# Brand Social Media Mistake Response: The First Hour

Your company account posted something wrong or offensive. A brand social media mistake response: first-hour steps, who signs off, what search and AI show.

---


When a company account posts something wrong, tone-deaf or offensive, the brand social media mistake response comes down to four decisions in the first hour: stop everything scheduled, capture the post as evidence, decide whether to edit, delete or keep it, and name one person who signs off on the public line. That public line is a correction, a clarification or an apology, and which one you use depends on what actually went wrong, not on how loud the reaction is.

This guide is about the company account specifically. A personal post you regret or an ambassador's scandal has its own handling. Here the account belongs to a business, several people can publish from it, and the mistake is the company's, whoever typed it. That changes the questions: who decides, who speaks, and what search engines and AI assistants will say a month later.

The scenario below is invented to show a decision, and it is labeled as such. No real company incident is used, and no number in it is a statistic.

<div class="box"><b>The short version</b>
<ul><li>In the first hour, freeze every scheduled post and automated reply, capture the post and its audit trail, and only then decide what happens to the post.</li><li>Match the response to the fault: correct a wrong fact, clarify a misread post, apologize for harm. Never apologize for a typo as if it were an injury, and never "clarify" an offensive post.</li><li>One incident owner and one approver, named in advance. The person who wrote the post is never the only approver of the response.</li><li>Deleting a post removes it from the account and from search on that platform, but screenshots, quote posts and coverage remain, so a deletion needs a statement beside it.</li><li>Afterward, search engines and AI assistants build the story from indexed pages. Publish one dated, factual statement on a page you own, and check the answers after a week and after a month.</li></ul></div>

Three figures from platform and search pages set the time frame for the decisions that follow.

<figure class="keyfacts">
<div class="keyfacts-grid">
<div><b>1 hour</b><span>how long X Premium subscribers can edit a post after sending it</span></div>
<div><b>5 edits</b><span>the most X allows in that hour, each with a public "edited" label</span></div>
<div><b>About 6 months</b><span>how long a Removals tool request lasts in Google Search, for pages on your own site</span></div>
</div>
<p class="fig-src">Figures from X's help page on editing posts and from Google Search Central's page on removing information. Policies change, so check both pages before relying on them.</p>
</figure>

## The brand social media mistake response in the first hour

Spend the first hour on containment and record, not on wording. A statement written in panic, while a scheduled promotion is still going out, makes the situation worse.

Marketing keeps running on its own schedule unless someone stops it, so the first act is a pause. Who pauses, how, and what happens next is where teams lose minutes. This is the order we use.

<figure class="fig">
<figcaption><b>The first-hour procedure for a company account</b></figcaption>
<ol class="fig-steps">
<li><b>Freeze the queue</b>Pause every scheduled post, automated reply and cross-post on every channel. Each scheduler keeps its own queue.</li>
<li><b>Capture the post</b>Record the post, its timestamp, its engagement and the scheduling audit trail before anyone touches it. The next section lists what to save.</li>
<li><b>Name the incident owner and the approver</b>The owner runs the hour. The approver signs off on anything public. They are two different people.</li>
<li><b>Classify the mistake</b>Factual error, offensive or tone-deaf content, wrong-account post, or unauthorized post. The decision table below turns each type into a first move.</li>
<li><b>Decide the post's fate</b>Edit, delete or keep with a correction beside it. Decide after capture, never before.</li>
<li><b>Publish a first line or hold deliberately</b>If the facts are clear, post the response. If they are not, post a one-sentence holding line that says you know and are looking into it, and say when you will update.</li>
</ol>
</figure>

A short, accurate line at minute 50 beats a polished essay at hour six, if it promises nothing you cannot confirm.

## What to capture before you touch the post

Capture the evidence before you delete, edit or reply, because afterward you can no longer show exactly what the post said, who approved it or when it went out. The record serves the internal review, your lawyer or insurer if needed, and a platform dispute if the post was unauthorized.

<aside class="callout">
<span class="callout-label">Rule</span>
<h4>Nothing is deleted before it is captured</h4>
<p>Capture first, delete second, and write down who did both and when. A post that disappears with no record leaves you unable to prove what it said, and unable to prove what you changed.</p>
</aside>

Save these items in a read-only folder that only the incident owner and the approver can reach:

- A full-page screenshot of the post showing the account name, the timestamp and the visible engagement figures.
- The direct link to the post, copied as text.
- The text of the post, any image text and any alt text, copied out so it is searchable later.
- The reply and quote counts, and screenshots of replies that raise a threat, a safety issue or a legal claim.
- The audit trail from the scheduling or publishing tool: who drafted the post, who approved it, when it was scheduled, when it went out.
- The list of who held publishing rights on the account that day, including agencies and connected apps.

