Building a Pan-India Real Estate Database: 1.27 Lakh Verified RERA Projects in 11 Weeks

Introduction

How a Series A proptech platform unified RERA data across 26 state portals — and turned it into its core moat

Actowiz Solutions  |  Case Study  |  Industry: Proptech / Real Estate

Client Snapshot

The client is a Bengaluru-headquartered proptech platform serving home buyers, investors, and channel partners across Indian metros. After closing its Series A funding, the company set an ambitious goal: become the most authoritative residential real estate database in India by anchoring every listing, project, and developer profile in RERA-verified data.

The client and exact platform name are anonymized.

The Business Challenge

The platform's existing project database was a stitched-together mix of broker submissions, builder uploads, and manual research. Three structural problems were blocking growth:

• Listing trust scores were inconsistent. Users frequently encountered projects with wrong carpet areas, outdated completion dates, or misattributed builders.
• The team had no automated way to verify whether a listed project was RERA-registered, and if so, whether the registration was active, expired, or under complaint.
• Sales and content operations spent more than 1,400 person-hours per month on manual RERA lookups across state portals — a cost line growing 9 to 12 percent month-on-month.

"We did not just want RERA scraping. We wanted RERA as a system of record that we could bet our product on." — CTO

Solution Architecture

1. State-specific extraction modules

Actowiz built 26 dedicated extraction modules, one per state RERA portal. Each module handled the portal's specific search flow, captcha behavior, pagination logic, and detail page schema. A shared scheduling layer coordinated polling cadence per state based on update frequency and portal stability.

2. Unified schema with state-specific extensions

Captured records were normalized into a single canonical schema covering 84 common fields, with a state_extensions JSON field preserving fields unique to specific portals such as Karnataka's plot-level data or Maharashtra's tower-level configurations.

3. PDF OCR and document indexing

Sanction plans, audit reports, and quarterly progress filings were downloaded, OCRed, and indexed. Numeric fields such as total project cost, construction expenditure incurred, and units booked were parsed out of free-form PDFs and pushed into the structured record.

4. Builder graph

A promoter resolution layer linked subsidiary companies to parent groups using PAN, director overlap, and address heuristics. This let the client display "projects by builder group" instead of fragmented company-by-company views.

5. Delivery to client systems

Two delivery channels operated in parallel: a low-latency REST API for user-facing project lookup, and a nightly PostgreSQL replication job that mirrored the full Actowiz RERA database into the client's data warehouse for analytics and ML use cases.

Results

• 1,27,800 active RERA projects loaded and continuously maintained across 26 states within 11 weeks of kickoff.
• Listing trust score on the client platform improved measurably. The internal "verified-listing" tag, computed against the Actowiz RERA dataset, lifted user-reported listing accuracy from 71 percent to 94 percent.
• Manual RERA lookup work in content and sales operations dropped by 87 percent, freeing more than 1,200 hours of monthly capacity for higher-value research and curation tasks.
• Search conversion to enquiry rose 18 percent after "RERA verified" badges and full project disclosures were added to listing pages.
• The client launched two new products on top of the RERA dataset within six months: a builder-comparison tool and a project-delay risk score, both differentiators no competitor could match without comparable data infrastructure.

"RERA data is now the spine of our product. Actowiz delivered what an in-house team would have taken 18 months to build." — Head of Data

Why It Worked

• State-by-state engineering, not a one-size script. Each portal got the bespoke handling it deserved, including separate retry strategies and captcha approaches.
• Schema unification with extension fields. The client team queried 84 normalized fields without losing state-specific richness.
• Operationalized integration. The dataset replicated nightly into the client warehouse meant analysts and ML engineers used it without friction. Adoption is what made the investment pay off.
• Continuous refresh discipline. Daily polling on new registrations and complaints kept the database from going stale, which would have re-introduced the original trust problem.