Tool AT-W31REV. 2026-A
E-mail header analysis: why did the mail land in spam or arrive late?
Why a mail landed in spam or arrived half an hour late is almost always written in its header. It is hard to read: every server adds a line, the newest at the top. The header analysis breaks the header down, works out the delay per hop and translates SPF, DKIM, DMARC, ARC and the spam scores of Microsoft 365 and SpamAssassin into plain English. You paste the header; your browser does the rest.
- free
- no sign-up
- reads only in your browser
01 / ANALYSIS
Paste the header, read the cause
Analysis · header of a received e-mail
The analysis reads what is in the header of the mail and explains it. It does not recalculate signatures, and it cannot see what a mailbox provider does not write into the header. It is built for the question of why a mail landed in spam or arrived late, not for the question of whether a mail is fraudulent.
Findings
Route of the mail
| No. | From → to | Received | Duration | TLS | Type |
|---|
SPF, DKIM, DMARC and ARC
| Check | Result | Domain | Meaning |
|---|
Sender details
Spam scores
Unsubscribe (List-Unsubscribe)
All header lines
| Name | Content |
|---|
Reads only in your browser. The header is not sent and not stored.
02 / HOW TO
How to get the header of the mail
Every e-mail has a header, but no e-mail program shows it by default. What matters is which one you use: the header of the mail as it arrived at the recipient. Almost everything relevant here is missing from the Sent folder, because server stamps and check results are only added along the way.
If your own mails land in spam at a customer’s, ask them for the header. Or send a test mail to your own account at Gmail or Outlook.com and take the header from there. The Microsoft 365 values (SCL, BCL, CAT) are, of course, only visible in a mailbox at Microsoft.
| Program | How to do it |
|---|---|
| Outlook (classic, Windows) | Open the mail with a double-click in its own window, then File > Properties. The header is in the “Internet headers” box. Click into it, Ctrl+A, Ctrl+C. |
| New Outlook and Outlook on the web | Open the mail, click “More actions” (three dots) at the top, then View > View message details. Select all and copy. |
| Gmail | Open the mail, click the three dots next to “Reply”, then “Show original”. It is best to use “Copy to clipboard”; copying the whole page also works. |
| Apple Mail (Mac) | Select the mail, then in the menu View > Message > Raw Source. Select all and copy. |
| Thunderbird | Select the mail, then View > Message Source (Ctrl+U). Select all and copy. |
| Phone | The apps usually do not show the header. Open the mailbox in a browser on a computer, or forward the mail as an attachment to a computer. |
The header contains addresses, IP addresses and the subject. The analysis reads it only in your browser. If you send it to someone, feel free to shorten the subject first; it does not matter for the evaluation.
Sources: Microsoft Support: view internet message headers in Outlook · Google Help: trace e-mails with full headers
03 / ROUTE OF THE MAIL
Read the Received lines from bottom to top
Every server that accepts a mail puts a Received line at the top of the header (RFC 5321, section 4.4). It says who it received the mail from, who it is itself, how the connection went and when. The bottom line is therefore the first hop, the top line the last. From the timestamps of two neighbours you can tell how long the mail sat between them.
In the example on the right, a customer sends a mail at 10:02:08 via the GMX webmailer. GMX takes over after 1 s, and the car dealership’s server only after 32 min. It states the reason itself: greylisting for 31 min.
A 2005 study involving test mails sent to thousands of domains showed why measuring is worthwhile: some were lost without any feedback, and some mails took days Afergan and Beverly 2005. Today, the most common reason for a noticeable delay is less serious.