Do not post these captures publicly.

## Delete, edit or keep the post

Delete a post that is harmful by itself, such as offensive content or a claim you cannot stand behind. Keep and correct a post whose only problem is a mild factual slip that is already widely seen. Edit only where the platform lets you, and expect the edit to be visible.

The platform's own rules shape the choice. Each platform's edit and delete options differ, so check them before you decide. On X, only Premium subscribers can edit, within a one-hour window, and an edited post carries a visible label.

### What an edit shows

According to X's help page on editing posts, only X Premium subscribers can edit, the edit window is one hour, and up to five edits are allowed in that hour. An edited post carries an "edited" label that everyone can see, and the earlier versions can be viewed. Promoted posts, polls, pinned posts and replies cannot be edited. An edit is therefore not a silent fix: it tells everyone that something changed and lets them see what.

### What a deletion does and does not do

X's help page on deleting posts says a deleted post is removed from your account, follower timelines and X search, and that reposts of it are removed too. It also says that posts copying your text, and reposts with a comment of their own, are not removed, and that X cannot remove posts that are not on its own apps or site, so cached or cross-posted copies may persist. Screenshots are not addressed at all.

So on X a deletion changes what a new visitor sees. It does not change what people who already saw the post have saved, quoted or written about.

<aside class="callout">
<span class="callout-label">Warning</span>
<h4>A deleted post with no statement reads as concealment</h4>
<p>People who saved the post will share it with the caption "they deleted it". Delete when the content is harmful, but publish the correction or apology in the same hour, so the deletion arrives with an explanation.</p>
</aside>

### When keeping the post is the better call

Keep the post, with a correction beside it, when the error is a minor fact (a wrong date, a wrong link) or when the post has been quoted so widely that deletion would only create a second story. Reply to the original with the corrected fact. Never keep a post that is discriminatory, that mocks a tragedy or that makes a claim you cannot support, however many people have seen it.

On X, a factual error can attract a Community Note whether or not you respond. X's Community Notes help page says a note becomes public only after contributors with differing viewpoints rate it as helpful. A fast, plain correction from you gives you the first word.

## Who signs off on the response, and how fast

Decide the sign-off before the mistake, and tie it to severity. The person who wrote or scheduled the post should draft the facts but should not be the only approver of the public response, because they are too close to it and often the first to want to delete everything or to defend it.

<div class="tablewrap"><table><thead><tr><th>Severity</th><th>Typical case</th><th>Who signs off</th><th>Target</th></tr></thead><tbody>
<tr><td><strong>Low</strong></td><td>Typo, wrong link, wrong image</td><td>Social lead</td><td>Fix and note within the hour</td></tr>
<tr><td><strong>Medium</strong></td><td>Wrong price, date or product fact that affects customers</td><td>Social lead and head of comms</td><td>Correction within the hour</td></tr>
<tr><td><strong>High</strong></td><td>Offensive, discriminatory or tone-deaf content, or a post tied to a tragedy</td><td>Head of comms and chief executive</td><td>Response within the hour, full statement the same day</td></tr>
<tr><td><strong>Critical</strong></td><td>Unauthorized post, regulated claim, a named person harmed, or leadership implicated</td><td>Chief executive and counsel</td><td>Holding line within the hour, counsel before detail</td></tr>
</tbody></table></div>

The targets are our practice, not a rule of any platform. Two things make the table work. The approver can be reached out of hours, because a rule that needs a sleeping person gets skipped. And the social lead may pause the queue and capture evidence without asking anyone.

Where the claim is regulated, such as a health claim, a financial promise or a price you are legally bound by, involve your own counsel before posting detail. This guide is information, not legal advice.

## Correct, clarify or apologize: choosing the response

Use the response that matches the fault. A correction fixes a fact. A clarification supplies context for a post that was accurate but misread. An apology answers harm or disrespect. Mixing them up is the most common way a manageable mistake becomes a story.

**Correct** when something factual was wrong. State the right fact in the first sentence, say the earlier post was wrong, and stop. A typo does not need an apology in the language of an injury.

**Clarify** when the post was accurate but the reading was unfair, or context was missing. Say once, in plain words, what you meant and what you did not mean. If many readers took it the same way, the post failed, and that calls for an apology instead.

**Apologize** when the post was offensive, careless about a real event, or insulting to a group. Say what was wrong, who it affected, what you have done and what will change. An apology has four parts, in this order: the specific fault, the people harmed, the action taken and the change that follows. Leave out "if anyone was offended", leave out the excuse, and leave out any blame on an agency, an intern or a tool. The company published it.

An invented example of a good apology line, after a scheduled promotion ran during a local emergency:

> We scheduled a promotional post that went out while people in our region were dealing with flooding. That was wrong, and we are sorry. We have removed it and paused all scheduled posts. We are changing how we review the queue when there is breaking news.

