Type "google maps data extractor" and look at what the search box suggests: free download, chrome extension, online, github. Those four qualifiers do not describe four products. They describe four different hopes about how simple the problem is. And the friction point is almost never the extraction itself. It sits somewhere else, and it can be measured: the gap between the field the extractor hands you and the field you need to book a meeting.
Five forms of extractor, five implicit contracts
Five very different objects share that one keyword. Confusing them is the leading cause of disappointment.
The browser extension. It reads the page in front of you and turns it into a table. It is by far the most widely adopted form: Instant Data Scraper shows 1,000,000 users on the Chrome Web Store (2026). Maps-specific extensions run on quotas: G Maps Extractor advertises 1,000 records per month on its free plan (2026).
The open source binary. A program you run yourself, with no account and no subscription. This is a real segment, not a curiosity: the gosom/google-maps-scraper project has 5,679 stars under an MIT licence (GitHub, read on 1 September 2026). You keep control over everything, including whatever breaks.
The official API. Google sells access to its place data, which settles the legitimacy question and raises the cost one. On the Places API, Text Search at the Pro tier is billed $32.00 per 1,000 requests on the first monthly band, with 5,000 free calls per month; the Enterprise tier, which unlocks extra fields, moves to $35.00 per 1,000 requests (Google Maps Platform, 2026). One request returns a page of results rather than one business, and the field you ask for decides which tier you are billed at.
The marketplace. A catalogue of extractors written by third parties, hosted and billed by a platform. Apify's Google Maps Scraper Actor advertises from $1.50 per 1,000 scraped places (2026), with per-event surcharges for reviews, contacts or email verification.
The dedicated service. You describe an area and an activity, the service returns a file. Outscraper advertises the first 500 businesses free, then $3 per 1,000 records up to 100,000 and $1 beyond that (2026). On the same ground, Scrap.io shows EUR 49 per month on monthly billing for 10,000 export credits, and reserves whole-country search for its top plan (2026).
The test that settles it: what is left to do after the export?
All these forms read the same surface. A Google Maps listing exposes a trading name, a postal address, a displayed phone number, a website link, a rating and reviews. That is exactly what you will get, whichever form you pick. So the useful question is not "which one extracts best", but what is still missing from the file at the moment I want to call someone.
Two things are always missing. The first is the link to the official register: a shopfront name is not a legal entity, and without a company number you know neither whether the business is still trading, nor its legal form, nor who runs it. In France that link is made against a free public source: INSEE describes the Sirene register as covering close to 25 million companies and 36 million establishments, free of charge and updated daily (2026). That is the job of a module for SIRET, SIREN, VAT and trade register enrichment, and it is also what makes it possible to then read the legal form and the names of directors before dialling a number.
The second is a usable point of contact. The listing gives you the switchboard, rarely a named address, and never an address whose deliverability has been tested. Finding and then verifying a professional email is a separate craft from extraction, with its own failure rates. Until that step is done, the number of exported rows tells you nothing about the value of the file, a point we set out in our definition of scraping.
Browser extensions, judged fairly
Three real strengths, and no other form combines them. They need no infrastructure: one install, one click, one CSV. They keep visual control in your hands, since you see on screen exactly the list that goes into the file, which makes a targeting mistake immediately obvious. And they cost nothing to try, which matters when you do not yet know whether the area you are after fits on one page of results or runs to several thousand rows.
For a local outreach push, an extension is often the right tool and there is no reason to change. We say the same in our comparison of alternatives to Instant Data Scraper.
They drop off on four points, all structural. Volume first: the list has to be scrolled to the end inside the browser, and Google only loads a limited number of results per query, which caps a single pass well short of a full county. Then resumption: close a tab, let a session expire, let the machine sleep, and the collection starts over. Then the dependency on your own workstation, which rules out scheduling and ties up your browser for the duration. On top of that sits the free-plan quota quoted above, which burns faster than expected as soon as you slice a city into districts.
One more point deserves naming without drama. An extension reads the page on your behalf, inside your session: responsibility for that reading stays with you. A third-party service takes on the extraction and documents it. Those are not the same position, and the difference matters the day someone internally asks where a file came from.
