TL;DR: To scrape emails for agency prospecting, a tool visits each business website and lists the public addresses it finds. Many local businesses publish no email at all, so a scrape returns fewer leads than the list size suggests. Verify what you collect, check the rules for sending, and treat the export as a starting list, not a send list.
Most scraped lists are smaller and riskier than the export file suggests, so filter before you send.
All five pages I analysed for this post end with a spreadsheet, and all five are vendor pages that sell the tool. They teach the same three steps: pick a source, let the tool visit the pages, export a CSV. That workflow is correct as far as it goes. It stops before the question an agency has to answer: are these addresses real, allowed to contact, and worth a message?
When I cold called local businesses for web clients, building the list took the most time. A scraper is faster, but it only changes how you collect the list. The checks that decide whether the list works come after the export.
How does a website scraper find addresses on a site?
A website email scraper reads the public pages of a business website and pulls out the email addresses it finds there. It does not search a database of contacts. Good tools check the homepage first, then the contact, about and team pages, because that is where a business usually publishes an address.
A crawler finds an address only where the business chose to publish one. Hubertino checked 65,024 top-ranked Google Maps listings and found an email on the business's own site for 63% of them (Hubertino).
Why does that matter for your list? A scraper that skips the contact page is like a waiter who checks the front door and calls the restaurant closed, while the kitchen door stays open.
Where does the address hide? Usually one click away. I learned this on those cold calls: an owner once read me an address from the contact page that my first look had missed. Tools that follow sub-pages catch more of these. Crawl depth matters.
LeadsAgent works the same way. It visits each business website, checks the homepage and the contact, about and team pages, and drops addresses that are placeholders, noreply inboxes or a website builder's own address. You get the public emails a business publishes, not guessed ones.
How many local businesses list an email on their own site?
Many US local businesses publish no email on their own site. In a 2026 study of top-ranked Google Maps listings, 56% of the US listings had one, against 81% in Germany (Hubertino). Put another way, about 44% of US listings gave a scraper nothing to find.
Your scrape will return fewer leads than the list size, so plan for the gap before you buy credits. GMaps Scraper, a vendor that sells scraping, reported an email for 36.8% of 1,174 US listings it checked. The rate ran from 12.3% for car dealerships to 60.4% for hair salons (GMaps Scraper). The sample is small and the source has a stake, so use the range as a rough guide.
The common advice says scrape, export, then send. The turn is that the export is only a starting list. Some of the gap is businesses with no website, and some have only a social or booking page. Use an ideal customer profile to decide which listings deserve a phone call before you spend time on them. Drop the listings with no site, call the rest, and send only to rows that pass the checks below. Where a business lists named staff on its own pages, you can start sooner, but you still run those checks.
Is it safe to scrape LinkedIn emails?
LinkedIn email scraping is not safe, because LinkedIn bans the method. Its User Agreement bars crawlers and scripts that scrape profiles or other data. Its help pages warn that accounts using such tools can be restricted or shut down. Some sales teams try it anyway, so know the risk before you start.
The biggest risk is your own LinkedIn account, and LinkedIn has taken legal action against a data vendor. LinkedIn sued Proxycurl in January 2025 over scraping through fake accounts, and the case ended in July 2025 with a permanent injunction (January filing; July ruling). Its Sales Navigator help page says features can be limited when it spots unusual activity (LinkedIn Help).
Think of LinkedIn as a directory of names and job titles, not a list of verified inboxes. Use it to learn who runs a business and what they care about. Then find the email on the company's own site or a published contact page, and send from there. If you searched for how to scrape email from LinkedIn, this is the honest answer: the step that pulls the address is the step LinkedIn forbids.
Which free allowance is worth testing first?
Start with the free allowance that fits your list size, not the one with the biggest headline. Hunter gives 50 credits a month, Lusha gives up to 40 credits a month, and Outscraper covers the first 500 contacts a month. Skrapp gives 50 credits that do not roll over, and Saleshandy gives 50 credits once, in a 7-day trial. Evaboot shows no free allowance.
In my reading, one free allowance tests one list, not a whole campaign. The pricing pages show how small the allowances are: Hunter, Skrapp, Lusha, Evaboot, Outscraper and Saleshandy. Check the current pages before you commit, because allowances change.
I call this The Fifty-Row Test: a small, repeatable sample before you spend anything. Pick one search, such as dentists in one city, the way a focused agency pipeline starts with map listings. Scrape the first 50 results, count the addresses, run them through your checks, and note the duplicates.
What do those 50 rows tell you? Whether the tool finds real addresses in your market, and how many survive the checks. Then run the same 50 through a second tool. The cheapest test is the one you can repeat. Start small.
How do you check a scraped email before you send?
Check each address in steps before it reaches your outreach tool. Run the cheap checks first, so the obvious failures drop out before anything else. LeadsAgent checks each address's syntax, then confirms the domain can receive mail. Each row gets a deliverability result, and the verified-only filter keeps the rows that pass.