Typical delays and how to recognise them
| Cause | In the header | Duration |
|---|---|---|
| Greylisting at the recipient | Large gap at the first hop at the recipient, sometimes with the line X-Greylist | Minutes to hours; window under RFC 6647 from one minute to 24 hours |
| Retry after a temporary error | Large gap at a handover, with no further hint | At least 30 minutes per attempt under RFC 5321; the sender gives up after 4 to 5 days |
| Queue or filter | Gap between two hops of the same system | Seconds to hours |
| Quarantine at Microsoft 365 | SFV:SKQ in X-Forefront-Antispam-Report | until released |
| Outbox at the sender | Date is well before the first Received line | any length |
| Server clocks | Negative duration between two hops | Seconds to minutes, no real delay |
The analysis reports a hop as noticeable from 2 min and as clearly late from 30 min; from there, at least one delivery attempt has usually failed. Sources: RFC 5321 · RFC 6647
Greylisting temporarily rejects mail from unknown senders on the first attempt. Proper servers try again later, many spam programs do not. That is why the method is widespread Pagani et al. 2016. How long the mail then waits, however, is up to the sender: RFC 5321 recommends at least 30 minutes until the next attempt, and especially with greylisting and blocklist messages, most large providers do not behave as they should Li et al. 2026. Half an hour for the first mail from a new sender is therefore normal. If every mail arrives that late, look at the recipient’s server.
Every timestamp comes from a different clock. If two servers differ by a few seconds, this can give a negative duration; clock jumps and drift between computers are more common than one would think Paxson 1998. The analysis reports this as a note and not as a fault of the mail.
Encryption per hop
Whether a stretch was encrypted is stated in the Received line: ESMTPS (the S stands for STARTTLS, RFC 3848), “using TLSv1.3” with Postfix or “version=TLS1_2” with Microsoft. The analysis evaluates this for each hop separately, because an encrypted first hop says nothing about the second. Large providers encrypt by themselves; with smaller servers this worked for only a good third in 2015 Durumeric et al. 2015. Encrypted also does not mean verified: many servers do not check the other side’s certificate at all Foster et al. 2015. If it says TLS 1.0 or 1.1, the server is outdated Holz et al. 2016. Google has required TLS from all senders since 2024.
04 / CHECK RESULTS
Reading SPF, DKIM, DMARC and ARC in the header
What the recipient has checked is written into the Authentication-Results line (RFC 8601). First comes who did the checking, such as mx.google.com; Microsoft leaves this part out. After that come the methods with their result, such as spf=pass, dkim=fail or dmarc=pass. Most recipients really do check, SPF up to 85 per cent Deccio et al. 2021, and act on the result.
| Result | SPF | DKIM | DMARC |
|---|---|---|---|
| pass | The sending server is in the SPF record of the envelope domain. | The DKIM signature is valid: the mail was not changed after signing. | DMARC passed: SPF or DKIM passed for a domain that matches the sender. |
| fail | The sending server is not in the SPF record, and the domain demands rejection (-all). | The DKIM signature no longer matches, usually because the text or subject was changed along the way. | DMARC failed: neither SPF nor DKIM passed for a domain that matches the sender. |
| softfail | The sending server is not in the SPF record; the domain only asks for marking (~all). | – | – |
| neutral | The SPF record makes no statement about this server (?all). | The DKIM signature could not be evaluated (syntax error or similar). | – |
| none | The envelope domain has no SPF record. | The mail carries no DKIM signature. | The sender domain has no DMARC record. |
| temperror | SPF could not be checked because of a temporary DNS error. | The DKIM key could not be retrieved for the time being. | The DMARC record could not be retrieved for the time being. |
| permerror | The SPF record is faulty, for example present twice or with more than ten lookups. | The DKIM check failed permanently, for example because the key is missing in DNS. | The DMARC record is faulty. |
Sources: RFC 8601 · RFC 7208 · RFC 9989
Alignment: the most common reason for dmarc=fail
For DMARC it is not enough that SPF or DKIM passes. The domain for which one of them passed must match the domain in the From address (RFC 9989; what counts is the organisational domain, so newsletter.your-company.de matches your-company.de). This is exactly where newsletter services fail: they send with their own envelope and their own signature, SPF and DKIM pass, but DMARC still does not. The analysis therefore shows the domain next to every result and says whether it matches. What is in your DNS is checked by the E-mail check.
Forwarding, mailing lists and ARC
When a server forwards a mail, such as a club mailing list, both checks often break. SPF then only applies to the forwarder, and if “[Vorstand]” is put in front of the subject, the DKIM signature no longer matches. Measurements on 20 forwarding services show how thoroughly forwarding undermines the assumptions behind SPF, DKIM and DMARC Liu et al. 2023.