What Google's terms say, and what the French framework adds
Two frameworks stack here, and they do not say the same thing. On the publisher side, the Google Maps additional terms, under their "Prohibited Conduct" section, prohibit mass download or bulk feeds of the content, and the use of Google Maps to create or augment a mapping dataset, a business listings database or a telemarketing list for a service that substitutes for, or is substantially similar to, Google Maps (last modified 27 January 2026). That second prohibition is narrower than it is usually read: it targets building a competing service, not every act of reading. It is worth reading in full before you decide.
On the French side, published business data stays company data as long as it identifies no one. A director's name, a named email address or an individual mobile number move into personal data. For outreach itself, the CNIL states that prospecting professionals can rest on legitimate interest where the subject of the approach relates to the profession of the person contacted, provided the person is informed and given a simple right to object (2026). No form of extractor exempts you from that duty, and none of them discharges it on your behalf.
What each form returns, and what it leaves to the next step
| Criterion | Extension | Open source binary | Official API | Marketplace | Dedicated service |
|---|---|---|---|---|---|
| Nature of execution | in your browser | on your machine | licensed server call | shared infrastructure | vendor infrastructure |
| Scope of one pass | whatever loads on screen | per configuration and runtime | per billed request | per scheduled run | per area and activity |
| Dependencies | machine on, tab open | technical skill, maintenance | key, billing account, quotas | platform account, Actor choice | subscription or credits |
| Resumption after interruption | no | depends on implementation | handled by the caller | handled by the platform | handled by the service |
| Link to the official register | later step | later step | later step | separate building block | varies by vendor |
| Verified email supplied | no | no | no | separately billed option | varies by vendor |
| Billing unit | subscription or quota | your time | the request | the event | the credit or the record |
The deciding row is rarely the price. It is the "link to the official register" row, because that is the one that prices the work you will still be doing after you have paid.
Where outsend fits, and where it does not
outsend is not one more extractor: it is a chain that starts from a Maps search and returns a file already connected to the register. Extraction is not confined to a single country, and the enrichment modules that follow it add the company number, legal data, directors, extra phone numbers found on the website, verified emails, the site's technology stack or reviews and the owner replies to them. Access is invitation-based alpha and free: you can request free alpha access and judge on your own areas.
It is only fair to say where others remain the better answer. If your sources change every quarter and go well beyond Maps, a marketplace like Apify covers ground no dedicated service matches. If you embed places inside your own product and need an explicit licence, the official API is the only defensible answer, whatever it costs. If you need a systematic sweep of an entire country rather than a qualified list, a coverage service such as Scrap.io is better sized for it. And for forty shops on one street, a free extension is unbeatable. Our logic is the chained pipeline, and it carries a setup cost that a throwaway list does not justify.
Questions people ask before downloading anything
Is a free extension enough to get started?
Yes, if your target fits in a few hundred rows and you accept filling the gaps by hand. It stops being enough the moment you have to sweep the same area every month, or cover a whole county without leaving a tab open all day.
Why is the official API more expensive than third-party services?
Because it does not sell the same object. A third-party service sells a file; the API sells a documented right of use and fresh data per call. At $32.00 per 1,000 requests on the first band of the Pro tier (Google Maps Platform, 2026), it earns its place when the licence is part of the requirement, not when you are after a prospect list.
Are open source extractors reliable?
They are exactly as reliable as your ability to maintain them. A live, heavily forked project like the one cited above works, but Google's rendering changes and the repair work lands on you. The real cost is your time, not your invoice.
What is the email field in a Maps export worth?
It rarely comes from the listing, which publishes none: it is inferred from the website when the tool offers one. Without a deliverability test it tells you nothing. That is the first number to measure before writing, as we explain in our piece on cold email and the GDPR.
Do I want an extractor or a ready-made file?
If your targeting moves often, take an extractor and accept the downstream work. If your targeting is stable and you would rather spend time selling than tooling, look for the form that hands back a file already connected. Both answers hold up; what does not is paying for the first while believing you bought the second.