Multi Tenant CRM SaaS for UK Real Estate Agencies

The Challenge

A UK proptech startup wanted to build a CRM specifically for independent estate agencies. Existing tools like Reapit, Alto, and Jupix were expensive, rigid, and designed for large chains. Smaller agencies with 2 to 15 agents needed something modern, affordable, and built around how they actually worked.

Lead leakage was the core problem. Agencies received enquiries from Rightmove, Zoopla, OnTheMarket, their own websites, phone calls, and walk-ins. Without a unified system, leads fell through cracks. Agents estimated they followed up on only 60% of inbound enquiries within 24 hours. The rest went cold.

Manual property matching wasted hours. When a new buyer registered their requirements, agents mentally scanned their listings or used basic keyword searches to find matches. There was no automated system that could match buyer preferences (budget, bedrooms, commute distance, school catchment) against live listings and proactively notify agents of strong matches.

No data driven prioritisation. Agents treated every lead equally, spending the same time on a casual browser as on a mortgage-approved buyer ready to move within 30 days. Without lead scoring, the highest-value opportunities often received the same (or less) attention as low-intent enquiries.

The Approach

RG INSYS designed a multi tenant SaaS platform where each agency operated in an isolated environment with custom branding, their own user roles, and independent billing. The architecture was built to onboard 100+ agencies without infrastructure changes.

Unified lead inbox: We built integrations with Rightmove, Zoopla, and OnTheMarket APIs to automatically ingest enquiries. Website lead forms, phone call logs (via Twilio), and manual entries all fed into a single timeline per contact. Duplicate detection merged records from multiple sources into one buyer profile.

AI lead scoring: A scoring model trained on historical conversion data assessed each lead's likelihood of progressing to a viewing and then to an offer. Signals included response time, enquiry specificity, price bracket alignment, mortgage status, and engagement patterns. Agents saw a priority-ranked lead list updated in real time.

Automated property matching: When a buyer's requirements were captured (or updated), the system ran a matching algorithm against all active listings across the agency. Matches factored in budget flexibility, location preferences with commute time calculations (via Google Maps API), and property feature alignment. Strong matches triggered instant agent notifications.

Timeline: Week by Week

Weeks 1 to 2: Discovery with 5 estate agencies. User journey mapping, data model design, multi tenant architecture with schema per tenant in PostgreSQL. CI/CD and infrastructure provisioning.

Weeks 3 to 5: Core CRM: contact management, property listings, viewing scheduling, offer tracking. Rightmove and Zoopla lead ingestion. Unified inbox with duplicate detection. Role based access control.

Weeks 6 to 8: AI lead scoring model (Python, trained on anonymised historical data). Automated property matching engine. Agent dashboards with priority lead lists. Email and SMS notification system (SendGrid, Twilio).

Weeks 9 to 11: White label branding engine (custom colours, logos, domains per agency). Stripe billing integration with tiered pricing. Reporting: pipeline analytics, agent performance, conversion funnels.

Weeks 12 to 14: Beta rollout with 8 pilot agencies. Feedback iteration, performance tuning, security audit. Production launch with onboarding automation.

Tech Stack

• Backend: Node.js 20, Express, TypeScript, Prisma ORM
• Frontend: React 18, TypeScript, Tailwind CSS, React Query
• AI/ML: Python (scikit-learn, XGBoost) for lead scoring, deployed as AWS Lambda microservice
• Database: PostgreSQL 16 (schema per tenant), Redis 7
• Integrations: Rightmove API, Zoopla API, OnTheMarket, Twilio (calls/SMS), SendGrid (email), Google Maps API
• Infrastructure: AWS (ECS Fargate, RDS, ElastiCache, S3, CloudFront, Lambda)
• Payments: Stripe Billing with tiered subscriptions
• AI tooling: Claude Code, Cursor IDE

Results

Key Features Delivered

• Unified lead inbox: Enquiries from Rightmove, Zoopla, OnTheMarket, website forms, and phone calls merged into a single contact timeline. Automatic duplicate detection and record merging.
• AI lead scoring: ML model scores every lead on likelihood to convert. Agents see a priority-ranked list that updates in real time as new signals arrive (email opens, viewing confirmations, price changes).
• Automated property matching: Buyer requirements matched against live listings using budget, location (with commute calculations), and feature preferences. Strong matches trigger instant push notifications to assigned agents.
• White label multi tenancy: Each agency gets custom branding (logo, colours, domain), isolated data, independent user management, and configurable workflows. New agencies onboard in under 10 minutes.