Why one search for a whole city misses businesses
A map search for "electrician Manchester" returns a ranked list, not a census. Map apps show a limited number of places per query (the Google Maps list typically stops at around 120), favour what is near the map's centre and filter by what they judge relevant at that zoom level. In a dense city, hundreds of genuine businesses never appear for the broad query.
Searching the same keyword from many points across the area changes that. Each point sits over a small neighbourhood, the map is zoomed in, and the ranking now competes over a few streets instead of the whole city. Run the keyword at every point, combine the results and remove duplicates, and coverage rises sharply. This tool produces the points; the searching can be done by hand for a few cells or by software for hundreds.
Making a grid
Choose the area
Pick a city preset, type the north, south, east and west edges of a bounding box, or switch to Centre and radius and give a point plus a distance in kilometres.
Set the spacing
The spacing is the distance between neighbouring points. Start with 1 to 2 km for town centres and 5 km or more for rural districts; the counter shows how many points that creates.
Pick a layout
Square places points in neat rows and columns. Staggered shifts every other row by half a cell, which covers the area with fewer gaps between circles of the same size.
Add your keyword
Optional: type the trade you are looking for, such as
roofer, and every point gets a map search link at the suggested zoom.Export
Download CSV for spreadsheets or job lists, KML to view the grid in Google Earth or My Maps, or GeoJSON for QGIS and other mapping tools.
Example: plumbers across central Manchester
Using the Manchester preset with 2 km spacing and a square layout gives 54 points in 9 rows, with a suggested zoom of 15 (at that zoom a typical map window shows about 2 km across). With the keyword plumber, each row of the CSV carries its own search link.
If a manual search in one cell still fills the whole result list, that part of the city is denser than the rest: generate a second grid for just that neighbourhood at 0.75 km and search it again. Overlapping results are normal and are removed when the lists are merged.
Area: Manchester preset (53.39 to 53.55 N, 2.32 to 2.15 W) Spacing: 2 km, square Keyword: plumber
id,row,col,latitude,longitude,zoom,keyword,search_url R001C001,1,1,53.39765,-2.310329,15,plumber,"https://www.google.com/maps/search/plumber/@53.39765,-2.310329,15z" R001C002,1,2,53.39765,-2.280197,15,plumber,"https://www.google.com/maps/search/plumber/@53.39765,-2.280197,15z" ... 54 rows in total
Choosing a spacing
There is no perfect number: it depends on how many businesses of that type exist and how tightly they cluster. These starting points work for most local trades; adjust after searching a few cells by hand.
| Area type | Spacing to try | Why |
|---|---|---|
| Dense city centre | 0.5 to 1 km | Many listings per street; small cells keep each search below the result cap |
| Inner suburbs | 1 to 2 km | Mixed high streets and housing |
| Market towns | 2 to 3 km | One centre plus outlying estates |
| Rural county | 5 to 10 km | Few businesses, long distances between villages |
| Specialist trades | Double the above | Rare categories (for example sign makers) rarely fill a result list |
How the points are calculated
Rows are laid out from the middle of the area outwards, a fixed number of kilometres apart, using 110.574 km per degree of latitude. Along each row the gap in degrees of longitude is widened or narrowed to match the real distance at that latitude (111.320 km multiplied by the cosine of the latitude), so a 2 km grid really is 2 km apart in Edinburgh and in Penzance. At city scale the error of this approximation is well under one per cent.
In radius mode, points are kept when they fall within the radius plus half a cell, measured along the Earth's surface with the haversine formula, so the edge of the circle is still covered. The staggered layout uses a row height of 0.866 times the spacing, which turns the points into a triangular lattice: every location is close to a search point with fewer points in total.
The suggested zoom is the level at which a map about 800 pixels wide shows roughly one cell. Searching at that zoom keeps the map's own idea of "nearby" in line with the grid. A grid stops at 5,000 points to keep the page responsive; widen the spacing or split the area if you reach it.
Getting clean results from a grid search
- Search a handful of cells by hand first and check that the results are actually local to each cell.
- Keep the keyword identical across cells so results can be compared and merged.
- Merge the lists and remove duplicates by place ID, website domain or phone number, not by name alone.
- Drop results that fall outside your area: map apps sometimes show strong matches from the next town.
- Record which cell each business came from, so a second, finer pass can target only the dense cells.
Questions people ask
How many results does a Google Maps search show?
The scrolling result list in Google Maps usually ends at around 120 places for a single query, and fewer in many cases. That cap is the main reason grid searching exists: breaking a city into small cells means each query covers a smaller area, so far more of the businesses that exist end up in at least one list.
What spacing should I use for a grid search?
Start with about 1 to 2 km in towns and cities and 5 km or more in rural areas, then search a few cells by hand. If a cell's search still fills the whole result list, the spacing is too wide for that neighbourhood; generate a finer grid for that part only.
Square or staggered: which layout is better?
Staggered covers an area more evenly, because every point is closer to its neighbours in all directions, so you need fewer points for the same coverage. Square grids are easier to read in a spreadsheet and to label by row and column. Either works if the spacing is right.
Can I open the grid in Google Earth or My Maps?
Yes. Download the KML file and import it into Google Earth or Google My Maps to see every point. The GeoJSON download works in QGIS and most web mapping libraries. Both files contain the point IDs so you can match them with your CSV.
Why is the grid not a perfect rectangle?
Points are spaced by real distance, not by degrees, so the number of points per row can change slightly with latitude, and rows are centred on the middle of the box. In radius mode, points outside the circle (plus half a cell) are dropped, which leaves a rounded shape.
Does the tool send my area to a server?
No. All the maths runs in your browser, and the page loads no map tiles, so not even the area you chose is sent anywhere. Only if you click a search link does your browser go to that map service.
Where your data goes
The grid is calculated by JavaScript on this page. No map tiles or location services are used, so the area you enter stays in your browser. Files you download are created locally. Search links open the map service only when you click them.
Checked against the tool on