ARC (RFC 8617) is the answer: the forwarder seals how the mail arrived at its end, and the next recipient can read that. Gmail and Outlook evaluate ARC, and most implementations work correctly Wang and Wang 2022. There is no guarantee, though. Whether a recipient follows the seal is its own decision; with Gmail you can tell by “dis=NONE” despite p=REJECT, with Microsoft by compauth with reason 130. The analysis checks the structure of the chain (instances, cv values); only the recipient recalculates the signatures themselves.
05 / SPAM SCORES
SCL, BCL and spam points: what the filters write into the header
Some providers write their verdict into the header in a readable form. Microsoft 365 does so in detail, and so do SpamAssassin and Rspamd on the servers of many hosts and companies. Gmail, Apple and GMX/WEB.DE do not, or only in lines whose content they do not describe. The analysis shows such lines raw, without interpreting them.
Microsoft 365: the category decides, not the SCL
In the line X-Forefront-Antispam-Report, Microsoft writes which rule applied (CAT), what the spam filter decided (SFV) and the older spam score SCL from −1 to 9. Since Microsoft has been filtering by category, the SCL in the cloud no longer says whether a mail counts as spam; CAT and DIR are what counts. For bulk mail there is a separate value, the BCL in X-Microsoft-Antispam.
| Value | Meaning |
|---|---|
| SPM | Spam |
| HSPM | Spam with high confidence |
| BULK | Bulk mail |
| SPOOF | Sender not verified (spoofing) |
| PHSH | Phishing |
| HPHSH | Phishing with high confidence |
| MALW | Malware |
| BCL | Meaning |
|---|---|
| 0 | not bulk mail |
| 1 to 3 | Bulk senders with few complaints |
| 4 to 7 | Bulk senders with a mixed number of complaints |
| 8 to 9 | Bulk senders with many complaints |
From BCL 7, Microsoft’s default rule sorts a bulk mail into the Junk folder, with the preset “Standard” from 6, with “Strict” from 5 and then into quarantine. A BCL of 7 on a newsletter therefore means: depending on the recipient’s setting, it lands in the Junk folder or just misses it.
| Value | Meaning |
|---|---|
| BLK | Filter skipped and blocked: the sender is on the recipient’s blocked senders list. |
| NSPM | The spam filter classed the mail as not spam. |
| SFE | Filter skipped: the sender is on the recipient’s safe senders list. |
| SKA | Filter skipped: the sender is on the allowed senders or domains list in the anti-spam policy. |
| SKB | Marked as spam: the sender or domain is on the blocked list in the anti-spam policy. |
| SKI | Filter skipped: the sending IP is on the allowed IP list. |
| SKN | Filter skipped because of a mail flow rule (transport rule). |
| SKQ | The mail was in quarantine and was released from there. |
| SKS | Already marked as spam before the spam filter, by a transport rule or an on-premises Exchange server. |
| SPM | The spam filter marked the mail as spam. |
Added to that is compauth in the Authentication-Results line: Microsoft’s overall verdict on whether the mail comes from the domain in the From address. It combines SPF, DKIM, DMARC and its own signals. A compauth=fail with reason 001 for a domain without a DMARC policy is the typical start of a classification as SPOOF.
| Reason | Meaning |
|---|---|
| 000 | DMARC failed, and the domain requires quarantine or reject. |
| 001 | Implicit fail: the domain has published no authentication or only a weak one (SPF ~all or ?all, DMARC p=none). |
| 100 | SPF or DKIM passed, and the envelope and sender domains match. |
| 101 | DKIM-signed by the domain in the From address. |
| 102 | SPF passed, envelope and sender domains match. |
| 108 | DKIM failed because an earlier, legitimate hop changed the text. |
| 130 | The ARC result of a trusted forwarder overrode the DMARC result. |
| 905 | DMARC not applied because of nested forwarding (for example via an on-premises Exchange or a third-party service). |
Sources, as of 14 November 2026: Microsoft Learn: anti-spam message headers · Microsoft Learn: Spam Confidence Level (SCL) · Microsoft Learn: Bulk Complaint Level (BCL)
SpamAssassin and Rspamd: points against a threshold
SpamAssassin awards points for every rule that applies and adds them up. From the threshold required, 5 by default, X-Spam-Status says “Yes” and the server flags the mail. Rspamd works similarly but has levels: greylisting from one value, a spam flag above it, rejection at the very top. Spam filters hit remarkably well Cormack 2008, but every server sets its own threshold. That is why the analysis always reads the score against the threshold of the filter in question and names the rules that contributed most.
