An email list website is any online source that supplies contact addresses for outreach: a searchable database, a finder, a list marketplace, a public registry, a directory, a job board, a community, a data cooperative, or a seller of scraped lists. They differ less in what they claim and more in what a verifier finds afterwards. Across the sources below, the share of addresses that could be confirmed deliverable ranged from above 90% for addresses the owner typed themselves to under half for scraped lists; the corpus average for non-freemail (business and ISP) addresses was 14.5% invalid, 2.3% catch-all and 27% unknown. This page ranks the nine source types and shows what to check before relying on any of them.
Ranked by confirmable share after verification, best first. Named-vendor invalid figures are the on-site numbers; "corpus" figures come from the BounceZero corpus, April to October 2026. Cost is per contact unless noted.
| Rank | Source type | Examples | Invalid on verification | Cost | Freshness |
|---|---|---|---|---|---|
| 1 | Your own signups and inbound | Forms, trials, webinars | 5 to 10% | Free | Days |
| 2 | Community and event member lists | Slack and Discord groups, conference attendee lists (with consent) | 8 to 15% | Free to low | Weeks |
| 3 | Job boards as company sources | LinkedIn Jobs, Indeed, niche boards, plus a finder | 15 to 25% after finding | Finder cost only | Days |
| 4 | Contact databases | Apollo, ZoomInfo, Cognism, Lusha | 5.8% invalid plus 18.4% catch-all | $0.05 to $1 | Months |
| 5 | Email finders | Hunter, Snov, FindyMail, Kaspr | 15 to 28% | $0.05 to $0.50 per find | Inferred now, verified then |
| 6 | Public registries and directories | Companies House, OpenCorporates, Clutch, G2, app marketplaces | High share of role addresses | Free | Months to years |
| 7 | Data cooperatives and intent networks | Bombora, G2 Buyer Intent, Demandbase | Varies; often company-level only | Subscription | Weeks |
| 8 | List marketplaces | Bought lists by industry or title | Corpus: about 1 in 7; .au lists 36.5% | $0.02 to $0.20 | Unknown |
| 9 | Scraped-list sellers | Bulk CSVs from unknown origins | Highest role and catch-all share; often a third discarded | Very low | Unknown |
Corpus: 6,514,884 non-freemail (business and ISP) addresses, 14.5% invalid, 2.3% catch-all, 27.0% unknown, 32.6% confirmed. Freemail on the same lists: 17.6% invalid, 74.0% confirmed. Full tables on the benchmarks page.
The best sources share one trait: the address was supplied or confirmed by its owner recently.
The owner typed it, with intent, this month. Typos and throwaways are the only loss, and a validation call at the form removes most of them. The constraint is volume: you cannot choose who signs up.
People who joined a group or attended an event in your category. Addresses are usually personal or current work ones, and the person is reachable on the platform if the email fails. Use only where the organiser permits contact; a member list is not a mailing list.
A posting is a company-level trigger with a date on it. Find the hiring manager or the role's boss, infer or find the address, verify. Fresh because the posting is, and relevant because the hire tells you what they are building.
Most lead generation at volume runs on contact databases and finders, ranks four and five. They are fast, filterable and broad. Their weakness is time: a database record was checked when it was indexed, and a finder's address was inferred from a pattern. The on-site Apollo accuracy study of 100,000 contacts found 5.8% of addresses were invalid at export and a further 18.4% were catch-all, an effective bounce exposure of 15 to 20% once catch-all bounce rates are counted; the finder comparison puts finder output at 10 to 25% invalid.
Those rates are workable if the export is verified before it is sequenced. They are not workable if the vendor's badge is taken at face value, because sending platforms enforce a 2% bounce limit and Google and Microsoft a 0.3% complaint rate at the sending domain and a single unverified export breaches it on day one. The two tools, finder and verifier, do different jobs; the distinction is spelled out in email finder vs email verifier.
Where the address sits matters as much as where it came from. These cuts explain why two lists from the same website can perform very differently.
| Segment | Addresses | Invalid | Catch-all | Unknown | Confirmed |
|---|---|---|---|---|---|
| gmail.com | 801,225 | 3.5% | 0.0% | 0.0% | 96.0% |
| hotmail.com | 253,445 | 16.5% | 0.1% | 0.0% | 70.5% |
| icloud.com | 50,895 | 73.0% | 0.0% | 4.9% | 20.0% |
| aol.com | 33,275 | 17.9% | 24.7% | 19.9% | 11.0% |
| .uk | 21,194 | 17.0% | 3.7% | 10.6% | 65.0% |
| .fr | 119,582 | 12.1% | 0.5% | 9.3% | 47.6% |
| .au | 560,378 | 36.5% | 6.3% | 17.6% | 37.3% |
| .de | 2,095,440 | 13.9% | 0.0% | 41.2% | 16.5% |
| .org | 56,401 | 24.6% | 12.4% | 25.0% | 36.2% |
BounceZero corpus, April to October 2026. An iCloud-heavy consumer list and a German B2B list are both poor email channels for different reasons: one is mostly dead, the other mostly unconfirmable.
