01 · The market
Delhi NCR alone is large enough to support a category-defining trust portal.
Delhi NCR registers roughly 280,000 residential transactions in a typical year across Gurgaon, Noida, Greater Noida, Faridabad, Ghaziabad and South Delhi. The Gurgaon residential market on its own is conservatively estimated at ₹1.4 lakh crore in annual sale value, with the Golf Course Road and Dwarka Expressway corridors carrying the bulk of the premium ticket sizes. Sector 137 in Noida and Phase 5 in Gurgaon together account for the bulk of secondary-market inventory churn that we observe weekly.
We resist the urge to publish a precise TAM. Our working ranges, deliberately wide: a primary-market broker-fee SAM in NCR of ₹1,800–2,400 crore a year, and a secondary-market lead-gen SAM of ₹400–600 crore at current portal economics. Capture even one percent of that flow at a higher quality bar and the unit economics work out — incumbents hold their gross margins above sixty percent precisely because the supply side has nowhere honest to go.
We are not pretending NCR is the whole opportunity. It is the wedge. The same trust mechanic carries cleanly into Mumbai MMR (deeper, harder), Bengaluru (cleaner brokers, less RERA leverage) and Pune. NCR is where the listing-trust pain is sharpest and where the regulatory tooling — HARERA, UP-RERA, the DLRC registry — gives us the most public-record leverage at zero cost.
Starting inventory · live count
270 listings scored, 79 gold-tier, 27 verified brokers. Seeded from public broker activity in Gurgaon, Noida and South Delhi. Small on purpose — every row was inspected before it shipped.
02 · The problem
Three failure modes the buyer sees, none of which the incumbents fix.
Bait listings. A 4 BHK in DLF The Crest priced at ₹3.2 Cr — twenty percent under any comparable in the building — is bait. The flat is either not for sale, sold months ago, or the broker will admit on the call that the real number is ₹4.5 Cr. You called for the price and the broker now has your number for the next eighteen months. The platform earned a lead. Everyone made money except you.
Refresh theatre. Same flat in Sector 49 Gurgaon, posted in November, refreshed weekly without a price change for five months. Each refresh resets the freshness badge and pushes the listing back to the top of the search. We documented one Greater Kailash listing that had its price tweaked by ₹1.5 lakh six times — every tweak earned it another freshness boost. Our public stale board currently flags every active listing where this pattern is detectable.
Duplicate clusters. One 3 BHK in DLF Phase 5, listed by five different brokers at five different prices ranging from ₹4.7 Cr to ₹5.2 Cr. The buyer thinks they are seeing five flats. They are seeing one flat and five competing brokers — and the platform is invoiced for five listings. We surface the cluster, show the price spread, and let the buyer pick the broker. Methodology page explains the matching.
App store evidence
Across the most recent thousand one-star reviews of 99acres and MagicBricks on the Play Store, the single most common complaint phrase is some variant of "fake listing" or "property already sold". The second is broker harassment after a callback. Both flow from the same root incentive — listing volume rewarded, listing quality not penalised.
03 · Why incumbents won't fix it
Their P&L is structurally hostile to a clean inventory.
Removing fake listings is not a hard engineering problem. The incumbents have larger data teams than we will for a long time. The reason they don't is that the incentives run the other way — and the deeper we read their last few earnings calls, the more obvious it becomes.
-
01
Lead-gen revenue scales with listing count, not listing quality.
The take rate is per lead, not per closed transaction. A bait listing that generates twenty calls is worth more on the platform's books than a real listing that generates two qualified ones. The cleanup roadmap dies in a planning meeting.
-
02
Brokers pay to refresh — refresh-blocking is a revenue cut.
Refreshes are a paid feature dressed up as inventory hygiene. A platform that detects refresh theatre and degrades the listing's ranking is, in the same motion, killing one of its most reliable upsell SKUs. No CFO signs off on that.
-
03
Hiding stale inventory makes the catalogue look bigger.
When we tracked fifty "Just Listed" Gurgaon flats for ninety days, only nineteen were actually new. The other thirty-one were republishes, price-update masquerades and cluster shuffles. The reason the incumbents don't expose this is that their listing-count graphs are the headline metric in every investor deck.
Read the full investigation: "I tracked 50 'just listed' Gurgaon flats for 90 days. 31 weren't new at all." on the Field notes blog.
