The challenge
A real estate brokerage was publishing its inventory through a website that had to be updated by hand every time a property was listed, priced, reserved, or sold. Agents sent property details and photos to whoever maintained the site, which meant listings went live late, stayed live after a property was gone, and frequently appeared twice when two agents submitted the same unit from different sources.
The public site was also invisible to the searches that mattered. Property pages carried no structured data, filtering was limited to a handful of coarse categories, and there was no stable URL scheme, so a listing that ranked would lose its position the moment it was re-entered. Buyer enquiries arrived as unattributed emails to a shared inbox with no record of who followed up or what happened next.
The brokerage needed agents to own their own inventory and their own leads, without giving every agent the ability to publish anything they wanted to the public site unreviewed.
Our approach
We built a public listings site and an agent back office against one shared property record, so a listing's public page and its internal state are the same data viewed through different permissions. Agents create and edit their own inventory, move a property through its lifecycle (draft, active, reserved, sold), and a listing disappears from the public site the moment its status changes — no separate step and no stale page left behind. Publishing runs through a review state, so an agent can prepare a listing without it going live unreviewed.
Duplicate control was treated as a data problem rather than an editorial one. Properties are keyed on building, unit, and location attributes, and a submission that matches an existing record is flagged for merge instead of silently creating a second page competing with the first. Property URLs are stable and independent of title edits, and every listing page emits structured real-estate markup so search engines can read price, location, and property attributes directly rather than inferring them from prose.
Photography is the heaviest part of a listing, so images are uploaded straight to cloud object storage and served in modern formats at the sizes each layout actually needs, keeping gallery-heavy pages fast on mobile connections. Enquiries are captured as assigned leads against both the property and an agent, with a follow-up trail, so a buyer question is a tracked record rather than an email in a shared inbox.
Outcome
Listings now go live and come down as agents change their status, rather than through a manual site update, and the public site reflects real inventory instead of a lagging copy of it. Duplicate submissions surface as merge candidates before publication, and every buyer enquiry is attached to a property and an owner with a visible follow-up history.