| Rule | Meaning |
|---|---|
| RCVD_IN_PBL | The sending IP is in a range from which, according to the Spamhaus PBL, no mail should come directly, typical of home and DSL connections. |
| RCVD_IN_SBL | A hop is on the Spamhaus SBL (known spam sources). |
| RCVD_IN_XBL | A hop is on the Spamhaus XBL (hijacked computers). |
| RDNS_NONE | The delivering server has no reverse DNS (PTR). |
| SPF_SOFTFAIL | SPF failed (softfail): the sending server is missing from the SPF record. |
| SPF_FAIL | SPF failed (fail). |
| DKIM_INVALID | A DKIM signature is present but invalid. |
| DKIM_VALID_AU | Valid DKIM signature from the sender domain. |
| DMARC_REJECT | DMARC failed; the domain requires reject. |
| DMARC_MISSING | The sender domain has no DMARC record. |
| MISSING_MID | The mail has no Message-ID. |
| MIME_HTML_ONLY | The mail consists of HTML only, with no text version. |
| BAYES_00 … BAYES_999 | Bayes filter: probability of spam, learned from earlier mails on this server. The number in the name is the lower limit in per cent. |
The rules with RCVD_IN and URIBL query blocklists. This has been routine for two decades; as early as 2004, such queries made up a seventh of all DNS queries at MIT Jung and Sit 2004. They hit the route, not the text. Most spam comes from a few areas of the address space Ramachandran and Feamster 2006, and lists based on IP reputation catch the majority before acceptance Esquivel et al. 2010. A copier that sends scans straight from a DSL line therefore lands in spam, however harmless the scan is. The remedy is to send via the provider’s mail server, with authentication.
| Action | Meaning |
|---|---|
| no action | Delivered, without a flag. |
| greylist | Temporarily rejected on the first attempt (greylisting); only arrives on the next attempt. |
| soft reject | Temporarily rejected; the sending server must try again later. |
| add header | Marked as spam; the mailbox usually sorts the mail into the spam folder. |
| rewrite subject | Marked as spam and the subject rewritten. |
| quarantine | Put into quarantine. |
| reject | Rejected. |
| discard | Discarded without telling the sender anything. |
Sources: Apache SpamAssassin: rule files (descriptions of the rules, default threshold 5) · Rspamd docs: actions and thresholds
06 / UNSUBSCRIBE
Unsubscribing with one click (RFC 8058)
Newsletters need an unsubscribe that the mailbox itself can offer. RFC 8058 sets out how this works with one click, and Gmail requires it of everyone who sends more than 5,000 mails a day to Gmail addresses. Without it, recipients are often left with only the spam button. And it is exactly such complaints that Microsoft measures with the BCL.
- List-Unsubscribe with an HTTPS address. A mailto address may also be there, but on its own it is not enough.
- List-Unsubscribe-Post: List-Unsubscribe=One-Click. Exactly like this, with nothing added. The mailbox sends this text to the HTTPS address, with no login and no redirect.
- A valid DKIM signature over both lines. Both must be in the h= of the signature, otherwise the mailbox does not offer the click.
For club mailing lists and similar lists, the Gmail requirement usually does not apply; the analysis reports a missing one-click unsubscribe there only as information. Source: Google: e-mail sender guidelines.
07 / METHOD
What to do with the result
- Blocklist, no reverse DNS, RCVD_IN_PBL. This is down to the sending route. Devices and your own servers should send mail through your provider’s mail server (SMTP with login) instead of directly from the connection.
- dmarc=fail on your own mail. Almost always a service is sending for your domain without DKIM: newsletter, shop, invoicing software. The record is set up at the service and in your DNS; what is in your DNS is shown by the E-mail check.
