How we keep the data honest

Every number on this site comes from anonymous strangers. That only works if the pipeline between "someone typed a rent" and "a median on a page" is defensible. This is that pipeline, in full.

What a pin is

A rent pin is a location + what the renter actually pays + the flat's basics. No name, no phone, no email, no account. We can't identify who pinned. By design, there is nothing to leak or sell.

Privacy protections on the data itself

• Location fuzzing: displayed pin positions are deterministically offset by ~40 to 120 m, so a rent can never be traced to a specific flat or tower. Exact coordinates never leave the server (they're used only to compute the locality label and matching distances).
• Hashed IPs: abuse-prevention systems (rate limits, report caps) store irreversible hashes, never addresses.

Quality controls on the way in

• Sanity bounds: rents outside ₹2,000 to ₹5,00,000/month are rejected.
• Geofencing: pins in parks, lakes, forests or outside the Noida/GN/GN West boundary are rejected. Nobody lives in the Yamuna.
• Bot traps: hidden-field and timing checks catch automated submissions.
• Burst detection: suspicious patterns (many pins from one network, clusters of near-identical rents appearing within minutes) are quietly held for human review before they can touch any statistic.
• Rate limits: per-network submission caps, enforced in the database.

Quality controls once it's live

• Community reports: any pin or listing reported by 3 different people disappears from the map pending review. A single network can contribute at most 2 of those reports, so one WhatsApp group can't vote an inconvenient rent off the map.
• Outlier warnings: a rent more than 3× (or under ⅓ of) its area median carries a visible caution label.
• Duplicate weighting: the same flat pinned twice (a renewal, two flatmates) is counted once in every statistic.

How we present numbers

• Medians, not averages, as the headline figure, so one luxury penthouse can't drag a locality.
• Sample sizes everywhere: every median shows how many pins it rests on, labeled low / medium / high confidence. We'd rather look thin than pretend.
• No scraped data: nothing on this site is lifted from listing portals. Their numbers are broker asking prices, which is exactly what this map exists to correct. During the bootstrap phase some pins are clearly-labelled seeded estimates (see the next section).

Seeded estimates: the bootstrap phase, in the open

A crowdsourced map has a cold-start problem: nobody pins on an empty map. Our answer is seeding, done transparently or not at all:

• Some pins are seeded market estimates: calibrated from locality-level market knowledge, jittered around area centroids. Not scraped listings, and never presented as a real person's rent.
• Every seeded pin says so on the pin itself. Open one and the "seeded estimate" note sits beside its date.
• The homepage counter says "rents mapped" and counts everything on the map, seeds included. It deliberately does not claim they are all real submissions. Seeds are excluded from the Pin & Win draw and every leaderboard.
• Locality medians may blend seeded estimates with real pins while samples are thin. The sample-size label on every median tells you how much data sits underneath.
• As real pins arrive, seeds are retired. The goal is a map with zero of them.

What we don't do

• No broker listings, no paid placement in data, no advertising deals that touch the numbers.
• No selling of data. Contact details (for listings/seekers only) are shared exclusively inside a private match, and nowhere else, ever.

Found something that looks wrong? Use the flag on the pin itself, or check the FAQ for how hiding works. The map is only as honest as the people on it, and the people on it have been remarkably honest so far.