Apollo.io’s database is large, cheap to search and better than most alternatives at the price, and its email accuracy is lower than Apollo’s marketing says. Those statements are all true at once, and the gap between them is what this guide measures. Apollo claims 98% email accuracy. The largest independent test published in 2026 found that 73.2% of 100,000 exported Apollo emails were valid, 18.4% went to catch-all domains and 5.8% were invalid (Cleanlist, BounceZero). The right way to use Apollo follows from that number: search broadly, export, verify, and only then send.
Apollo’s 98% claim versus the third-party test
| Source | Sample | Valid | Catch-all | Invalid | Who published it |
|---|---|---|---|---|---|
| Apollo’s accuracy claim | Not stated | 98% | Not broken out | Not broken out | Apollo (vendor claim) |
| Cleanlist / BounceZero study, 2026 | 100,000 contacts exported in H1 2026 | 73.2% | 18.4% | 5.8% | Email verification vendors |
Both numbers come from parties with something to sell: Apollo sells the data, and the study’s authors sell verification. Read them as bounds rather than truth. The study’s method, running every exported address through SMTP-level verification, is the standard one, and its sample is large enough that the split is unlikely to be noise. It is also consistent with the rate at which cold-email tools report bounces on raw Apollo lists. What the 98% figure most plausibly describes is the share of addresses Apollo marks “verified” that are deliverable, not the share of all addresses in an export. Measured figures from Apollo Exporter’s own export volume are published on the methodology page rather than quoted here, so the numbers on this page stay third-party.
How Apollo gets its data
Apollo describes three inputs to its database. The first is a contributor network: users who connect their inbox or CRM share contact data back, which Apollo aggregates. The second is public web sources, including company websites and professional profiles. The third is data partners that license records to Apollo. Apollo then runs its own verification and assigns each email a status such as verified, guessed or unavailable. This description is Apollo’s own and is not independently audited. The contributor model explains both the breadth and the aging: a contact contributed in 2023 may have changed roles twice since, and Apollo’s refresh cycle is not published. It also explains why “guessed” emails exist, since a pattern inferred from a company’s known addresses is cheaper than a confirmed one.
How big is Apollo’s database?
Apollo publishes contact and company counts on its website in the hundreds of millions of people and tens of millions of companies. Treat those as Apollo’s claims; no third party has audited them, and the headline figure counts records, not current, reachable people. For a buyer the useful question is coverage of your segment, which you can test for free: Apollo lets you search and see result counts without spending credits, so run your target filters and look at the count and the share of rows that show an email status before paying anything. Depth within a single search is capped at 100 pages of 25 results, as explained in Apollo export limits.
What catch-all means, and why 18% matters
A catch-all domain is a mail server configured to accept messages for any address, real or not. When a verifier checks jane@example.com and the server says yes to everything, the verifier cannot tell whether Jane’s mailbox exists, so it returns “catch-all” rather than “valid”. In the 2026 study, 18.4% of Apollo exports fell into this bucket. They are not invalid; a meaningful share of them deliver. But they bounce at higher rates than verified addresses, and some catch-all servers accept mail and then discard it silently. Most verification tools let you choose: send to catch-all addresses on a separate, lower-volume sending domain, or drop them. Apollo marks some of these as “verified” and others as “guessed”, which is why Apollo’s own status column is not a substitute for verification.
Why verification before outreach matters
The arithmetic is simple. Send 1,000 raw Apollo emails and, on the study’s numbers, 58 bounce outright, about 184 go to catch-all servers with unknown outcomes, and 732 land. Verify first, and the sends go to the 732 valid addresses plus whatever share of the catch-alls you choose to risk, which keeps bounces near zero and protects the domain. Verification also changes the real price of a contact: a plan that looks like $24 per 1,000 emails works out to about $32 per 1,000 valid ones, as calculated in Apollo pricing and credits. Apollo Exporter runs verification on every email before the download is built, so the email status column in its CSV or XLSX reflects a live check rather than Apollo’s stored label; the verification method and its measured pass rates are on the methodology page. The same discipline applies to any list, from any source.
“Buy Apollo leads” and the Apollo lead database
People searching for “buy apollo leads” usually want Apollo’s data without an Apollo subscription. Two things to know. Apollo’s terms prohibit selling or distributing its contact data to third parties, so anyone selling “Apollo leads” as a file is breaching those terms, and the files on offer are typically cached exports of unknown age, which is where the worst accuracy complaints in this niche come from. Reviews of done-for-you lead vendors in the Apollo space describe lists that were half invalid and orders that never arrived. The alternative that keeps the data current is to run the search yourself on your own Apollo account, which is free to search, and export it: the Apollo Exporter export tool takes the URL of a search you run, collects the rows through your own logged-in session, verifies the emails and bills per row. It does not sell lists, hold a database of Apollo contacts, or export anything you cannot see yourself. For email-specific extraction and phone availability, see the Apollo email scraper page.
Practical checklist
Before relying on Apollo data for a campaign: run the search and check the result count and the email-status mix for free; export only what you need, within your plan’s credits or through a verified export; verify every email, separating valid, catch-all and invalid; send to valid addresses first and catch-alls, if at all, from a secondary domain; and re-verify any list older than about 90 days, since job changes are the main reason valid addresses go stale. The methods for getting data out of Apollo, with their costs and risks, are compared in how to scrape Apollo.