EmailCheck / Questions问题页 / How accurate is bulk email verification?
How accurate is bulk email verification?批量邮箱验证到底有多准?
Per address, exactly as accurate as a single check — the checks are identical. What changes in bulk is the mix: on a real B2B list, roughly 30–40% of addresses usually sit on catch-all domains, where no verifier can give a definite yes or no. So the honest output of a bulk run is a segmented list — valid, risky and invalid, each row scored — not one accuracy percentage. Measure accuracy on your own list: send to the valid segment first and compare hard bounces band by band.
逐个地址看,和单查一样准——跑的检查完全相同。批量时变的只是名单构成:真实 B2B 名单里通常有约 30–40% 的地址在 catch-all 域名上,那部分任何验证工具都给不出确定答案。所以一次批量跑诚实的产出是一份分好段的名单——valid、risky、invalid,每行带分数——而不是一个准确率百分比。准确率要到你自己的名单上量:先发 valid 段,再按分数段对比硬退信。
Why a bulk run has no single accuracy number为什么批量跑没有单一准确率
Single-check benchmarks already cap out lower than the marketing suggests: in Hunter's public benchmark — 3,000 real business emails run through 15 verification services — the best service scored 70%. A bulk run inherits that per-address ceiling and adds a second problem: composition. Two lists run through the same verifier will report different "accuracy" purely because one has more catch-all domains, more role accounts or more typos in it. The verifier didn't change; the list did.
That is why a single percentage for a bulk job is the wrong unit. The useful output is the distribution: how many rows came back valid, how many risky and why (catch-all risk, disposable domain, a typo suggestion), and how many invalid. Those bands are what you actually act on.
单查的基准本来就比营销话术低:在 Hunter 的公开基准里——3,000 封真实商务邮箱、跑 15 家验证服务——最好的一家也只有 70%。批量跑继承这个逐地址天花板,还多一个问题:构成。两份名单跑同一个验证工具,报出来的「准确率」会不同,纯粹因为其中一份 catch-all 域名更多、角色账号更多或拼写错误更多。变的不是工具,是名单。
所以给一次批量任务报单一百分比是用错了单位。有用的产出是分布:多少行 valid、多少行 risky 以及原因(catch-all 风险、一次性域名、拼写纠正建议)、多少行 invalid。这些分段才是你真正要动手处理的东西。
How to measure accuracy on your own list怎么在自己的名单上量准确率
- Seed it first. Add a few addresses you already know are good and a few you know are dead. If the verifier mislabels the seeds, stop there.
- Keep the bands separate. Don't merge risky back into valid to make the list look bigger — the bands are the measurement.
- Send in order. Mail the valid band first, the risky band later and separately, and never mail the invalid band.
- Compare hard bounces per band. After the send, line the bounce rate up against the score bands. A verifier is accurate *for your list* when the bands predict the bounces: near-zero hard bounces in valid, visibly more in risky.
- 先埋种子。放几个你已知是好的地址、几个已知是死的。验证工具把种子都标错,就不用往下走了。
- 分段别合并。不要为了让名单看起来大,把 risky 并回 valid——分段本身就是测量。
- 按顺序发。先发 valid 段,risky 段晚些单独发,invalid 段永远不发。
- 按段对比硬退信。发完之后,把退信率和分数段对齐。对你的名单来说准的验证工具,应该是分段能预测退信:valid 段硬退信接近零,risky 段明显更多。
What EmailCheck returns per rowEmailCheck 逐行返回什么
Every row gets the same domain-level checks as a single lookup — format, domain and MX health, disposable and role-account signals, a typo suggestion when one is likely, domain age, and catch-all risk — combined into a 0–100 score and a valid / risky / invalid verdict, with the fired signals named. The bulk email checker runs a whole list in one call (up to 100 addresses per batch request) and returns the rows in order, so the bands survive the trip. Rows whose signals run out — typically catch-all domains — come back scored as risky, not silently passed as valid.
每一行跑的域名级检查和单查完全相同——格式、域名与 MX 健康、一次性与角色账号信号、疑似拼写时的纠正建议、域名年龄、catch-all 风险——合成 0–100 分和 valid / risky / invalid 裁决,并写明触发的信号。批量检查器一次调用跑完整个名单(每批最多 100 个地址),按原顺序逐行返回,分段不会在路上丢掉。信号到头的行——通常是 catch-all 域名——会以 risky 分数返回,而不是悄悄当成 valid 放行。
Try it on a real address拿真实地址试一下
Live call to the production API — the same check described on this page.
实时调生产 API——就是本页说的这个检查。
What gets checked: address format, domain and MX health, disposable and role-account signals, likely typos with a correction, domain age from registration records, and catch-all risk — combined into a 0–100 score and a valid / risky / invalid verdict.
查什么:地址格式、域名与 MX 健康、一次性与角色账号信号、疑似拼写错误及纠正建议、来自注册记录的域名年龄、catch-all 风险——合成 0–100 分与 valid / risky / invalid 裁决。
Related questions相关问题
- How accurate are email verification APIs, really?
- Bulk email checker — verify a whole list in one call
- How do I verify a list of emails in bulk without calling the API one by one?
Running this in bulk? The 3,500 Credits Pack ($9.99, one-time) runs EmailCheck through the API at 1 credit per check — credits never expire and work across every connected NewKwLab API.
要批量跑?3,500 积分包($9.99,一次性)通过 API 跑 EmailCheck,每次扣 1 分——积分永不过期,在全部已接入的 NewKwLab API 通用。