Most mistakes need less than a second email
The worth of a correction depends on what the mistake would make a reader do. A typo makes them do nothing. A broken link makes them fail to reach a page, which can often be fixed at the page. A wrong date, price or code makes them act on false information, which may need telling. A message sent to the wrong people raises a different question altogether. Classifying the mistake before reaching for the send button is most of the work.
A second email is not free. It doubles the volume to the same people in a short time, it draws attention to a slip many readers never noticed, and every extra marketing email carries some risk of exits and complaints. The deliverability guide explains that complaints weigh heaviest in a sender's reputation, and Google's sender guidelines, listed in sources, ask senders to keep the spam rate reported to them below a documented threshold. An unnecessary correction to a large list can do more harm to deliverabilityWhether sent email reaches the inbox rather than spam or rejection, decided by sender reputation. than the mistake did to anyone. The steps below apply to every campaign type in the campaigns hub, from a newsletter issue to a promotion or an announcement. A press release sent to journalists is a different case, with embargoes, a newsroom page and writers who may already be at work, and the tutorial on correcting a press release after it goes out takes it from there.
| Kind of mistake | Usually needs | Why |
|---|---|---|
| Cosmetic: a typo, an awkward sentence, a slightly wrong image | Nothing, beyond fixing the template | Nobody acts differently because of it |
| A broken or wrong link | A fix at the destination: a redirect or a corrected page | Readers who click later reach the right place without another email |
| A personalisation failure: an empty or wrong first name | Usually nothing; a short note only if it was widespread and jarring | Embarrassing rather than harmful, unless it showed one person another's details |
| A wrong date, price, offer or code | A fix at the destination where possible, and often a short correction | Readers may act, or fail to act, on false information |
| Sent to the wrong segment | Depends on what they received; often nothing | People who received something irrelevant need no further email about it |
| Sent to people who had opted out | No correction to them; suppression, a record and a data protection review | A consent matter first, and another email would compound it |
Before you start
- Access to the sending platform with permission to pause sends and flows, and someone who knows where the documented way to pause automated flows is kept.
- The send's own recipient report, which lists who was sent the email and who it was delivered to, and the final version of the email as sent.
- The original brief and approval record, so the mistake can be compared with what was intended and approved.
- Someone who can change the destination: the website, the checkout, the booking system or the landing page, because the best fix is often there.
- The service team told, with a short note on what went wrong and what to say, before any correction goes out.
- A contact for data protection, in case the mistake reached people who had opted out or showed one person's details to another.
- One person in charge of the incidentA live error in marketing work, data or systems that reaches customers, spend or reports, such as a wrong price, a wrong segment or a broken purchase tag., with a single place where everything found is recorded.
The steps
Stop the send and any flow using the same content
Pause anything still sending: later waves of a staged send, a resend to people who did not open, a scheduled send in another time zone. Then look for every automated flow that uses the same template, block, code or link, because a mistake in a shared content block keeps going out long after the campaign has finished. The automation hub describes keeping a kill switch, a documented way to pause every flow at once, for exactly this kind of moment; if only one flow is affected, pause that one. Note the time the mistake was found and the time sending stopped, and keep the faulty version rather than overwriting it, because the later steps need it.
Find out exactly what went wrong
Open the email as it was sent, not the draft, and write down precisely what is wrong: which link, which price, which code, which field, which segment. Compare it with the brief and the approved version, because the difference tells you whether the error was in the brief, in the build or in a late change after approval. Check the destination too: a link that worked in testing may point at a page that was changed afterwards, and a code that worked last week may have expired. Look at the plain-text part and the preheader as well as the HTML, since the same mistake often appears in places nobody reads during a panic. Record the answer in a sentence anyone in the business could understand.
List who received it
Take the recipient list from the send's own report, not from the segment as it stands now, because segments change as people join, buy and unsubscribe. Note how many were sent the email and how many it was delivered to, and whether any waves were stopped before sending. Check the list against the suppression listA list of addresses or contacts excluded from every send, such as unsubscribes, hard bounces and spam complaints.: if anyone on it received the email, the mistake includes a consent problem, and those people are handled separately in the next step. This list, and only this list, is the audience for any correction. Do not use the people recorded as opening it; the email metrics and testing guide explains why opens are unreliable, and a correction sent only to openers misses readers whose mail client blocks images.
