for a brokerage handling commercial and residential properties across Indonesia.

The Problem

Coldwell Banker Indonesia operated with 15+ active agents across Greater Jakarta using a fragmented toolset — spreadsheets, a separate CMS, and an external CRM. Consolidating these into a single system required handling a highly complex property data model with dozens of structured and semi-structured fields, media-heavy workflows (images, videos, PDFs, automated watermarking), and advanced search filtering across multiple dimensions with low latency requirements. Data normalization needed to be balanced against query performance under read-dominant workloads, all delivered within a 3-month timeline with evolving requirements.

15+

Active Agents

Fragmented Tools

3 months

Timeline

Services Delivered

Dedicated Team

Unified CRM & Listing Management Platform for Real Estate Operations

Architecture Overview

Data Layer

PostgreSQL Redis

Backend & Orchestration

Laravel Inertia.js

Frontend

Next.js (SSG)

Infrastructure

DigitalOcean Docker

Key Technical Decisions

System Design

The system separates write-heavy admin operations from read-heavy public traffic, optimizing each independently. The backend uses Laravel with a repository pattern for maintainable domain logic. The admin interface is built with Inertia.js (React), enabling rapid full-stack development without a separate API layer. The public-facing property website uses Next.js with static site generation for fast, SEO-optimized listing pages. PostgreSQL handles relational data with targeted indexing, while Redis provides application-level caching and shared state across requests.

Key Decisions

Inertia.js was chosen for the admin interface to balance development velocity with modern UX, avoiding the overhead of a standalone SPA architecture. A known Inertia limitation around global state was mitigated by using Redis as a shared state layer. Next.js SSG was selected for the public website for performance and SEO, though deploying outside Vercel required manual optimization. The most significant architectural decision was denormalizing search-specific fields: accepting write complexity in exchange for low-latency multi-dimension filtering across property listings.

Implementation Highlights

Multi-layer caching (application-level + Redis) reduced backend load for high-traffic listing endpoints. Denormalized search tables support complex filtering across location, price, type, and facilities without expensive joins. Background job queues handle email notifications, report exports, and media processing — including automated image watermark generation. The database indexing strategy was designed specifically around property search query patterns. The admin interface was built mobile-first for agent usability in the field on phones and tablets.

Results & Validation

Achieved 99.5% uptime in production across the full 15+ agent deployment.

Consolidated three fragmented systems — spreadsheets, CMS, and external CRM — into a single operational platform.

Significantly improved agent efficiency through centralized property listings, client data, and media workflows.

Reduced backend infrastructure requirements through effective multi-layer caching strategy.

Delivered a fully operational system within a 3-month timeline despite evolving requirements.

Key Insights

Search performance in real estate systems is primarily a data modeling problem — precomputed, denormalized search fields outperform even heavily optimized relational queries for multi-dimension filtering.

Inertia.js accelerates admin development significantly but requires an external shared state layer (Redis) when global state is needed across requests.

Separating write-heavy and read-heavy traffic at the architecture level allows each layer to be optimized independently, avoiding the compromise of a single unified data path.

Background job queues are essential for media-heavy workflows — synchronous processing of images, watermarks, and PDFs blocks user-facing operations at scale.

Deploying Next.js outside Vercel is viable with proper static generation and CDN configuration, but requires explicit optimization work that would otherwise be handled by the platform.

Who This Applies To

This solution is applicable to real estate brokerages managing high volumes of listings and agents, businesses with fragmented operational systems needing consolidation, and platforms requiring fast, search-heavy user experiences with cost-efficient infrastructure. It is particularly relevant for organizations balancing internal workflow complexity with public-facing performance requirements.

Real Estate CRM & Operations Search-Heavy Systems Media Workflows System Consolidation

Technologies Used

Backend

Laravel Inertia.js

Frontend

Next.js

Infrastructure

DigitalOcean Docker

Data & Integrations

PostgreSQL Redis

Patterns & Techniques

Repository Pattern SSG Denormalization Background Jobs Role-Based Access Image Watermarking

Tools

GitHub

Building something similar?

We specialize in dedicated team for real estate companies. If you're facing challenges like the ones we solved for Coldwell Banker Indonesia, let's talk.

Book a Call View All Case Studies