An invented example of a good correction line, after a wrong end date on a sale:

> Correction: our sale ends Sunday, not Saturday. The earlier post had the wrong date. Anyone who has already ordered keeps the sale price.

Both are short and specific. Humor can read as mockery of the people who were upset, so do not use it where anyone was harmed.

### Where to publish the response

Publish on the channel where the mistake happened, as a new post, and pin it for a short period. If customers were directly affected, add an email or an in-app message, because a pinned post only reaches people who return to the account. Reply to the original post only if you are keeping it.

## Decision table by type of mistake

The four types of company-account mistake differ in their first move, their response form and what lingers afterward. The table is the one-page version.

<div class="tablewrap"><table><thead><tr><th>Type</th><th>First move</th><th>Response form</th><th>What lingers</th></tr></thead><tbody>
<tr><td><strong>Factual error</strong></td><td>Capture, then correct beside the post or edit where allowed</td><td>Correction: right fact first, no apology theater</td><td>Low if fixed fast; wrong figure may be repeated by others</td></tr>
<tr><td><strong>Offensive or tone-deaf content</strong></td><td>Freeze the queue, capture, delete, post the response in the same hour</td><td>Apology: fault, people harmed, action, change</td><td>High: screenshots, quote posts and coverage that name the brand</td></tr>
<tr><td><strong>Wrong-account post</strong></td><td>Capture, delete, find how access was shared</td><td>Short apology or correction, depending on content</td><td>Medium: the "who has the keys" question follows the brand</td></tr>
<tr><td><strong>Hacked or unauthorized post</strong></td><td>Secure access, capture, then delete the unauthorized posts</td><td>One line: not us, access restored, posts removed</td><td>Medium to high if the claim is doubted; proof decides it</td></tr>
</tbody></table></div>

## Wrong-account posts, agency posts and unauthorized posts

Three variants turn on access, not on words.

### The wrong-account post

A staff member posts a personal opinion from the company account, or a message meant for one brand goes out from another. To the public it is the company's post. Handle it according to its content, then fix the cause: separate logins and a dedicated browser profile for the company account.

### The agency, freelancer or intern post

When an outside agency or a junior team member posts the mistake, the brand answers publicly and the matter is settled privately. Never name the agency or the person in the public response. Inside, review the approval chain: a post with no second reader is a process fault, and the fix is a second reader, not a scapegoat.

### The unauthorized or hacked post

Claim a hack only when you can show it. Login history from a device you do not recognize, changed recovery details or an app with posting rights you never granted are the kind of trace that supports it. X's help page on compromised accounts tells owners to change the password, secure the linked email address, revoke unfamiliar apps and delete the posts made by the unauthorized party. It also notes that odd activity can come from a buggy third-party app rather than a break-in. If the login history shows only your own people and devices, treat the post as an internal mistake, because a hack story that unravels becomes the larger story.

## What search engines and AI assistants show afterward

Search engines and AI assistants do not replay the platform's timeline. They assemble an answer about the brand from pages they have indexed, so after a mistake the story is told by whatever is on those pages: coverage, commentary, forum threads, and your own statement if you published one on a page that can be indexed.

This matters because a deleted post is not a deleted story. Third-party pages that quote or screenshot the post are outside the platform's reach, as X's deletion page itself says of copies beyond its own apps and site. A search for the brand name plus the incident can keep surfacing those pages after the original is gone, without your side of the account.

### How AI assistants build the answer

AI assistants that search the web, such as ChatGPT search (per OpenAI's help page), find sources by searching the web from the prompt and cite them in the answer. Google's documentation on AI features in Search says a page must be indexed and eligible to be shown with a snippet to be a supporting link, and that the systems may issue multiple related searches across subtopics to build a response. So a question such as "what happened with this brand's post" can pull in more than one page: the commentary, the coverage and any statement of yours that is indexed. We have not tested how any assistant weights those sources, so treat this as how the systems are described, not a prediction.

### What to do about it

1. **Publish one dated statement on a page you own.** State what happened, what was wrong, what you did and what changes. Plain text, no spin. It gives search and AI systems a first-party source to find.
2. **Leave it up.** Do not remove the statement when the attention fades. It is the page that answers the question the next time someone searches the incident.
3. **Check the answers after a week and after a month.** Search the brand name with the incident words, and ask the main AI assistants what happened. Note any wrong detail and which page it comes from. Fix the source page if it is yours; report it to the provider or the publisher if it is not. The [AI search reputation service](/ai-search-reputation) covers how those answers are audited.
4. **Use removal tools only where they apply.** Google's page on removing information says the Removals tool can take a page on your own site out of Google results in about a day, that such requests last about six months, and that a permanent removal needs the content removed or updated, protected by a password or marked noindex. It covers pages you control, not third-party coverage, and it should never be used on your own honest statement.
5. **Keep watching.** A mistake can resurface when someone reposts it, and [reputation monitoring](/reputation-monitoring) catches that early.