Decide whether the mistake needs a correction
Use the table above. Ask what a reader would do differently because of the mistake, and whether fixing the destination removes that difference. A cosmetic slip needs no correction. A broken link usually needs a fixed page, not an email. A personalisation failure usually needs nothing, unless it showed one person another person's details, which is a data protection matter to raise at once with whoever handles it; the AI mistake guide explains why they hear first. A wrong price or offer needs care: whether a price shown in error has to be honoured is a legal question that varies by market, so involve whoever owns commercial and legal decisions and take advice, and keep to the principle the planning promotions guide sets out, that the advertised price should be one people can actually pay. Mail sent to people who had opted out is a consent matter: re-apply the suppression, record what happened and who was affected, and never send them a correction. The suppression list entry explains why opted-out addresses stay on file so they can be recognised and blocked, the list building guide explains that an unsubscribe has to persist, and the ICO's direct marketing guidance, listed in sources, explains the UK rules on electronic marketing, including offering a way to opt out in every message. Nothing here is legal advice.
Fix what can be fixed without emailing
Most fixes belong at the destination. Redirect a broken link to the right page. Correct a wrong date or description on the landing page. Make a code work at checkout on the terms the email advertised, where the business decides to honour them. Add a short note on the landing page if readers arriving from the email need one. A destination fix reaches everyone who acts on the email from now on, including people who have not read it yet, without adding to anyone's inbox. Test the fix by following the email's own link from a seed addressAn inbox or test contact the business controls, added to a send or run through a workflow so the team sees exactly what real recipients would receive., not by visiting the page directly. If the fix is complete, the decision from the previous step may change: a mistake readers can no longer act on often needs no correction at all.
Write the correction
If a correction is still needed, write it short. The subject line says plainly what it is about, such as a corrected date or a code that now works, without the drama of a large apology or the trickery of a subject line designed to be opened. The body states what was wrong, what is right, and what the reader should do, in that order, in a few sentences. The apology is proportionate: one sentence, owning the mistake, without excuses. Over-apologising for a small slip, or adding a discount by reflex, teaches readers that mistakes pay and makes the next correction more expensive. A correction to a marketing email is still a marketing email, so it carries the usual footer and unsubscribe link; it does not become transactional emailAn email triggered by a person's own action or account, such as a receipt, password reset or shipping notice, rather than a marketing send. because it corrects an earlier one, and the promotional versus transactional comparison explains that purpose, not the label, decides which a message is. Run it through the QA checklist like any other send.
Send it only to the original recipients
Send the correction to the recipient list from the third step, minus anyone on the suppression list and anyone who has unsubscribed since the original went out. Not to the whole list, because people who never received the mistake would be told about it, and not to openers only, for the reasons above. Send it once. A correction that itself needs correcting is the point at which readers stop trusting the sender's email, so the second reader the newsletter design and copy guide recommends before every send matters more here than anywhere. If the original was a staged send that was stopped part-way, send the corrected version of the original to the remaining waves instead of the faulty one plus a correction.
Handle replies and complaints
Replies will come, to the original and to the correction, and some will come through the service team. The service recovery and complaints guide sets out the sequence: acknowledge quickly, apologise without excuses, understand what the person needs, fix the original problem, make proportionate amends where the mistake cost them something, follow up and record the cause. A customer who missed a deadline because the email gave the wrong date has a real loss; a customer who saw a typo does not. Watch unsubscribesShare of delivered emails that led the recipient to opt out. and spam complaintsShare of delivered emails that recipients marked as spam. on both sends against the usual level for that kind of send; the unsubscribe rate spike diagnosis helps locate the exits if they rise. Add everyone who unsubscribes or complains to the suppression list at once.
Find the cause
Once the incident is contained, trace why it happened. The five whys suits a single, traceable failure like this one: state what went wrong, ask why, and keep asking why of each answer, on evidence, until the chain reaches a process or rule the team can change rather than a person to blame. "The code was wrong" becomes "the code expired before the send date" becomes "codes are created with a default expiry and nobody checks it against the send date" becomes "the QA checklist checks that links work but not that codes work". Compare the final email with the approved version from the approval record: if the error arrived after approval, the cause is a change that bypassed review, not the reviewer. When the mistake is that people received the same email twice, the diagnosis for subscribers receiving duplicate emails separates a repeated send from duplicate records, two sending systems and flows that collide.