- Bulk mail (CAT:BULK, high BCL). Keep the recipient list clean, build in one-click unsubscribe, send in a more targeted way. You can ask individual recipients to add you as a safe sender.
- Delay at the handover. Once, after greylisting, is normal. If every mail arrives late, the recipient’s server (exception list) or your own (queue, retry interval) needs a look.
- Nothing found. Gmail and Apple do not write their spam reason into the header. Then a test mail to your own account at Microsoft helps, whose header reveals more.
Where the analysis stops
- It does not recalculate signatures. Whether DKIM and ARC are valid is decided by the recipient with the key from DNS. The analysis reads its verdict and checks the structure.
- It sees only what is in the header. Internal ratings by Gmail, Apple or GMX stay hidden; lines without a public description it shows raw.
- Alignment is simplified. It determines the organisational domain from a short list of endings, not via the DNS lookup from RFC 9989. The recipient’s DMARC result takes precedence.
- No fraud detection. Whether a mail is forged or dangerous is not something it answers.
The background is explained at leisure in the guide Why do my e-mails land in spam?. If you would rather not tackle it yourself: we set up domain and mailboxes as part of our IT support, and connect newsletter and shop services at the same time. With ongoing IT support, we keep an eye on delivery.
08 / FOR AI ASSISTANTS
Using this tool from an AI assistant
Assistants such as Claude or ChatGPT that speak the Model Context Protocol (MCP) can call this page’s calculation directly instead of rebuilding it. The server is public, needs no sign-in and stores nothing. The answer always includes the link to this page, the date of the data and the limits of the result.
- Endpoint
https://aureum-tech.com/mcp- Tool
werkzeug_e_mail_header_analyse- Transport
- Streamable HTTP, read-only
| Name | Required | Meaning |
|---|---|---|
kopfzeilen | yes | The complete header of a received e-mail as text, as shown by Outlook (“Internet headers”), Gmail (“Show original”) or Apple Mail (“Raw Source”). A whole .eml file also works; the content after the first blank line is ignored. |
alle_felder | no | Optional: true additionally returns all header lines with decoded values (subject, names under RFC 2047). Default false. |
Questions where an assistant should call it
- “Why does our newsletter land in the Junk folder for customers on Microsoft 365? Here is the header.”
- “This mail arrived half an hour late. What caused that? (header pasted)”
- “What do SCL 5, BCL 7 and compauth=fail reason=001 mean in this mail header?”
- “Why does Gmail show dmarc=fail although the mail came through our club mailing list?”
How the connection works and what MCP can do in your own business is explained on the page about MCP integrations.
09 / COMMON QUESTIONS
Common questions about the e-mail header
What is in the header of an e-mail?
The header sits above the text of every e-mail. It holds the sender, recipient, subject and date, plus a Received line from every server that passed the mail on, with the time and the encryption used. The recipient’s server adds what it checked (SPF, DKIM, DMARC), and many spam filters add their verdict. E-mail programs hide all this, but every one of them can show it.
How do I find the header in Outlook?
In classic Outlook, double-click the e-mail to open it and choose File > Properties; the header is in the “Internet headers” box. In new Outlook and Outlook on the web, click the three dots at the top and choose View > View message details. In Gmail the option is called “Show original”. Always take the header of the received e-mail, not the one from the Sent folder.
Why does my e-mail land in spam although SPF, DKIM and DMARC pass?
Passed checks only prove that the mail comes from your domain. Whether it is wanted is decided by filters based on reputation and content: blocklists for the sending IP, links to suspicious domains, many complaints about bulk mail, or a Bayes filter that takes the text for spam. Microsoft writes this into the header as CAT and BCL, SpamAssassin as rules with points. Gmail does not give the reason in the header.
What do SCL and BCL mean in Microsoft 365?
SCL (Spam Confidence Level) runs from −1 to 9; from 5 a mail is generally considered bad. In Microsoft 365, however, it is no longer the SCL but the category CAT that decides whether a mail counts as spam. BCL (Bulk Complaint Level) measures from 0 to 9 how many complaints a bulk sender generates. From 7 the default rule sorts the mail into the Junk folder, with stricter settings from 6 or 5.
Why did an e-mail arrive half an hour late?
