Qvantor logo QvantorInfrastructure, explained plainly
Email Infrastructure

Why Email Forwarding Breaks SPF and What ARC Fixes

A reader wrote to me in October about invoices that reached half his list and vanished for the rest. The pattern looked random until we sorted the failures by recipient domain.

A desk with a laptop showing email authentication headers beside a printed diagram of a forwarded message path

A reader wrote to me in October about invoices that reached half his list and vanished for the rest. The pattern looked random until we sorted the failures by recipient domain. Every missing message had gone to someone with a forward in place: a university alumni address pointing at Gmail, a shop keeping an old mailbox alive at a provider it left years ago, two staff at a company that routes everything through a filtering gateway before final delivery. Those invoices passed SPF, DKIM and DMARC on the way out of his server. They failed DMARC on the way in.

Forwarding is the one case where honest mail fails authentication by design. Nothing is misconfigured, no attacker is involved, and the sender sees a clean report on his own side. I burned most of a week on this in 2022 before I accepted that the repair is not on the sending end at all. What follows is what breaks, what survives, what the Authenticated Received Chain actually does about it, and the one piece of advice I wish people would stop repeating in forums.

The envelope changes hands and SPF notices

SPF does not check the address your reader sees. It checks the envelope sender, the address given in the MAIL FROM command during the SMTP conversation, against the list of IP addresses your DNS record authorises. Those two addresses are usually the same for mail you send directly, which is why the distinction almost never matters until it does.

A plain forward keeps the envelope sender intact and sends the message on from a different machine. Gmail receives a message claiming MAIL FROM at your domain, arriving from a university mail relay that your SPF record has never heard of, and records softfail or fail. The content is untouched. The verdict is still negative, because the only question SPF can answer is whether that IP was allowed to use that envelope address, and it was not.

Mailing lists behave differently and usually come out better. Most list software rewrites the envelope sender to its own bounce address, so SPF passes for the list rather than for you. That is called sender rewriting, and it is the reason your mail to a discussion list rarely has this problem while a one line forwarding rule does.

DKIM survives the trip more often than people expect

DKIM signs a hash of the body and a chosen set of headers with a private key, and publishes the matching public key in your DNS. A forward that copies bytes faithfully leaves that signature valid, which is why DMARC can still pass on alignment through DKIM even after SPF has failed. In practice the plain Gmail or Fastmail style forward preserves DKIM somewhere around nine times out of ten in the traces I have looked at.

The failures cluster around middleboxes that edit the message. A gateway appending a legal disclaimer to the body breaks the signature instantly. So does a scanner that rewrites links to its own click tracking host, a system that changes the subject line to add a tag like [EXTERNAL], and the long standing habit some filters have of re-encoding the body from quoted-printable to base64. Each of those is a byte level change inside something you signed, and the hash no longer matches.

Two things reduce that risk and cost nothing. Sign a short, stable header list and skip volatile ones. Avoid the l= tag that limits how much of the body gets signed, because it invites a different problem where someone appends content below your signed portion and keeps the signature valid.

What ARC adds, and who decides whether it counts

ARC exists to carry an earlier authentication verdict across a hop that would otherwise destroy the evidence. When a forwarder supports it, it stamps three headers on the message before passing it along: ARC-Authentication-Results, which records what it saw when the message arrived, ARC-Message-Signature, which signs the message as it looked at that moment, and ARC-Seal, which signs the chain itself so later hops can tell whether anyone tampered with the record.

Add a second forwarder and you get a second instance of all three, numbered i=1, i=2 and upward. The final receiver can walk the chain backwards and read a sentence close to this one: Gmail is telling me that when this message reached Gmail it passed SPF and DKIM for your domain, and Gmail is willing to sign its name to that claim.

Here is the part that disappoints everyone. ARC is a statement, not a guarantee. The receiving system has to trust the forwarder that made the claim before the claim carries any weight, and nobody publishes the list of forwarders they trust. Google, Microsoft and Fastmail generally honour chains from each other and from well behaved large relays. A small institutional server that nobody has a reputation record for can seal a chain perfectly and still have it ignored. You cannot publish anything in your own DNS that makes receivers trust your forwarders.

Stop adding the forwarder to your SPF record

The advice you will find in half the threads on this subject is to look up the forwarding server and add its IP range to your SPF record. I understand the reflex. It is still the wrong move, for three reasons that compound.

You do not control that server, so its address range changes without telling you and your record silently goes stale. Every include you bolt on eats into the ten DNS lookup ceiling, and a record that trips that limit returns permerror, which is treated worse than a plain fail. Worst of all, you have now authorised an unrelated machine to send mail claiming to be from your domain, which hands a usable asset to whoever eventually gets access to it.

