Mapbox Integration for Location Search
When the answer is a place.
A map is easy to add and hard to make useful. The difference is whether it is a picture of your results or the instrument people search with — and that is decided in the data model, not in the map library.
Mapbox runs the map on the LA28 venue hub for Visit California, where a delegation filters training facilities by distance to the Athletes’ Village and reads candidate venues against official competition sites.
Mapbox is vector maps under full control: styling, markers and layers, geocoding and directions, and rendering that stays fast with thousands of points on screen. Most maps on content sites are decoration. Pins get dropped, the map sits beside a list, and no one’s decision changes because of it. A map earns its weight when the user’s actual question is spatial — how far is this from the thing I care about, what else is near it, which of these are the real options — and answering that requires the underlying records to carry structured location and the map to be wired to the same filters as the list, so the two are one interface rather than two views that disagree. Mapbox is a good tool for that. It is also a good tool for decoration, which is why the tool is rarely the reason a map works.
Our Work with Mapbox
How the Work Splits
Vector maps with full control over styling, markers and layers, geocoding and directions APIs, and rendering that stays fast with a large number of points on screen.
The part that decides whether the map is useful — structured location on the records themselves, distance filters computed against the reference point users actually care about, map and result list driven by one filter state, visual distinction between categories of pin that mean different things, and event instrumentation so map use is measured rather than assumed.
A map a delegation can act on. On the LA28 venue hub, facilities are filterable by distance to the Athletes’ Village, the map separates candidate training venues from the state’s official competition sites so the two are never mistaken for each other, and results and map move together as filters change. Map interactions are measured alongside searches and exports.
The work in practice
Adding Mapbox to a site takes an afternoon. Making the map matter takes the work that happens before you open the library — and that work is almost entirely about the data behind the pins.
What makes a map worth having
A spatial question people are actually asking. “Where is this” is a caption. “Which of these is within an hour of the place I have to be every day” is a question, and only a map answers it well. If your users are not asking the second kind, a map is decoration and an address is cheaper.
Distance to the thing that matters, not distance from the viewport. Generic proximity — near me, near the map centre — is the default and is usually the wrong reference point. The useful filter measures against whatever the user’s decision actually revolves around, which is domain knowledge rather than a map feature.
One filter state, two renderings. The most common failure is a map and a list that drift apart, each with its own idea of what is selected. They should be one interface. A filter change moves both, and what is on the map is exactly what is in the results.
Pins that mean different things, and look like it. When a data set mixes categories with different authority — candidates versus confirmed, ours versus theirs — collapsing them into one marker style quietly misleads. The visual distinction is a correctness requirement, not a design flourish.
Instrumentation, so you find out. Map use is the easiest interaction to assume and the hardest to see. If nothing records pans, zooms and pin opens, no one can say whether the map earned its place, and it stays in the build forever on the strength of the fact that it looks good.
The LA28 venue hub
Ahead of the 2028 Games, national Olympic and Paralympic committees and sport federations scout California for pre-Games bases — pools, tracks, high-performance facilities, meeting space, somewhere to put a hundred people. Geography is not a nice-to-have in that search; proximity to the Athletes’ Village is one of the first constraints a performance director applies.
So the map is wired to the search rather than sitting beside it. Facilities carry structured location, distance to the Athletes’ Village is a filter alongside sport, fields of play, recovery amenities and accommodations, and the map distinguishes candidate training facilities from the state’s official competition venues so a delegation never reads one as the other. A training-facility search within 200 miles of the Village returns 62 candidates today. Map use is measured along with searches and exports, so the map’s contribution is a number rather than an assumption.
Built with Payload CMS behind it, a Next.js front end and Algolia search; the engagement is Where the World Trains Before the Games.
If you are adding one
Start with the question, not the library. Write down the spatial decision your user is making and check whether your records carry the structured location to answer it — most do not, and retrofitting that is the actual project. If the answer is that people mostly want an address and a link to directions, give them that. It is faster, cheaper and more useful than a map no one filters.
Practice leadership
Jack Hartman
Jack Hartman’s specialty is cultural and destination platforms, where a map is how someone decides what to go and see.