An alternative to scraping business listings that does not break every month
Why homemade scrapers fail, and what to use instead: a places database for the bulk, a live map search for the newest, a browser task for the rest.
Part of the guide Location intelligence automation
The practical alternative to scraping business listings is to stop writing code against other people's web pages and use three things instead: a places database for the bulk of businesses with positions, ratings and types already cleaned, a live map search for places that opened in the last few days, and an AI browser task only for the one niche directory or register that nothing else covers. Each fails in fewer ways than a scraper, and none needs a developer on call when a page changes.
This is for operations and growth teams who have, at some point, paid a freelancer for a scraper and watched it die. It explains why scrapers break, what each replacement does, how to choose between them, a made-up example that uses all three, and how to run the result as an AI employee. The location intelligence automation guide covers the data side in more depth.
Why scrapers keep breaking
A scraper is a program that reads a web page and pulls out the parts you want by their position in the page's code. It works on the day it is written. Then:
- The page changes. A redesign moves the phone number into a different box, and the scraper returns blanks without saying anything is wrong.
- The site pushes back. Too many requests from one address get slowed, blocked or met with a human check.
- The terms forbid it. Many sites do not allow automated collection, which turns a technical problem into a legal one.
- The data is raw. Names come with stray text, addresses have no position, categories are free text, and the same business appears three times under three spellings.
- Nobody owns it. The person who wrote it has moved on, and the code is a mystery the next time it fails.
The cost is rarely the scraper itself. It is the quiet weeks of bad data before someone notices.
What to use instead, and when
| You need | Use | Why it holds up |
|---|---|---|
| Most businesses of a kind in an area, with position, rating, chain and type | A places database (the Nearby places block) | Already structured and refreshed monthly; no pages to parse |
| Places that appeared on online maps this week | A live map search (the Map business scan block) | Searched live, kept only inside your area, positions checked |
| A list that lives only on one directory, register or portal | A browser task (the Browser block) | Described in words, so a layout change does not break it |
| To know when a page you watch changes | Web page watch | Compares each run with the last and says what is new |
Which tool to reach for. Most needs are met by the first row.
Nearby places reads xMap's places database of more than 300 million places, sorted into more than 5,000 types, each with its distance, brand, chain or independent, rating, reviews, opening date and contact details when known. Map business scan searches online maps live, inside a city, a circle, your own boundary or around your own places, and can cover a whole area with a grid of searches.
The Browser block is for what is left: a trade association's member directory, a council licensing portal, a supplier catalogue with no export. You write the task in plain words ("search this directory for members in Madrid and bring back name, address and registration number"), and xMap AI drives a real browser step by step. Because it reads the page the way a person would, a moved button does not break it the way it breaks a scraper. It reads public pages only unless you sign in yourself, leaves alone sites whose robots file asks robots to stay away, never types a password or payment details, and a task that brings nothing back fails loudly instead of returning blanks.
A worked example
This example is made up. An insurer in Madrid offers cover for veterinary clinics. Its old scraper copied clinic listings from a directory site and broke twice a year. The team rebuilds it as one flow:
- A Schedule runs the flow on the first Monday of each month.
- Nearby places searches veterinary clinic types across the districts of Madrid in an Area of interest, keeping places with a phone or website.
- Only new drops clinics the flow already sent in earlier months.
- The Browser block, run for each new clinic, looks it up in a public professional register the team names in the task and brings back whether a registration is shown.
- An xMap LLM block writes one line on each clinic for the underwriting team, and an Email sends the list.
Nothing in the flow depends on where a box sits on a page. When the register redesigns its search form, the browser task still describes the same job in words.
Mistakes when switching away from scraping
- Using a browser task for bulk lists. It is slower and costs more than a database. Use it for the narrow slice nothing else holds.
- Expecting the database to have this morning's openings. It is refreshed monthly; add a live map search when days matter.
- Keeping the old scraper's fields. Start from what you use. A shorter list of reliable fields beats a long list of half-empty ones.
- Skipping duplicate removal. Combining a database search and a live search finds some places twice; Remove duplicates keeps one.
Running it as an AI employee
Deploy the flow and it becomes an AI employee with a Team inbox thread, where each run leaves a short note: how many places were found, how many were new, and any browser task that could not finish. Templates to begin from: New outlets to CRM for a live map search into a CRM, Businesses with low ratings and no website for a database-led list, and What the best-rated cafes near you say on their own websites, which pairs Nearby places with the Browser block.
Questions
Is a places database legal to use where scraping is not?
A database is a product licensed for your use, so you are not collecting from someone else's pages yourself. Always check the terms of any site you point a browser task at.
What if the business I need is not in the database?
Add a live map search for recent openings, or a browser task for a directory that lists it. Combine them with Remove duplicates.
Does the Browser block get past CAPTCHAs?
There is an optional setting that lets the browser service solve bot checks, meant only for sites whose terms allow it. Off, a bot check ends the task and the step says so.
How fresh is the data compared with a scraper?
xMap's places database is refreshed monthly; a live map search reflects online maps at the moment it runs. A scraper is only as fresh as its last successful run.
Templates that do this
Read next
More on business and places data · Every article · Templates