There is also no end to the exercise. Forwards are created by recipients, one at a time, without telling you. The population of servers forwarding your mail on any given Tuesday is unknowable, so a record built to cover them is a record built to be wrong.

What you see in the headersWhat actually happenedReasonable response
spf=fail, dkim=pass, dmarc=passA clean forward, DKIM carried alignmentNothing. This is the system working.
spf=fail, dkim=fail, no ARC headersA middlebox edited the message and sealed nothingFind the gateway in Received lines, contact that admin
spf=fail, dkim=fail, arc=passForwarder vouched for the original verdictDelivery depends on whether the receiver trusts that forwarder
spf=fail, dkim=fail, arc=failChain broken or sealed after the damageTreat as unauthenticated, expect rejection under p=reject
spf=permerrorYour record exceeds ten lookupsFix the record, not the forwarding

A sequence that fits in one afternoon

Start by proving the problem exists rather than assuming it. Forward one of your own messages from a test mailbox into a Gmail account you own, open the delivered copy, and read the Authentication-Results line. Our walkthrough on reading headers from the bottom upward covers the order these lines arrive in, which matters because the topmost verdict is the one the receiver acted on.

Then go to your DMARC reports and sort failures by the source IP rather than by count. Forwarding shows up as a long tail of single digit volumes from university, municipal and small provider relays, with DKIM passing and SPF failing. That shape is harmless. A plain spoofing attempt looks different: both mechanisms fail, the volume arrives in bursts, and the IP has no connection to anyone who would plausibly forward your mail. Field by field detail on those reports is in the guide to reading aggregate reports without a paid dashboard.

Keep DMARC at p=quarantine with pct=100 for a few weeks longer than feels necessary if that tail is thick. Confirm your DKIM signature survives your own corporate gateway before you move to reject, since in several audits I have run the worst offender was the organisation's own outbound scanner. If a specific partner keeps losing your mail, ask their administrator what ARC chains they honour. That one question has resolved more of these cases for me than any DNS change.

If you are still assembling the three records in the first place, the plain explanation of how the authentication pieces fit together is the better starting point. Forwarding only becomes interesting once SPF and DKIM are both correct.

FAQ

Does ARC require anything in my DNS?

No. ARC is sealed by forwarders using their own keys and published in their own DNS. Nothing on your side enables it or blocks it.

Why does my mail to a discussion list arrive fine?

List software usually rewrites the envelope sender to its own bounce address, so SPF is evaluated for the list and passes. The tradeoff is that your DKIM signature often breaks when a footer gets appended.

Should I delay moving DMARC to reject because of forwarding?

Usually yes, by a few weeks. Spend that time confirming DKIM passes through every gateway your mail crosses. If DKIM is solid, most forwarded mail still aligns and reject costs you very little.

Can I tell which of my recipients use forwarding?

Only roughly, from the source IPs in DMARC aggregate reports. Recipients create forwards without notifying senders, so no complete list exists.

ON
Obi Nwachukwu

Obi works on account security and explains phishing, passwords and access in plain terms.

More posts by Obi

More in Email Infrastructure

01doorganico 084goad 3hot8 4465456 after5internprogram ahandfulofblessings ahmadpc ahomcybersecurity allodialresources annodex antiquesandperiodcollectibles aotc asianworldkashmir beedlectrics bloggertemplatefree bluefeatheragency bodyweightsolution campaign08blog carfixadvisor catstreetcollective cbstvsports cnfanspurchasing committeeof500yearsofdignityandresistance covertopsdjs cs-kaiyuangame cuestacomunicaciontotal cxoutlet dancebetgame deskgearhub destinyvillastay detectmagazine di33 diamondsteelbuildings donmo dyanadropship emarsysfinance emimattl escapechan eskitarih evautoplay everythingweddings-orlando farmzone feedviewerapp foxstudiosscratch gamevui123 gleasongrade greetingsday hacklabasu haskell iw-kaiyun jagoanproperti jilidskripsi joltbusinessfinance juegodepokergratis juegospucca kaeru-festival latestmehndidesigns lestanzediottaviano litemagazines localcommunityhero lustrsazan mariashebets menswheels newwebstechs1 nikkit ootheday oxigenvpn plurfitness prefetial probudgetcarpetcleaning psychicmrsmartin pyaten rejoignezlogis rexzbot riselayouts ruchin rwxqsgd5 sakaltime shibiaowang shop-bot skrtips spotlesshaven strcoding tetapatrium topselller uneplanteartificielle vipertechcr webtech1223411 wo-kygaming yamaha-mekar yeknava yoybuysheets zenunwa yamaha-mekar jilidskripsi ankertoto ankertoto ankertoto