## A worked scenario: the scheduled post that ran during an emergency

This scenario is invented. Harbor and Pine is an imaginary regional furniture retailer with a small marketing team and an outside agency that manages its scheduling tool. A promotional post, scheduled a week earlier, goes out during a local flooding emergency with the caption "Our biggest sale ends tonight". Replies begin within minutes, and everyone is reachable. Times are illustrative, not measured.

<div class="tablewrap"><table><thead><tr><th>When</th><th>Action</th><th>Owner</th><th>Result</th></tr></thead><tbody>
<tr><td><strong>Minutes 0 to 10</strong></td><td>First replies read. Queue paused on every channel and in the agency tool.</td><td>Social lead</td><td>Next two scheduled posts stopped</td></tr>
<tr><td><strong>Minutes 10 to 25</strong></td><td>Screenshots, replies and audit trail saved. Post deleted.</td><td>Social lead, head of comms</td><td>Record exists before removal</td></tr>
<tr><td><strong>Minutes 25 to 45</strong></td><td>Apology drafted by the social lead, approved by head of comms and chief executive. No excuse, no blame on the agency.</td><td>Head of comms, chief executive</td><td>Approved text</td></tr>
<tr><td><strong>Minutes 45 to 60</strong></td><td>Apology posted as a new post and pinned. Automated replies stay off.</td><td>Social lead</td><td>Public line out within the hour</td></tr>
<tr><td><strong>Hours 2 to 24</strong></td><td>One named person answers real questions. Statement page published on the company site. Staff told what to say.</td><td>Head of comms</td><td>One source of truth</td></tr>
<tr><td><strong>Days 2 to 7</strong></td><td>Review why the post was not paused. New rule: no scheduled posts during a regional emergency.</td><td>Head of comms, agency</td><td>Rule in writing</td></tr>
<tr><td><strong>Weeks 2 to 4</strong></td><td>Search the brand name and incident words, ask AI assistants, check the answers against the statement page.</td><td>Head of comms</td><td>Wrong details traced to source</td></tr>
</tbody></table></div>

## Put a pause rule and a sign-off line in writing this week

The next step is a one-page protocol, written before the next mistake and tested once with the people who would use it. It holds these points:

- Who may pause the queue without asking, and how, on every account and every tool.
- The severity table above, with real names and out-of-hours contacts.
- Who holds publishing rights on every account, including agencies and connected apps, with unknown apps revoked.
- A rule that nothing scheduled goes out during breaking news that touches your region, sector or customers.
- A two-reader rule for any post that mentions a tragedy, a group of people, a price or a health or financial claim.
- A skeleton for correction, clarification and apology, so the first draft is not blank.

Reputation Partners works through exactly these questions with brands before and after an incident, including how the story reads in search results and AI answers a month later.

<div class="box hot"><b>If this is happening to you now</b>
<p>If a post from your company account is in front of people and you are not sure whether to delete it, correct it or apologize, start with the capture and the queue freeze, then talk to us about the response. Our <a href="/crisis-management">crisis management desk</a> can take the first decisions with you.</p></div>

## Questions people ask

### How long can a company wait before responding to a social media mistake?

Respond as soon as you know the facts, and sooner with a holding line if you do not. In our practice the first hour is the target for a high-severity mistake. Silence while the screenshots spread leaves others to tell the story. If the facts are unclear, say you are looking into it and name when you will update.

### Should we reply to every angry comment?

No. Replying to each one with the same script looks automated and feeds the thread. Assign one named person to answer real questions in plain words, ignore pure abuse, and report threats. Turn off automated replies on the affected post, because a cheerful auto-reply under an apology reads as the brand not listening.

### Can we say it was the agency or the intern?

Do not say it in public. The company published the post, and the public holds the company responsible. Naming a person or a supplier looks like deflection and can expose a junior employee to harassment. Review the contract and the approval chain privately, and tell the public what changes in the company's process.

### What if the post was a joke that landed badly?

Treat it by its effect, not its intent. If it offended people, apologize for the fault, not for how they felt. Do not explain the joke or tell readers they missed the point. A short, sober apology usually avoids the second wave of mockery that a defensive explanation brings.

### When can we go back to normal posting?

Resume ordinary posts after the response is published and the main wave of replies has slowed, usually within a day or two for a low-severity mistake and longer for a serious one in our experience. Start with plain, useful posts, not jokes or promotions, and keep the pause on any topic connected to the incident.