Change the check that missed it
Change the check, not only the template. Add the missing line to the team's copy of the email campaign QA checklist, such as testing every code at checkout against the send date, or reading the plain-text part, or reconciling the segment count after suppressions. If the mistake was a change made after sign-off, change the workflow so that a late change sends the email back for review, as the email campaign send SOP does by asking the approver to sign off the final test; the approvals and compliance reviews guide explains that incidents that got through point to a missing row in the approval matrix, and that prices and promotions belong with a reviewer who checks them. Record the incident, the decision on correction and the change made, so the next person facing the same mistake starts from the answer. Where the same gap could let a mistake through in other channels too, the tutorial on reviewing a marketing error after the fix runs a fuller review and keeps one error log for every channel. The marketing incident management playbook sets who leads, who is told and when that log is read across every kind of incident, of which a wrong email is one.
A florist's wrong discount code (illustrative)
Stemquiver Florist's incident log
Stemquiver Florist, an invented online florist, sends its Mother's Day email in two waves. The first wave goes to past bouquet buyers with a discount code that, it turns out, expired the previous week. The incident log:
| Time | What happened |
|---|---|
| 09:00 | First wave sent to 4,200 past bouquet buyers; second wave of 3,100 scheduled for 13:00 |
| 09:40 | Two customers reply that the code is rejected at checkout |
| 09:45 | Second wave paused; the cart reminder flow, which uses the same offer block, paused too |
| 10:10 | Send report: 4,200 sent, 4,150 delivered, 0 recipients on the suppression list; checkout log shows 38 failed code attempts |
| 10:30 | Decision: reactivate the code at checkout on the advertised terms until the stated end date, and send one correction, because readers who tried and failed are unlikely to try again |
| 11:20 | Correction sent to the 4,150 delivered recipients: subject "The code in this morning's email now works", three sentences and a one-line apology |
| 13:00 | Second wave sent with the working code; cart flow resumed |
| Next day | 126 orders used the code from the first wave; 24 replies, all answered by the service team the same day; 9 unsubscribes on the original, 5 on the correction, 1 spam complaint |
The five whys traced the cause to codes being created with a default one-month expiry that nobody checked against the send date. The florist added "test every code at checkout on the send date" to its QA checklist and changed the default expiry to the promotion's own end date.
Verify it worked
Check each step against the incident record and the platform.
- Confirm nothing is still sending. No later wave, resend or flow using the same template, block, code or link should be live.
- Read the incident record. It should hold the time found, the time sending stopped and the faulty version; the log in the illustrative example above shows the first customer reply and the pause.
- Compare the recipient list with the send's own report, then check it against the suppression list.
- Follow the email's own link from a seed address. It should reach the corrected destination.
- If a correction went out, confirm it went once, to the original recipients only. Compare unsubscribes and complaints on both sends with the usual level.
If it goes wrong
- Faulty content keeps going out after the pause. A shared block feeds a flow you missed. Search every flow for the block, code or link, and pause each one.
- The fixed link or code still fails. You tested the page, not the email. Retest through the email's own link from a seed address.
- Suppressed or opted-out people received the email. Treat it as a consent matter. Re-apply the suppression, record who was affected, tell whoever handles data protection and send them no correction.
- Unsubscribes or complaints rise after the correction. Check the audience was the original recipients only. Add everyone who exits to the suppression list at once.
Next steps
- Read the guide to handling an AI mistake when the error came from AI-produced work.
- Read the newsletter design and copy guide to build a second reader into every send.
- Read the promotional versus transactional comparison to judge what kind of message any follow-up is before you send it.
Common mistakes
- Apologising to the whole list. People who never received the mistake are told about it, and the volume brings exits and complaints.
- Correcting only the openers. Opens are unreliable; the recipient report is the audience.
- Emailing people who had opted out to apologise. The original was the problem; a second email compounds it.
- Sending a correction before fixing the destination. A redirect or a working code often removes the need for an email.
- Leaving the flow running. The campaign stopped, but the shared block keeps going out through an automation.
- Over-apologising. A long apology and a reflex discount for a typo make the slip bigger than it was.
- Promising to honour a wrong price without advice. Whether an error has to be honoured varies by market; the decision belongs with whoever owns commercial and legal questions.
- Blaming the person who pressed send. The cause is the check that missed it, and only changing the check prevents the next one.
