Maps Scraper

A maps scraper turns a "what, where" search into a table of businesses. Enter something like "dentists in Leeds" and you get back each listing with its name, street address, category, star rating, review count, phone number and website, in a form you can sort and filter rather than scroll through one card at a time. It is the same data a maps search shows, arranged for work rather than for browsing.

The usual reason to want it is a list: local outreach, competitive mapping of a category in a city, or checking how many businesses of a given type publish a website at all. The last of those is often the most useful output. A category where half the listings have no website is a category where a web designer has fifty conversations to start, and that ratio is visible in the summary before you read a single row.

Results come from the Places index rather than by driving a browser through the maps interface, which matters for two reasons. It is stable, because there is no page layout to break, and it is bounded, because five pages of ten is what the index returns before it starts repeating listings. Deduplication is on name plus address rather than name alone: a chain has one name and many genuinely distinct branches, and collapsing those would delete most of what a local search is for.

Two limits are worth being direct about. Fifty listings is a slice of a dense category, not a census - "restaurants in London" has thousands and you will see the first fifty by relevance, so narrow the search by neighbourhood if you need coverage rather than a sample. And the data is only as fresh as the index: a business that closed last month may still be listed, and phone numbers go stale faster than addresses. Verify before you call. There is also a daily cap, because each search costs real credits, and email addresses are not included because the index does not carry them.

A worked example shows what the Google Maps scraper tool puts in each cell. Search "coffee shops in Portland", click Has website, and the CSV button shows how many rows the download will hold, because the export follows the active filter. A listing rated 4.6 from 212 reviews lands as 4.6 in the Rating column and 212 in Reviews, a listing with no rating leaves both cells empty, and a name like Joe's "Best" Coffee, Inc. is written as "Joe's ""Best"" Coffee, Inc." so the comma and quotes stay inside one cell. That is the whole data set: seven columns of map data, no review text and no emails.