Five checks that take an afternoon and save a quarter of wasted sends.
Owner-supplied, indexed from the web, inferred from patterns, bought from a partner, scraped. A seller who cannot answer is selling rank nine whatever the website says.
Not the vendor's sample. Pull 200 rows at random from a real export or a trial, run them through the verifier, and read the invalid, catch-all and unknown shares. The free tier covers this twice a month.
A UK B2B sample should land near 17% invalid and 65% confirmed. If it comes back 30% invalid, the source is old or scraped. If it reports zero unknown, the vendor is guessing on that share.
Above 10% role addresses means directory scraping. Above 10% catch-all means a vertical heavy in .org or .edu domains, which needs a non-email channel for that segment.
Rows times (1 minus invalid share) times (1 minus unknown share you will not email) is the number you are paying for. Cheap lists stop being cheap at this step.
Own signups and community members gave an address for a purpose; using it for a different one needs either consent or a documented legitimate interest and a clear opt-out. Database and finder contacts are work addresses, and B2B outreach on legitimate interest is defensible in the UK and EU if the source is stated and opt-outs honoured; PECR adds consent requirements for individuals and sole traders in the UK. Bought and scraped consumer lists are the problem category: no lawful basis for marketing to the individuals on them in the UK or EU, and a CAN-SPAM opt-out obligation in the US that the seller has not met. None of this is legal advice; it is the shape of the rules as they apply to each source.
Rank a list website by what a verifier finds in its output, not by what its homepage claims. Two hundred random rows and an afternoon settle it.
Seven sources ranked, build versus buy
The sellers compared, with sample-check rules
Marketplaces compared and the checklist before paying
How finders work and which to use
Building a list you own instead of renting one
The data behind every number on this page
The best sources are the ones where the owner supplied the address recently: your own signups, community and event lists used with permission, and job boards combined with a finder. For volume, contact databases such as Apollo, ZoomInfo and Cognism and finders such as Hunter and Snov are the workhorses, with 10 to 25% of their output invalid until verified. List marketplaces and scraped-list sellers rank last.
Buying a B2B list of work addresses and emailing those people about something relevant to their role can rest on legitimate interest in the UK and EU, with the source stated and an opt-out honoured. Buying consumer lists for marketing is not lawful in the UK or EU without consent, and in the US the seller rarely has the CAN-SPAM opt-out history you inherit. Verify before sending in all cases.
In corpus data from April to October 2026, bought business lists came back about one invalid address in seven, with Australian lists at 36.5% invalid and a further 27% of non-freemail (business and ISP) addresses unconfirmable across all sources. Accuracy depends on age and origin, which is why a random 200-row sample verified before purchase is the only reliable test.
A contact database (Apollo, ZoomInfo, Cognism) lets you search by filters and export contacts on a subscription, with records indexed over time. An email list website in the narrow sense sells a fixed CSV by industry or title. Databases are fresher and filterable; list sellers are cheaper and older. Both need verification before the first send.
Run it through a bulk verifier and read four numbers: invalid, catch-all, unknown and role. Drop invalid, route catch-all and unknown to non-email channels or low-volume secondary domains, remove role addresses from personal outreach, and compare the shares against the corpus benchmark for the list's country and provider mix. A vendor that reports no unknowns is guessing on that share.
Upload a random sample, get invalid, catch-all, unknown and role shares back, and compare them with the benchmarks before you pay for the rest.
Ayoub built BounceZero's 5-stage validation pipeline, its dedicated BGP-announced IP infrastructure, and the Patroni HA PostgreSQL cluster behind every verification. Previously built high-volume email delivery infrastructure. Trained at 1337 Benguerir (École 42 network, 2019). Open-source: bgp_analyzer.
Where lists come from, what bounces, and what to check before you pay
A B2B leads database is only as good as the share of its contacts you can still reach
Seven B2B contact database providers compared on coverage by region, email accuracy at export, refresh...
Sellers compared, with sample-check rules
How databases are built and where they decay
Thinking of buying a B2B email marketing list? Corpus data from 7.93M verified addresses shows what...
Where to buy leads for business in 2026
Continue through related topics