资源 · 34

提高电子邮件发件人的信任度和送达率

检查发件人身份、邮件类别、退订行为和拒收原因。

已更新 · 16 min

无人设备间内的通信机架与网络布线 由人工智能生成的编辑插图

本指南有助于实现的目标

  • 映射流程
  • 验证域
  • 启用撤回功能
  • 诊断拒收原因

快速检查

  • 哪些系统使用该域发送邮件?
  • 实际邮件是否通过了预期的 SPF、DKIM 和 DMARC 检查?
  • 每封邮件是促销邮件还是事务邮件?
  • 退订功能是否正常?
  • 退订邮件是否按原因分组?

分步指南

  1. 1

    发送清单

    区分新闻简报、自动化邮件、收据和提醒邮件。 记录每个流程的域名、提供商、流量、用途和所有者。

    可交付成果:流程和域名映射。

  2. 2

    检查身份验证

    对已交付的样本测试 SPF、DKIM、DMARC 和发件人域名对齐。 预期的 DNS 配置并不能证明每个发送路径都有效。

    可交付成果:检查的邮件头和身份验证结果。

  3. 3

    符合收件人预期

    检查注册、发送频率、承诺的内容以及促销流量和事务流量的分离情况。 避免在服务邮件中出现意外的促销信息。

    可交付成果:按邮件类别划分的规则。

  4. 4

    测试退订

    对于适用的促销邮件,验证可见链接和一键式行为是否符合提供商要求和 RFC 8058。测试退订是否实际生效。

    可交付成果:端到端退订证据。

  5. 5

    监控信誉

    跟踪永久性和临时性拒收、投诉以及收件人-提供商信号。 按流程和提供商进行分析,而非使用全局平均值。

    可交付成果:未送达原因分析。

  6. 6

    修复并重新测试

    首先处理身份验证、不合格列表、技术错误和退订功能失效等问题。 根据实际情况逐步增加处理量。

    可交付成果:纠正计划和观察到的结果。

管理指标

指标衡量内容首要行动
身份验证包含预期 SPF/DKIM/DMARC 结果的真实邮件修复发送路径
拒收按流程统计拒收率和拒收原因清理或修复邮件列表
撤回已应用促销请求修复邮件头和处理问题
投诉按活动跟踪报告修改目标定位和发送频率

常见错误

  • 假设已发布的 DNS 可证明每个流程
  • 混合事务性邮件和促销邮件
  • 混淆打开状态和收件箱位置
  • 显示取消订阅链接但实际上并未移除收件人

常见问题解答

仅 SPF 是否足够?

否。 收件人要求因邮件量和上下文而异;还应检查 DKIM、DMARC 和适用的对齐方式。

每封邮件都需要一键取消订阅吗?

提供商规则区分促销邮件和事务性邮件,因此请先对每条邮件流进行分类。

打开状态是否能证明邮件已送达?

否。打开状态指标有其自身的局限性,不能替代送达结果。

邮件安全:身份、对齐与传输

此语言版本的详细参考说明以英语提供。

Domain authentication, transport confidentiality and inbox placement are different questions. SPF, DKIM and DMARC inform domain identity; MTA-STS and DANE concern server-to-server connections; reports help observe failures. None of these signals proves a message is benign.

References and status checked on .

Which mechanism answers which question?

Which mechanism answers which question?
Mechanism and sourceWhat is checkedWhere to find evidenceLimit to retain
SPF

SPF · RFC 7208

Authorisation of the sending IP for the SMTP envelope domain; DMARC uses MAIL FROM.The evaluated domain’s TXT record and receiver result. Evaluation allows 10 DNS-querying terms, rather than simply 10 network requests.Forwarding may change the observed IP. Selecting multiple SPF records produces permerror; an SPF pass does not prove alignment with From.
DKIM

DKIM · RFC 6376

DKIM cryptography · RFC 8301

Cryptographic verification of signed message parts for the signature’s d= domain.DKIM-Signature, selector s=, domain d= and key under selector._domainkey. RFC 8301 requires RSA ≥ 1,024 bits, recommends ≥ 2,048 bits and excludes rsa-sha1.A valid signature from another domain may not satisfy DMARC. Changes to signed parts can break verification; DKIM does not encrypt the message.
DMARC

DMARC · RFC 9989

Links the visible From domain to at least one SPF- or DKIM-authenticated, aligned identifier.The TXT policy discovered through _dmarc and trustworthy results produced at reception. Record alignment mode and the domain actually evaluated.A pass neither guarantees delivery nor endorses content. Receivers retain local policy; DNS errors and non-conforming messages require separate handling.
DMARC aggregate / failure reports

DMARC aggregate reports · RFC 9990

DMARC failure reports · RFC 9991

Observation of streams using the domain: aggregate statistics and, separately, details about failing messages.rua / ruf destinations, covered period and received reports. Their formats are now specified in separate RFCs published in May 2026.Missing reports do not prove no mail was sent. Detailed reports may expose message data; define suitable access, minimisation and retention.
MTA-STS