Usually because of greylisting: the recipient’s server temporarily rejects mail from unknown senders on the first attempt and only accepts it on the second. When that second attempt comes is decided by the sending server; the standard recommends at least 30 minutes between attempts. In the header you see this as a large gap between two Received lines, sometimes with an X-Greylist line. For the first mail from a new sender this is normal.
Why does the analysis show a negative delay?
Every Received line carries the time of the server that wrote it. If two servers’ clocks differ by a few seconds or minutes, the later hop is, on paper, before the earlier one. This says nothing about the mail. The analysis reports it as a note once the difference reaches a minute.
What does dmarc=fail mean for a forwarded mail?
Forwarding and mailing lists often break both checks: SPF then only applies to the forwarder, and a changed subject invalidates the DKIM signature. DMARC then fails even though the original sender set everything up correctly. With ARC the forwarder seals the original result; Gmail and Microsoft can follow it, but do not have to. A mailing list that rewrites the sender or leaves the mail unchanged is safe.
Where does the header analysis stop?
It reads and explains what is in the header, and only in your browser. It does not recalculate signatures, and it cannot see what a provider does not write into the header. It is not built to tell whether a mail is fraudulent. If your own mails land in spam at your customers’, we check your domain, sending services and mail servers and set them up correctly, on site in the Vogtland or remotely.
10 / STUDIES AND SOURCES
What the tool is based on
The guide values and notes above are based on these 14 scientific papers. Each one has been checked via its DOI or the publisher’s page (most recently on 14 November 2026): it exists, it has not been retracted, and its findings still hold for this question today.
Afergan, M.; Beverly, R. (2005): The state of the email address. ACM SIGCOMM Computer Communication Review 35 (1), pp. 29–36
Finding For a month, test mails were sent to popular, random and Fortune 500 domains and traced like a traceroute. Result: noticeable losses without any feedback, transit times of up to several days and surprising errors.
In the tool Explains why the analysis measures each hop separately: that mail “just works” is not a given.
doi.org/10.1145/1052812.1052822 · Journal article
Li, R.; Li, E.; Liu, M. et al. (2026): When Delivery Meets Error: Exploring Email Delivery Retry Strategies and Defects. Proceedings of the 2026 ACM Internet Measurement Conference, pp. 487–494
Finding Examined how 11 mail providers and 8 free mail servers redeliver after errors. Many implementations have flaws: for blocklist and greylisting messages, most providers do not retry appropriately, and six systems spread the attempts poorly across the MX servers over IPv4 and IPv6.
In the tool Explains in the result why a delay at the recipient often depends on the retry behaviour of the sending server and not only on the block itself.
doi.org/10.1145/3777912.3839798 · Conference paper
Pagani, F.; De Astis, M.; Graziano, M. et al. (2016): Measuring the Role of Greylisting and Nolisting in Fighting Spam. 46th Annual IEEE/IFIP International Conference on Dependable Systems and Networks (DSN), pp. 562–571
Finding First comprehensive measurement of greylisting and nolisting: worldwide spread, effect against spam malware and consequences for normal mails. Both methods exploit the fact that spam programs often do not follow the retry rules of SMTP.
In the tool Greylisting is no rarity: the analysis therefore checks every delay at the handover to the recipient to see whether it fits a greylisting pause.
doi.org/10.1109/DSN.2016.57 · Conference paper
Paxson, V. (1998): On calibrating measurements of packet transit times. Proceedings of ACM SIGMETRICS 1998, pp. 11–21
Finding Transit times measured with two separate clocks are error-prone: in a large number of TCP connections, clock jumps and drift between sender and receiver were so frequent that they lead undetected to large measurement errors.
In the tool Every Received line comes from a different clock. Negative or implausible intervals are therefore reported by the analysis as clock deviation, not as a fault of the mail.
doi.org/10.1145/277851.277865 · Conference paper
Durumeric, Z.; Adrian, D.; Mirian, A. et al. (2015): Neither Snow Nor Rain Nor MITM… An Empirical Analysis of Email Delivery Security. Proceedings of the 2015 Internet Measurement Conference, pp. 27–39
Finding The large providers encrypt and authenticate mail by themselves. Of over 700,000 other mail servers, only 35 per cent set up encryption successfully; because SMTP falls back to unencrypted delivery if necessary, over 20 per cent of incoming Gmail messages arrived in plain text in seven countries.
