An interactive map can be the most useful thing on a website: store locations, project sites, property listings, service areas, live assets in the field. It can also be the heaviest thing on the page by a long way. Here is how we keep maps fast without giving up what makes them useful.
Why maps get slow
Most slow maps share the same few causes:
- The map library loads on every page, even pages where nobody scrolls to the map
- All the data loads at once, often as one large file with full detail for every shape
- Thousands of markers are drawn individually, which strains phones in particular
- Map tiles come from a distant or rate-limited service
None of these are hard to fix. They just need to be planned from the start rather than patched later.
How we build them
1. Load the map only when it is needed
If the map sits halfway down a page, there is no reason to load it before someone gets there. We show a lightweight static preview first and load the interactive map when it scrolls into view or someone taps it. The rest of the page stays fast either way.
2. Use vector tiles
Instead of sending the whole dataset, vector tiles split it into small pieces by area and zoom level. The browser only downloads what is on screen. Zoomed out, it gets simplified shapes. Zoomed in, it gets the detail.
We often package tiles as a single file in the PMTiles format and serve it from Cloudflare storage, so there is no separate map server to run or pay for.
3. Simplify the shapes
Boundaries and routes exported from GIS tools are often far more detailed than a screen can show. Simplifying them for each zoom level can shrink the data dramatically with no visible difference.
4. Cluster dense points
When hundreds of locations sit close together, we group them into a single counted marker that splits apart as you zoom in. It reads better and draws far faster.
5. Draw with the GPU
Libraries like MapLibre GL draw maps using the device’s graphics hardware, which handles large datasets and smooth zooming much better than older approaches.
The fastest map is the one that only sends what’s on the screen.
Keeping the data live
“Live” means different things for different clients, so we match the setup to how often the data actually changes:
| How often it changes | What we use |
|---|---|
| A few times a year | Data built into the site, updated with a quick redeploy |
| Weekly or daily | Data pulled from your CMS, spreadsheet or database at build time, on a schedule |
| Constantly | A small API on Cloudflare that serves fresh data, cached for seconds or minutes |
The aim is that your team updates the data in a tool they already use, and the map follows. Nobody should have to edit map files by hand.
Making maps work for everyone
A map on its own isn’t accessible to everyone. Screen reader users, keyboard users and people on poor connections can’t always use one. So every map we build has a companion:
- A searchable list of the same locations, with addresses and details
- Keyboard controls for panning, zooming and selecting
- Text summaries for anything the map communicates visually, like coverage areas
The list also helps search engines and answer engines, which can’t read a map but can read a list of locations.
Where this is useful
- Multi-location businesses with branch or stockist finders
- Property, land and development projects
- Councils, utilities and infrastructure teams sharing assets or works
- Data platforms that need to explain where things are happening
Maps are one of our specialities, and our team includes GIS and backend engineering experience. If you have location data you’d like people to explore, we’d love to hear about it.