AI Lead Scoring & CRM Automation: Building a $6M Pipeline Engine for a Multi-City Broker Network

The Problem Was Not the Leads. It Was What Happened to Them.

Our client — a real estate broker network operating across five cities in India — was spending $10,000 per month on lead generation through portals like MagicBricks, 99acres, and Facebook Ads. They were generating 3,000+ leads per month. Their closure rate was 0.6%.

The root cause was not poor lead quality. It was zero prioritisation. Every lead — whether a casual browser saving listings for later or a buyer already approved for a home loan — hit the same queue. Agents called in FIFO order. The serious buyers grew impatient. The tire-kickers consumed hours.

When we audited their pipeline, we found that 22% of their leads were responsible for 91% of their closed deals. The other 78% were burning agent time and payroll. The fix was not more leads. It was knowing which ones were worth calling first — before a human ever picked up the phone.

The Challenges

• Lead Fragmentation: Leads were arriving across six channels — MagicBricks, 99acres, Facebook Ads, Google Ads, the company website, and direct WhatsApp messages — and being manually copy-pasted into spreadsheets by a dedicated admin. Leads were falling through the cracks daily.
• Zero Behavioural Data: The business had no visibility into what a lead had done before calling. Had they viewed the property three times? Searched for the same locality for six months? Switched from renting to buying filters? This intent data existed but was completely unused.
• Unequal Agent Load: Top-performing agents were getting the same volume of unscored leads as junior agents. There was no intelligent routing — no mechanism to send high-intent leads to senior closers and nurture-stage leads to junior team members.
• No Follow-Up Automation: The team was sending manual follow-up messages via personal WhatsApp accounts with no tracking, no templates, and no escalation logic when leads went cold.

The Solution: Custom CRM + AI Scoring + WhatsApp Automation

We did not recommend an off-the-shelf CRM. Every major CRM on the market — Salesforce, HubSpot, Zoho — is built for a generalised B2B sales motion. Real estate has its own data model: listings, localities, RERA compliance, site visit scheduling, loan eligibility, and a buyer journey that plays out over weeks or months. We built for that specific context.

• Unified Lead Ingestion Pipeline: We built API connectors for every channel the client used — MagicBricks Common Lead API, IndiaMART, Facebook Lead Ads, Google Ads webhook, and a WhatsApp Business API integration. Every lead, regardless of source, lands in one structured database within 30 seconds of submission.
• AI Intent Scoring Engine: We trained a machine learning classification model on 18 months of their historical data — which leads closed, how many touchpoints they had, what localities they searched, how many times they viewed listings, and whether they had engaged with EMI calculators or loan enquiry forms. The model assigns every new lead a score from 1-100 with a predicted closing probability. Leads above 65 are flagged as Hot and routed to senior agents immediately.
• Intelligent Agent Routing: Hot leads go to senior agents during business hours. Warm leads enter a 5-step WhatsApp nurture sequence. Cold leads get a monthly digest of new listings in their saved locality. Each tier gets an experience matched to where they actually are in their buying journey.
• WhatsApp Automation Sequences: We deployed a WhatsApp Business API-powered follow-up system with 12 pre-built templates — initial response, site visit confirmation, property shortlist, EMI estimate, and deal closure nudge. Every message is personalised with the lead's name, preferred locality, and budget range from the CRM data.

Lead Routing Flowchart: From Inquiry to Closed Deal

The diagram below shows the complete lead journey through our system. The critical moment is Stage 3 — the three-way split based on AI score — because this is where most real estate businesses haemorrhage agent time. By routing leads intelligently before a human is ever involved, senior agents spend their day exclusively on conversations that can close.

The Impact: Measurable Outcomes