Property Map Search With Sub-50 ms Responses and 2x Engagement Using Typesense
We rebuilt a property platform's map search on Typesense geo search, so radius, drawn-area, and nearest-first queries return together with filters in under 50 ms.
map search response, filters included
map-based engagement within two quarters
query for location and all filters
A property map that lagged every time a visitor panned, zoomed, or filtered
Our client is a real estate listings platform that connects property seekers with homes for sale and rent across web and mobile. Leadership saw map-led discovery as the way to stand out from larger portals. But every pan or zoom returned slow, imprecise results, and adding price or amenity filters made it worse. Visitors gave up on the map, and often on the site.
The client partnered with Ksolves to build a search layer that answers location questions as quickly and precisely as keyword ones, across listings in multiple cities with price, property type, and amenity attributes.
C H A L L E N G E
01
Slow Location Queries Every pan or zoom returned sluggish results, leaving users looking at stale pins while the map caught up.02
Inaccurate Radius Results Location matching was approximate, showing properties outside the requested distance and missing some inside it.03
No Drawn-Area Search Users could not sketch a neighborhood or commute zone, and location search stopped at coarse city and locality fields.04
Location and Filters Out of Step Adding price, type, and amenity filters slowed responses further and produced filter counts that did not match the map pins.05
No Nearest-First Ordering Results could not be ranked by distance from a chosen point, so users scanned the whole map to find the closest options.06
Index Lagging Behind Listings New, updated, and delisted properties took time to appear correctly, which eroded trust in what the map showed.S O L U T I O N
One fast query that answers every location question alongside the filters
Ksolves, an AI-first DevOps services company, built a dedicated search layer with Typesense at its core. They remodeled listings data for geo search, built a lightweight Node.js API to translate map interactions into Typesense queries, and built a React front end around an interactive Mapbox map to keep the map, filters, and results moving together. Each fix below answers the challenge with the same number.
- 01 Geopoint indexing in Typesense Each listing's latitude and longitude is stored as a geopoint, so lookups run against an in-memory index instead of slow relational queries.
- 02 Exact radius filtering Geo filters on the geopoint field return only properties within the distance the user selects, removing the false matches and misses.
- 03 Polygon search from the map Shapes drawn on the Mapbox map become polygon coordinates through the Node.js API, so users can search inside any custom area.
- 04 Combined geo and faceted search Location, price range, property type, and amenities resolve in one Typesense query, so facet counts always match the pins.
- 05 Distance-based sorting Results sort by distance from the user's chosen point, with the nearest listings first and the distance shown on every result card.
- 06 Automated index sync A Node.js indexing worker upserts and removes documents as listings change, so the map stays current without manual reindexing.
Architecture
Technology stack
| Category | Tehnology |
|---|---|
| Search Engine | Typesense (geopoint index) |
| Map Interface | Mapbox |
| Frontend Framework | React |
| Search API and Indexing Worker | Node.js |
D E L I V E R Y
How a map interaction becomes one Typesense query
Ksolves delivered a dedicated search layer with Typesense at its core. A lightweight Node.js API turns every map pan, zoom, drawn shape, and filter into a single query, and the React and Mapbox front end keeps the map, filters, and results in step. An indexing worker updates the index as listings are added, edited, or delisted. The same foundation can power saved-area alerts, commute-time search, and location-aware recommendations as the platform grows.
From laggy maps to search that answers by location in under 50 ms
Ksolves replaced slow, approximate location search with a Typesense geo search layer that answers radius, drawn-area, and nearest-first queries together with filters in one call. Map searches return in under 50 ms, map-based engagement doubled within two quarters of launch, and the index stays in sync with live listings. Map-led discovery is now a dependable differentiator for the platform.
- <50 ms Map Search Response Panning, zooming, or filtering used to return slow, laggy results. Map searches now return in under 50 ms with all active filters applied.
- 2x Map-Based Engagement Users used to drift to list views or leave the site. Map-based engagement doubled within two quarters of launch.
- 1 Query for Location and Filters Location and attribute filters used to run as separate steps with mismatched counts. Geo, price, type, and amenity filters now resolve in one query with accurate facet counts.
- 2 Ways to Search by Area Location search used to stop at city and locality fields. Users can now search within any radius or hand-drawn polygon.
- 0 Manual Reindexing New, updated, and delisted listings used to take time to show correctly. The indexing worker now keeps the map current automatically.
T E A M
Meet the Engineers on Your Project
Every engagement is staffed by senior engineers from our DevOps team, who will design, build, and support your platform.
STILL LOSING USERS TO LAGGY MAPS?
Fix Slow Map Search and Broken Filters Instantly
Eliminate map bottlenecks with a sub-50 ms Typesense architecture tailored to your stack. Our DevOps team will audit your setup and optimize your search layer.