Harbourline sells waterfront. Their buyers do not think in municipal boundaries — they think in terms of which side of the bay, how far from the marina, whether the dock is deeded. Every off-the-shelf IDX widget they had tried forced those buyers into a dropdown of city names that meant nothing to them.

We were asked for one thing: a search that matches how a waterfront buyer actually looks. Everything else on the site exists to support that.

The search is a map first and a list second. Buyers draw a polygon freehand — around a bay, along a stretch of shore, across two townships if that is what they mean — and results filter live inside it. Those shapes are saveable, so a buyer working three stretches of coast keeps three named searches rather than starting over each visit.

Underneath, listing geometry is indexed for point-in-polygon queries so a redraw returns in well under a hundred milliseconds even across a few thousand active properties. Waterfront-specific facets the MLS carries but nobody surfaces — frontage in feet, shore type, dock and mooring rights — became first-class filters alongside the usual beds and baths.

The design bet is that drawing a search is a stronger signal of intent than filtering one. A named, saved polygon says more about what a buyer is looking for than any combination of dropdowns, and it persists — which is what makes it worth building alerts on top of.

It also changes how agents share listings. A link to a saved shape communicates in one click what otherwise takes a paragraph and a screenshot.