PropstackX: Building a Real Estate Marketing Automation CRM from the Ground Up

The Real Estate Sales Problem

Real estate sales is a high-value, long-cycle, relationship-driven business — and most developers in India were running it like a 2010 contact centre.

The typical sales operation before PropstackX:

• Leads arriving from multiple channels (99acres, MagicBricks, Housing.com, walk-ins, referrals, digital ads) landing in different spreadsheets managed by different team members
• No unified lead record: the same buyer enquiring from multiple sources became three separate "leads" assigned to three different agents
• Follow-ups tracked via sticky notes, calendar reminders, and WhatsApp forwards between agents
• Campaign blasts sent manually — the same message copy-pasted to 5,000 contacts in batches on WhatsApp Web, with no targeting, no personalization, no tracking, and constant risk of number bans
• Zero funnel visibility: no way to see how many leads were in active negotiation, how many site visits had happened this week, or which campaigns were actually generating bookings

The business had inventory, marketing budget, and sales headcount. What it lacked was a data layer to orchestrate them.

This is our product. We designed and built every module from scratch.

What We Built

PropstackX is a multi-tenant marketing CRM and campaign automation platform for real estate developers and brokers.

Lead Management

• Unified lead record with source attribution (portal, referral, direct, campaign, walk-in)
• Pipeline with configurable stages: New → Contacted → Site Visit Scheduled → Site Visited → Negotiation → Booked
• Rule-based and round-robin agent assignment
• Full activity timeline per lead: every call, message, visit note, and status change with timestamp and actor ID

Campaign Engine

• Audience builder: filter leads by any combination of attributes — pipeline stage, budget range, property type interest, source, last activity date, opt-in status
• Campaign types: WhatsApp Business broadcast, email, SMS
• Template management with variable substitution (Hi {{first_name}}, the {{property_name}} unit you enquired about...)
• Throttled delivery: campaign messages queued and delivered at Meta API-compliant rates
• Delivery tracking via webhook callbacks: sent → delivered → read per recipient
• Unsubscribe handling: opt-out keyword triggers opted_out: true flag with cross-campaign suppression

Flow Automation

• Event-triggered sequences: "When a lead fills the contact form → send welcome WhatsApp immediately → wait 24h → if no response, assign to senior agent → wait 48h → if still no response, enrol in re-engagement campaign"
• Visual flow builder with branch conditions
• Flow analytics: drop-off rate per step, time-in-step distribution, conversion rate per flow variant

WhatsApp Integration

• Meta Cloud API for all WhatsApp Business messaging (official channel — no third-party bulk senders)
• Inbound message routing: customer replies land in the assigned agent's inbox
• Pre-approved template management with variable mapping

Technical Architecture

Engineering standards enforced throughout:

• Soft deletes on all entities — every record has a deletedAt timestamp; nothing is hard-deleted. Data is always recoverable.
• Full audit trail — createdAt, updatedAt, createdBy, updatedBy on every table. Who changed what, and when, is always queryable.
• CorrelationId on every request — the request ID flows through all downstream service calls and appears in every log line. Any production incident is traceable in under 10 minutes.
• Tenant isolation at query layer — tenant ID is baked into every service query. Cross-tenant data access is structurally impossible.

Campaign Engine: The Data Architecture Detail

The campaign engine received the most care in design because it handles state, compliance, and delivery at scale.

Audience resolution is evaluated at execution time, not at campaign creation time. A campaign targeting "leads interested in 3BHK, budget ₹80–120L, last contacted > 7 days ago" is evaluated against the live database when the campaign fires — so it always reflects current data, not a stale snapshot from when the campaign was set up.

Delivery compliance is structural. Every outbound message uses a pre-approved WhatsApp Business template. The campaign service validates template status before execution and refuses to send non-approved templates. This is not a UI warning — it is a hard block at the service layer.

Deduplication: before delivery, each recipient list is checked against the opted_out flag and against in-flight campaigns for the same recipient (rate limiting per contact to prevent spam complaints).

Delivery tracking: each message gets a unique messageId. Meta webhook callbacks update delivery status in real time. Campaign performance reports show message-level delivery funnels, not just aggregate sent counts.

Results

Before PropstackX, a developer running a campaign manually extracted a list, pasted numbers into WhatsApp Web, and sent messages one by one — no targeting, no tracking, constant compliance risk.

After PropstackX:

• 4.2x more qualified lead touchpoints per campaign — because campaigns now target warm leads (active in the last 30 days, matching property criteria) rather than the full database
• Lead response time dropped 78% — flow automation fires within minutes of lead creation vs agents picking up the queue manually the next morning
• Full funnel visibility for the first time — developers see exactly how many leads are at each pipeline stage and which campaigns drove site visits and bookings
• 100% WhatsApp compliance — all messages sent via official Meta Cloud API with pre-approved templates, eliminating the risk of business number bans

What This Demonstrates

PropstackX is a product we built and operate under real production load, with real clients, real data, and real consequences when something breaks.

The design decisions — campaign engine architecture with audience-at-execution-time resolution, compliance enforcement at the service layer, soft deletes and audit trails by default, multi-tenant isolation, flow automation with retry logic — are the same principles we bring to every data and automation engagement.

We don't design for the happy path. We design for the audit, the incident, and the scale.