04 · Our wedge
A 0–100 trust score, computed from seven signals. No black box.
Every listing on Estavera carries a deterministic score out of one hundred, broken down on the listing page itself. The signals and weights are below — they are the same numbers the production scorer uses, and every buyer can audit which signals their listing earned. Tier breakpoints: gold ≥ 80, silver ≥ 60, bronze ≥ 35.
-
25 pts
RERA project match
Cross-checked against the HARERA / UP-RERA project ID. Public record, free to verify.
-
20 pts
Broker reputation tier
Up to twenty for Verified Pro; fourteen for Verified; six for Basic; zero for Unverified.
-
15 pts
Photos verified
At least eight hi-res photos, perceptual hashes checked against reverse-image database.
-
15 pts
Price within ±15% of locality median
Live rolling median, same BHK and listing type. Bait pricing fails this signal automatically.
-
10 pts
Freshness
Real refresh with price change, or a recent price reduction. Refresh theatre earns nothing.
-
10 pts
Phone verified
OTP-verified handset. Filters out the throwaway broker numbers that dominate the bait segment.
-
5 pts
No active duplicate cluster
If the listing matches sibling listings on photo hash, address and BHK, all members of the cluster lose this five — but none get hidden. Buyer picks the broker, not the listing.
05 · Traction proxy & unfair advantages
We are not a "marketplace with AI". We are a property portal that does the diligence the others won't.
-
01
Deterministic scoring, not ML mystique.
Every signal is auditable in apps/trust/scoring.py. A buyer who disagrees with a score can read the file. An ML model with the same job would be both slower to ship and impossible to defend in a regulatory conversation.
-
02
Zero-cost diligence, because the records are public.
RERA project filings are public. Most broker-tier signals are public. The encumbrance certificate is public. We are not paying for a data moat — we are doing the work the incumbents could do and choose not to.
-
03
Local sourcing of inventory, walked street by street.
The seed dataset on this site was hand-collected from brokers in Phase 5, Cyber Hub, Sector 49 Gurgaon and Sector 137 Noida. We know which listings are real because we walked into the units. That sample size is small, the conviction is high.
-
04
The editorial voice is the marketing.
Field notes is the only acquisition channel we are paying any attention to in year one. Investigative pieces that name builders and brokers are the kind of content the incumbents structurally cannot publish.
-
05
The public stale board is a growth loop.
Every flagged listing on /stale/ is a piece of distribution. Buyers send the page to each other. Brokers complain about specific listings and accidentally do the marketing for us.
06 · What we're building next
Twelve-month roadmap, in priority order.
We are deliberate about what we are not building. No EMI calculator competing with a bank. No vanity feed. No "AI agent" that hallucinates a price.
-
Q3 · this year
Live RERA integration.
Today we cross-check against a seeded snapshot of HARERA and UP-RERA project IDs. Q3 ships the live scrape and the registration-matching logic, so a project that gets struck off RERA loses its 25 points the same week.
-
Q4 · this year
Seller-side workflow.
A direct-from-owner posting flow with society NOC upload, EXIF-checked photos and OTP-verified contact. Owners pay nothing; brokers pay only for verified leads.
-
Q1 · next year
Broker incentives layer.
A clear, published path for a Basic broker to become Verified Pro — not pay-to-promote, but verify-to-promote. Lifts inventory quality without lifting our headcount.
-
Q2 · next year
Transaction layer.
Token amount, agreement-to-sell, registration coordination — the parts of the buyer journey that today route through five separate humans. This is the part of the business that earns more than per-lead pricing, and it is the natural endpoint of a trust portal.
07 · The ask
Pre-seed, deployed against Delhi-NCR product depth and the first 10,000 organic listings.
We are raising a pre-seed round. It funds a four-person team for eighteen months — two engineers, one editorial / investigations lead and one operations person walking Gurgaon and Noida. The work it pays for, in priority order: live RERA integration, the seller-side workflow, the first ten thousand organic listings sourced and scored by hand, and the editorial backbone that makes Field notes a real publication rather than a content-marketing exercise.
We are not optimising for a maximum cheque or a maximum valuation. We are optimising for two or three investors who will use the product, send us a list of broker behaviours they want flagged, and not flinch when we publish an investigation that names a builder they happen to hold equity in. If that is the conversation you want to have, the contact link is in the footer.
Decks lie. Listings don't. Spend 5 minutes on /trust before this.