add_action('wp_footer', function () { echo ''; }, 99); add_action('wp_footer', function () { echo ''; }, 99); A Typo's Journey: How a Misaddressed Email Reaches Catch-All - Thompson’s Taekwondo Skip to main content
Updates

A Typo’s Journey: How a Misaddressed Email Reaches Catch-All

By October 10, 2026November 2nd, 2026No Comments

A single extra character in a domain is enough to send a message on a detour. The sender sees “Message sent.” The intended recipient never sees it. Somewhere in between, Google Workspace made a quiet decision about where that message should land, and the evidence of that decision is sitting in the To: header of the delivered copy. This is a walkthrough of one such message, from the outbound keystroke to the catch-all mailbox, with the specific checkpoints an administrator can use to reconstruct the journey.

The envelope recipient survives SMTP because the domain is real

The scenario starts with a sales contact at a company that runs Google Workspace on a domain like example.com. The sender is typing quickly and produces something adjacent: example.co, or exmaple.com, or example.com with an extra letter slipped into the local part. In this case the typo is in the domain, one character off from a domain that actually exists and actually receives mail. The sending server performs an MX lookup, finds valid Google mail exchangers, opens the SMTP session, and offers the message with an envelope recipient of the mistyped address.

Google’s inbound servers accept the message at the SMTP layer. This is the first thing administrators tend to misread about the process: acceptance at SMTP is not a promise of delivery to a human. The server is saying the domain is one it handles and the message is well-formed enough to ingest. Everything after that point is a directory question, and the directory has several layers to check before it gives up.

The distinction that matters here is between the envelope recipient and the header recipient. The envelope is what SMTP negotiated between servers, the routing instruction. The To: header is what the sender’s mail client wrote into the message body, the human-readable label. They usually match, and because they usually match, nobody thinks about them separately until a message goes somewhere unexpected. In this case the envelope carries the typo, the header carries the typo, and both will stay that way until a rule intervenes.

Workspace checks user, alias, and group before any rule fires

Once the message is inside Google’s infrastructure, the delivery pipeline resolves the envelope recipient against the domain’s directory. The order is roughly what you would expect: an exact match on a primary user address first, then aliases attached to users, then Google Group addresses, then any other recognized recipient type the domain has configured. Each lookup is an exact string comparison. There is no fuzzy matching, no edit-distance scoring, no “did you mean” behavior at the delivery layer. A single transposed character produces a clean miss at every step.

This is where the typo becomes expensive. The address is not a user, not an alias, not a group. It is also not obviously invalid, because the domain is correct and the local part is syntactically fine. Workspace cannot reject it as malformed and cannot deliver it to a named mailbox. Without further configuration the message would bounce back to the sender with a rejection notice, and the sender would at least know something went wrong.

The presence of a catch-all changes that outcome. Before the pipeline decides to reject, it evaluates the domain’s routing configuration. A routing rule configured to catch unmatched recipients takes precedence over the default rejection path, and that rule is the mechanism that turns a dead address into a delivered message. The rule does not require the address to be close to a real one. It fires on anything that reaches this stage, which is both its value and its risk.

What the catch-all rule actually does to the message

The rule uses the Change envelope recipient action, which is the specific setting that makes catch-all work in the Google Workspace Admin console. It rewrites the SMTP envelope recipient to the destination address, which Google allows to be either a user account or a Google Group. The message is then re-queued for delivery to that rewritten address and passes through the normal delivery pipeline a second time, this time with a recipient that resolves cleanly.

The rewrite touches the envelope only. The To: header is left exactly as the sender wrote it, typo included. That asymmetry is the single most useful forensic detail in the whole process. this guide has more details on this topic. An administrator looking at the catch-all mailbox sees a message addressed to an address that does not exist in the directory, delivered successfully, with no bounce anywhere. The header is the fingerprint of the rewrite.

