BariKhoj — Property Marketplace & Rent Ledger

The build

Dhaka's rental market runs on three fictions: the listing that is 'available' months after it went, the advance whose return is negotiated at move-out with no record of what was ever held, and the viewing arranged by nine phone calls. BariKhoj rebuilds each as a data object with an expiry and a witness — every listing carries a dated photo whose verified badge quietly lapses without a refresh; every flat has one rent ledger showing collected, due and advance-held that landlord and tenant read simultaneously; every viewing is a scheduled appointment with honest travel time. The result is not a prettier portal. It is the first portal where a bad listing is impossible to keep, a moved-out dispute has a paper trail by default, and the app earns the advance conversation instead of avoiding it.

What we handled

• Photo-expiry verification: 'verified' is dated evidence that decays, not a badge a broker buys once and keeps forever
• One rent ledger, two readers: collected ৳90,000 / due ৳45,000 / advance held — the move-out argument starts from shared arithmetic
• Advance as first-class state: security money is tracked from deposit to itemised return, not remembered in a notebook
• Viewings as appointments: slot, guests, drive time and send-address in one action — phone-tag becomes a fallback, not the system
• Ghost-listing economics broken: listings that never refresh lose their badge and their ranking, so stale inventory hurts its owner
• Tenant needs nothing: the ledger, receipts and notice window live in a link they can open on any phone browser

A portal optimises leads; a market needs truth

Property platforms earn on listings and phone-tag exposure, so their incentives point at volume of inventory, not quality of inventory. A ghost listing costs the lister nothing and costs every renter an afternoon. Meanwhile the money in the relationship — the two-to-six-month advance, the monthly rent, the itemised return at move-out — lives outside the platform entirely, negotiated back into existence only when someone wants a fight. Software that never touches the money cannot referee the market.

BariKhoj inverts both. A listing is dated evidence, not an advertisement: the verification badge carries the photo's timestamp and quietly expires ninety days later, which demotes the listing until the owner re-checks. And the ledger is the product's centre of gravity — every flat has exactly one rent account showing collected, due and advance-held, readable by landlord and tenant simultaneously. Discovery brings the parties together; the ledger is what makes leaving the platform expensive for both of them, which is how a marketplace actually retains supply without paying for it.

The badge that ages is the badge that means something

Verification here is deliberately unglamorous: a photo of the actual unit taken at the actual address, a claim from the owner or authorised agent, and a match against contact and duplicate fingerprints — photo hashes and phone reuse across multiple 'available' flats get flagged on write, not in a quarterly audit. What makes it trustworthy is decay. Ninety days without a refresh and the badge lapses, rank drops, the UI says so in plain words. Brokers can live with it; owners marketing their own flats like it; renters read it as the first honest signal in the category.

There is no title-registry fantasy in this design. The platform verifies presence and price honesty, not legal ownership — the listing page says exactly that in one line, because a trust badge that promises what a browser cannot check burns trust instead of building it. Where deeper ownership documents are in scope, they ride the onboarding visit as a human step with an outcome recorded in the same expiring structure.

The advance is not a rumour: money with a home

Security money is modelled as first-class state: held, itemised against documented deductions, and returned on a clock the platform nudges — because the move-out argument is not really about the deductions, it is about nobody having written down what was held in the first place. The monthly ledger works the same way as a courier's settlement sheet: January paid, February paid, March due in four days, advance held ৳90,000, collected versus due reconciled in the footer. Both parties open the same numbers before anyone opens their mouth.

Payment evidence attaches to ledger lines — a bKash or Nagad reference, a photo of a receipt, a landlord confirmation — and a late flag is a property of the line, not a mood. Reminders are the only 'notification feature' in v1 because everything louder becomes the tenant's enemy and the product dies in a group chat. Rent automation was deliberately kept off the roadmap until the ledger itself is boring; a system that chases money before it can record it is how fintech features become trust liabilities.

A viewing is a time, not a phone number

The viewing scheduler treats visits like dispatch treats runs: real slots, guest counts, and honest travel time computed from the searcher's usual origin, so '10 minutes from the office' stops being sales copy. The landlord confirms once; the address, gate instructions and a map pin send themselves; a no-show writes back to the listing's response quality, which feeds rank. It is the smallest feature with the largest manners effect in the whole platform, and it is why the demo page ends on a calendar, not a gallery.