In the tool That is why the analysis shows TLS for each hop separately: an encrypted first hop says nothing about the next.
doi.org/10.1145/2815675.2815695 · Conference paper
Foster, I. D.; Larson, J.; Masich, M. et al. (2015): Security by Any Other Name: On the Effectiveness of Provider Based Email Security. Proceedings of the 22nd ACM SIGSAC Conference on Computer and Communications Security, pp. 450–464
Finding Measured at every hop of the mail route: more than half of the 20,000 largest receiving servers supported TLS but did not check certificates. SPF was widespread but rarely enforced; few senders signed with DKIM, and even fewer recipients rejected invalid signatures.
In the tool Justifies two limits in the result: TLS in the header means encrypted, not verified, and a failure of SPF or DKIM has very different consequences depending on the recipient.
doi.org/10.1145/2810103.2813607 · Conference paper
Holz, R.; Amann, J.; Mehani, O. et al. (2016): TLS in the Wild: An Internet-wide Analysis of TLS-based Protocols for Electronic Communication. Network and Distributed System Security Symposium (NDSS) 2016
Finding Internet-wide scans and passive measurement of mail and chat servers, including between servers: large providers protect their users well, but most traffic in transit is weakly secured, with flaws in cryptography and authentication.
In the tool The analysis marks hops with TLS 1.0, 1.1 or SSL as outdated, rather than counting every encryption as equally good.
doi.org/10.14722/ndss.2016.23055 · Conference paper
Deccio, C.; Yadav, T.; Bennett, N. et al. (2021): Measuring email sender validation in the wild. Proceedings of the 17th International Conference on emerging Networking EXperiments and Technologies (CoNEXT), pp. 230–242
Finding First large measurement of which receiving servers actually check senders: up to 85 per cent check SPF, more than half all three methods SPF, DKIM and DMARC, in part with deviations from the SPF standard.
In the tool That is why the analysis gives the check results in the header great weight: most recipients act on them, and a failure has consequences.
doi.org/10.1145/3485983.3494868 · Conference paper
Liu, E.; Akiwate, G.; Jonker, M. et al. (2023): Forward Pass: On the Security Implications of Email Forwarding Mechanism and Policy. 2023 IEEE 8th European Symposium on Security and Privacy (EuroS&P), pp. 373–391
Finding Measurement of 20 forwarding services (16 large providers, four mailing list services): automatic forwarding and mailing lists break the assumptions on which SPF, DKIM and DMARC rest.
In the tool If the analysis detects forwarding or a mailing list, it explains a DMARC failure with that first, and not with a wrong record at the sender.
doi.org/10.1109/EuroSP57164.2023.00030 · Conference paper
Wang, C.; Wang, G. (2022): Revisiting Email Forwarding Security under the Authenticated Received Chain Protocol. Proceedings of the ACM Web Conference 2022, pp. 681–689
Finding Analysis of 600,000 mails and controlled experiments: ARC is not yet widespread but is used by large providers such as Gmail and Outlook; most implementations are correct, individual providers misread ARC results.
In the tool The analysis shows ARC as the forwarder’s second opinion and adds that the recipient decides for itself whether to follow it.
doi.org/10.1145/3485447.3512228 · Conference paper
Jung, J.; Sit, E. (2004): An empirical study of spam traffic and the use of DNS black lists. Proceedings of the 4th ACM SIGCOMM Conference on Internet Measurement, pp. 370–375
Finding Recordings at MIT from 2000 and 2004: mail volume rose by 866 per cent, and queries to DNS blocklists made up about 14 per cent of all DNS queries in 2004, against 0.4 per cent in 2000.
In the tool Blocklists have been routine in mail reception for two decades. If a blocklist rule fires, the analysis therefore puts it first.
doi.org/10.1145/1028788.1028838 · Conference paper
Ramachandran, A.; Feamster, N. (2006): Understanding the network-level behavior of spammers. Proceedings of ACM SIGCOMM 2006, pp. 291–302
Finding Over ten million spam mails from 17 months: most spam comes from a few areas of the IP address space, often from short-lived hijacked computers. Filters based on network features are more stable than filters based on content.