MTA-STS · RFC 8461

Recipient-domain policy for SMTP TLS transport and expected MX hosts._mta-sts TXT, HTTPS policy at /.well-known/mta-sts.txt and certificate. Three modes: testing, enforce and none.Applies to supporting servers and transport, without end-to-end encryption. Initial discovery and caching have their own limits.
TLS-RPT

SMTP TLS Reporting · RFC 8460

Reporting of successful and failed TLS negotiation or validation between servers._smtp._tls TXT, reporting destinations and received JSON data, correlated with transport logs.Observation does not enforce encryption. TLS reports and DMARC domain-identity reports describe different events.
DANE for SMTP

DANE for SMTP · RFC 7672

TLS transport authentication using DNSSEC-protected TLSA records.Validated DNSSEC chain, SMTP service TLSA record and a match with the presented certificate or key.DNSSEC validation is required on this path. DANE does not validate From or replace SPF, DKIM and DMARC.
ARC

ARC · RFC 8617

Carries authentication assessments along paths involving intermediaries.ARC header sets, chain validation and a trust decision about intermediaries.RFC 8617 is Experimental. ARC is not a trust framework and does not automatically make a message safe or deliverable.

DMARC in 2026: documentation changes

Reference
RFC 9989, published in May 2026, replaces RFC 7489 and RFC 9091. Aggregate and failure reporting are separated into RFC 9990 and RFC 9991.
Discovery and alignment
The specification defines DNS Tree Walk discovery. Do not infer organisational domains by naively retaining the final two labels.
Testing a deployment
pct is removed; the new t=y requests handling one level below the published failure policy: reject becomes quarantine, and quarantine becomes none. t=n is the default.
Interoperability
The standard advises against p=reject for general-purpose email domains. Inventory forwarding and mailing lists and check actual implementations before changing policy.

Publication of an RFC does not show that every provider already implements that version. Check documentation, observed results and exceptions in your own streams. These examples are educational and provide no ready-to-deploy DNS record. RFC 9989 →

Understand DMARC logic with a scenario

Simplified model: a conforming From message, an already discovered DMARC policy, one SPF MAIL FROM result and one examined DKIM signature. You supply alignment based on the evaluated domains and mode. The tool performs no DNS resolution or signature verification and predicts no delivery.

(SPF pass AND aligned) OR (DKIM pass AND aligned)

With multiple DKIM signatures, check each result / d= pair: one valid, aligned signature can suffice. Never combine one signature’s pass with an aligned domain from another failing signature. DNS discovery errors, absent policy and invalid From fields are outside this model. All choices stay in your browser.

Six educational cases, readable without JavaScript

Six educational cases, readable without JavaScript
ScenarioSPF: result / alignmentDKIM: result / alignmentExpected interpretation
Aligned SPF is sufficientpass — authenticated
Yes
fail — observed failure
No
pass — at least one authenticated and aligned path

The SPF path meets both conditions.

Aligned DKIM is sufficientfail — observed failure
No
pass — authenticated
Yes
pass — at least one authenticated and aligned path

The DKIM path meets both conditions independently of the other path in this model.

Two passes without alignmentpass — authenticated
No
pass — authenticated
No
fail — no authenticated and aligned path in this scenario

Two third-party domains passing technically does not validate From.

Aligned domains, no authenticationfail — observed failure
Yes
fail — observed failure
Yes
fail — no authenticated and aligned path in this scenario

A matching domain does not replace successful authentication.

Incomplete evidenceResult not supplied
Alignment not supplied
fail — observed failure
No
Undetermined — complete the observations

Missing evidence on one path cannot be invented.

Proven DKIM, unknown SPFResult not supplied
Alignment not supplied
pass — authenticated
Yes
pass — at least one authenticated and aligned path

One complete positive result suffices even when the other path is unknown.

Prepare an actionable email review

Prepare an actionable email review
StreamEvidence to correlateUseful decision
Alerts, receipts and resetsProvider and From domain, envelope and DKIM identities, tracking ID, SMTP response and available delivery event.Separate transport failure, authentication error and absence of reading. Define an alternative for a critical service.
Campaigns and newslettersDomains and subdomains, alignment results, representative samples from every platform, withdrawal mechanism and stream-specific errors.Repair the affected sending path before changing a global policy; do not equate authentication with recipient expectations.
Forwarding and mailing listsPath before and after the intermediary, message changes, SPF/DKIM outcomes and ARC information when present.Explain the observed break and local policy; do not assume a valid ARC chain compels acceptance.
Transport to MX hostsRelevant MTA-STS policy or TLSA, certificates, DNSSEC validation for DANE, TLS reports and timestamped logs.Associate failures with the relevant host and period. Test changes within authorised scope and retain rollback.

Continue with domain and DNS governance →

Primary sources and publication dates

Dates below identify RFC publication months, separately from the consultation date. Descriptions and cases are editorial summaries rather than complete normative extracts.

官方参考文献

参考文献支持该方法。 请根据您的实际情况调整检查;它们并非认证。 原始参考文献标题和源文档可能使用其他语言。