A DNS check catches dead domains, but it cannot prove that a person reads the inbox. LeadsAgent does not run a mailbox-level check yet, so a verified result means the domain is set up to receive mail. One vendor benchmark of 1,413,405 sends reported 1.27% true hard bounces, against a 3.04% headline counter before the vendor sorted the rows (RevenueFlow).
I call that gap The Dead Inbox Problem: an address that is spelled right, sits on a live domain, and reaches nobody. It is like a letter addressed to an empty house. You usually find it after the send, when a bounce or a silence tells you. Saleshandy's guide sets its own rules here: remove catch-all domains, keep bounces under 2%, and re-verify any list older than 60 days (Saleshandy). Those are its thresholds, not an industry standard, so write down your own.
LeadsAgent marks each address that fails as not deliverable and keeps it in the file, so you can see what you lost. Count before you send.
Before you use any list, including one you collected yourself, run the checks in the due-diligence guide for bought lists. It walks through a small sample test with a stop rule before you send to the full file.
If you want those checks built into the same search, LeadsAgent runs them on every address it finds.
Are generic inboxes like info@ fair game for outreach?
Often yes, but the rules change by country, so check the market your list comes from. In the FTC guidance I read, US rules do not single out generic inboxes. The French regulator is different. It has its own tests for a generic address, and those depend on how the address was collected and what your message says.
If you email people in France, the regulator treats a generic B2B address differently from a named person. The CNIL's 2009 guidance allowed prospecting a generic address such as info@ or contact@ without prior agreement, if the person could object when the address was collected and the subject related to their job (CNIL). That same guidance also barred prospecting from addresses taken off public web pages. The CNIL's 2022 page I read allows B2B prospecting on an opt-out basis, with each message naming the sender (CNIL). It does not repeat the web-page bar, so check the current rule before you rely on either version.
A vendor's data points the same way from a different angle. BounceZero's sample of cold-list checks found info@ addresses valid 13.4% of the time, against 3.3% for contact@ (BounceZero). That sample came from one vendor's checks, not from published addresses, so it shows that role inboxes carry more risk, not how often a given business reads them.
My working rule: use a generic inbox only when the business has no named person on its own pages, and send one short message with a clear opt-out. If the page names an owner or manager, write to that person instead. Ask first, then send.
Is it legal to email what you scrape?
Two separate decisions apply here: collecting a public address, and sending a cold message to it. The rules for sending depend on where the recipient is and what the message says. This is general information, not legal advice, so check the rules in your own market before you send.
In the United States, cold email does not need prior consent, but it does need an opt-out. The FTC's guide says each separate email that breaks the law can cost up to $53,088. Each message also needs a valid postal address and a clear opt-out that works within 10 business days (FTC). In an FTC staff blog post, one staffer named Christopher put it plainly: "In other words, there is no opt-in requirement" (FTC). He also warned that "buying lists like that can be risky" (FTC).
In the EU, the key rule for cold email is the right to object. GDPR Article 21 lets people object to direct marketing at any time, and processing must stop once they do (EUR-Lex). The French regulator fined the data company Kaspr EUR 240,000 in December 2024, partly for collecting LinkedIn contact details whose owners had restricted visibility (CNIL). That case was about collection, not cold email as such, so scraping and emailing carry separate risks.
Before a campaign, write down your legal basis for each market, keep the unsubscribe link working, and show a valid postal address. Then have a lawyer in each market confirm the details.
FAQ: common questions about scraping emails
Do I need a crawler or a database tool?
I could not find an independent comparison that measures accuracy by method. The 2026 benchmarks I found rank tools by verified output but do not sort them by method, and they disagree on the same tools. Run both kinds on the same 50 businesses and count the addresses that pass your checks. Keep the headline in mind: about 44% of US listings in one study had no email on their own site (Hubertino).
Can I scrape emails from Facebook?
Vendors say they can pull emails from public Facebook pages and groups without logging in. None explains its method in enough detail to check. I could not open Meta's own terms page during research. Secondary reports quote its terms as banning automated collection without permission, so treat those vendor claims as a risk, not a method (Social Media Today).
Which Chrome extension should I install?
Read the store listing before you install. Of six email scraper extensions I checked in the Chrome Web Store, three declared non-trader status, two declared trader status, and one would not open. Some privacy panels contradict their own descriptions, so read the panel, not just the summary.
What does LeadsAgent cost?
LeadsAgent is free forever: 1,000 free leads every month, with emails and lead verification included, and no credit card required. Paid plans are $29/month for 10,000 researched leads, $99/month for 50,000, and $199/month for 100,000. Billing is monthly per lead researched.
Next step: Pick one search, scrape the first 50 results, and count the addresses that pass your checks. When you are ready to run a full search with emails, verification and website checks in one list, try LeadsAgent on a small batch first.