In the tool Explains why a copier or server that sends directly from a DSL line lands in spam because of its IP, whatever the mail says.
doi.org/10.1145/1159913.1159947 · Conference paper
Esquivel, H.; Akella, A.; Mori, T. (2010): On the effectiveness of IP reputation for spam filtering. 2010 Second International Conference on Communication Systems and Networks (COMSNETS), pp. 1–10
Finding Distinguishes filters before acceptance (IP reputation) and after (content). Proprietary reputation lists, built among other things from SPF records, correctly identified around 90 per cent of spam and legitimate mail; but the lists have to be renewed continuously.
In the tool The analysis separates causes in reputation (IP, blocklist, reverse DNS) from causes in content, because they are fixed in different places.
doi.org/10.1109/COMSNETS.2010.5431981 · Conference paper
Cormack, G. V. (2008): Email Spam Filtering: A Systematic Review. Foundations and Trends in Information Retrieval 1 (4), pp. 335–455
Finding Overview of spam filters and how well they work: filters are remarkably accurate even though spam can hardly be defined formally. The filter is only one building block in a larger system of users, senders and services.
In the tool That is why the analysis always reads a spam score against the threshold of the filter in question and names the rules that carried it, instead of inventing an overall grade of its own.
doi.org/10.1561/1500000006 · Review
Standards, authorities and manufacturer data
- RFC 5322: Internet Message Format · RFC Editor (IETF)
- RFC 5321: Simple Mail Transfer Protocol · RFC Editor (IETF)
- RFC 3848: ESMTP and LMTP Transmission Types · RFC Editor (IETF)
- RFC 2047: encoded words in header lines · RFC Editor (IETF)
- RFC 8601: Authentication-Results · RFC Editor (IETF)
- RFC 8617: Authenticated Received Chain (ARC) · RFC Editor (IETF)
- RFC 9989: DMARC (replaces RFC 7489) · RFC Editor (IETF)
- RFC 7208: Sender Policy Framework (SPF) · RFC Editor (IETF)
- RFC 6376: DKIM signatures · RFC Editor (IETF)
- RFC 8301: rsa-sha1 retired in DKIM · RFC Editor (IETF)
- RFC 8058: one-click unsubscribe · RFC Editor (IETF)
- RFC 8314: TLS for mail (clause “tls” in Received) · RFC Editor (IETF)
- RFC 8996: TLS 1.0 and 1.1 deprecated · RFC Editor (IETF)
- RFC 6647: greylisting · RFC Editor (IETF)
- Microsoft Learn: anti-spam message headers · Microsoft
- Microsoft Learn: Spam Confidence Level (SCL) · Microsoft
- Microsoft Learn: Bulk Complaint Level (BCL) · Microsoft
- Microsoft Support: view internet message headers in Outlook · Microsoft
- Google Help: trace e-mails with full headers · Google
- Google: Email sender guidelines · Google
- Apache SpamAssassin: rule files · Apache Software Foundation
- Rspamd docs: actions and thresholds · Rspamd project
11 / MORE TOOLS
Still have a question?
- AT-W2 · Home & business
Windows 10 check
Does this PC still get security updates, and what needs doing by October 2027? - AT-W3 · Home & business
Wi-Fi planner
How many access points or mesh units does your home, office or warehouse need? - AT-W4 · Businesses
IT emergency check
How quickly could your business get back to work after an outage? - AT-W5 · Businesses
E-mail check
Are your e-mails landing in spam? Your domain’s SPF, DKIM and DMARC in plain language. - AT-W6 · Businesses
UPS calculator
Which UPS keeps your server, router and till running long enough during a power cut? - AT-W7 · Home
Digital legacy planning sheet
Which accounts, devices and contracts should your family know about?
You now know what it hinges on and want to get rid of it? Send us the copied result. We set up domain and sending services so that your mails arrive, in Rodewisch and throughout the Vogtland.
+49 3744 4399760[email protected]Mon–Fri 8:00–17:00 · Straße der Jugend 5, 08228 Rodewisch