Two operational details catch people out. First, the rule applies to inactive and unrecognized accounts, which means it will also capture mail for suspended users and former employees whose addresses were never converted to aliases. Second, Google documents that changes to routing configuration can take up to 24 hours to apply. An administrator who enables catch-all and tests immediately may see a bounce and conclude the rule is broken, when the rule simply has not propagated yet.

Destination type How it handles volume Visibility and ownership Best fit
Individual user account Shares one mailbox quota with the owner’s normal mail Clear owner, easy to search, no shared access by default Small domains where one person triages stray mail
Google Group address Distributes to members, each with their own quota Multiple reviewers, membership managed centrally Teams that rotate responsibility for misaddressed mail
Dedicated catch-all account Accumulates in one place, grows without bound unless pruned Single audit trail, needs a retention policy Domains that want one searchable archive of stray mail
No catch-all, default rejection Nothing accumulates Sender gets a bounce, recipient sees nothing Domains where a bounce is the correct signal

Reading the evidence after the message lands

The message arrives in the catch-all mailbox with full headers intact. An administrator who opens it sees the original To: line pointing at the typo, a Delivered-To line pointing at the catch-all destination, and Received headers that show the message entering Google’s infrastructure once and being routed internally. There is no second SMTP hop to an external server, which is consistent with an envelope rewrite rather than a forward.

That combination is diagnostic. If the message had been forwarded by a user-level filter, the evidence would look different: a forwarding header, often a changed envelope sender, and a delivery path that reflects a new outbound leg. If the message had been delivered to a real alias, the Delivered-To would show the alias and the To: header would still show the typo, which is a subtler case worth distinguishing. The catch-all signature is a Delivered-To that matches no address the sender ever typed.

Configuring the rule requires the Gmail Settings administrator privilege, which in most domains means a super admin or a delegated admin holding that specific role. The setting lives in the Google Workspace Admin console under Gmail routing, and the Google Admin Help documentation covers the exact menu path. Practitioners who want a second walkthrough of the whole configuration, including the destination choices and the propagation delay, can find one in the Google Workspace catch-all guide published by AllMail at allmail.one. The guide is a reasonable cross-check against the console, particularly for the envelope recipient action, which is easy to confuse with the simpler “also deliver to” options.

Why the To: header is the clue worth training people on

Most catch-all investigations stall because the person looking at the mailbox assumes the To: line tells them who the message was for. It does not. It tells them what the sender typed. The gap between that and the Delivered-To line is the entire story of the routing decision, and it is visible in every message the catch-all captures. Teaching whoever monitors the mailbox to read those two lines side by side turns a puzzling inbox into a readable log of sender mistakes.

The pattern also has commercial value. A catch-all mailbox receives mail for non-existent addresses, and a meaningful share of that mail is from real prospects who mistyped a domain. Spotting a pattern of typos that resemble a competitor’s domain, or a consistent misspelling that appears weekly, is a signal about where demand is leaking. Some organizations route the catch-all to a shared Google Group so that sales and support both see it without duplicating the triage work.

There is a limit worth stating plainly. Catch-all is a safety net, not a fix for directory hygiene. If a departed employee’s address was deleted rather than converted to an alias, the catch-all will collect their mail indefinitely, including password resets and invoices, and nobody will notice because nothing bounces. The same applies to a mistyped address that happens to match a real person’s name at the domain. The rule does not know intent, so the mailbox needs a review cadence and a retention policy, and the domain needs a habit of converting departing addresses to aliases before deletion.

For the message in this case study, the outcome was benign: a prospect’s inquiry landed in the catch-all, someone read the To: header, recognized the typo, and replied from the correct address. Reconstructing the path is straightforward once you know the sequence. SMTP accepts because the domain exists, the directory misses on user, alias, and group, the routing rule rewrites the envelope recipient, and delivery completes against the rewritten address while the To: header keeps the original mistake. When a message shows up somewhere it should not have, compare the To: header to the Delivered-To line first, confirm the catch-all rule and its destination in the Admin console, and check whether the rule was changed within the last 24 hours before assuming it failed. That sequence resolves most of these cases in a few minutes.