# Online Reputation Crisis Management: A Working Guide

Online reputation crisis management in order: sort what is true, run three tracks, pick a response by crisis type, and know what removal can and cannot do.

---


Online reputation crisis management is the work of deciding what is true, what to say about it, what to correct, and what people will find when they search your name while the story is still moving. It runs on three tracks at once, a public statement, a record of facts and corrections, and the state of search results and AI answers, and the order in which you start them matters more than any single tactic.

Much crisis advice is written for companies with a communications department and treats the crisis as a messaging problem. For a person, a founder, an athlete or a family, it is also a search problem and a legal problem, and a mistake in one track damages the other two. A statement that is clever but unchecked becomes evidence. A takedown request filed in anger becomes a second story. A search plan with no statement behind it leaves the first page to whoever is loudest.

This guide is written from the desk of a former newsroom editor, so it takes the journalist's side of the table seriously: what a reporter needs, what makes a correction credible, and why a true story about you cannot and should not be made to disappear. It is general information, not legal advice, and where a rule depends on the country or the platform, we say so and name the source we read.

<div class="box"><b>The short version</b>
<ul><li>Sort the situation before you act: false, true but incomplete, or true and damaging. The three call for different responses, and mixing them up is the most common cause of a crisis getting longer.</li><li>Run three tracks together: say (statement and spokespeople), fix (facts, corrections, takedowns, legal steps) and find (search results and AI answers). Start with a short holding position, not a full defense.</li><li>Removal is a narrow tool. Search engines remove results under specific policies and legal rules, they do not delete pages, and accurate reporting of public interest is usually kept.</li><li>Do not feed the story: no deleted posts without a record, no fake accounts, no pressure on reporters or complainants, no reviews written by friends.</li><li>Plan in weeks, not hours. Most crises have a peak of days, a long tail of weeks, and a recovery that is measured in months of accurate material accumulating.</li></ul></div>

## Online reputation crisis management starts with sorting what is true

The first decision is not what to say but what kind of crisis you have. Everything later, from tone to takedown routes, depends on it, and people get it wrong because the sorting is emotionally hard. When you are the subject, almost everything feels false.

Use three buckets and be honest about each claim separately, because a single story often contains all three.

**False.** The claim is factually wrong: wrong person, invented event, doctored screenshot, fabricated quote. This is the only bucket where correction, takedown and legal routes have real force. Even here, the first move is to gather proof, not to attack.

**True but incomplete or out of context.** The facts exist, but the framing misleads: a real incident from years ago presented as current, a quote without the sentence before it, one side of a dispute. The route is context and, where a publisher is willing, a correction or update. You are adding information, not removing it.

**True and damaging.** The facts are accurate and the public has a legitimate interest. A court record, a regulatory finding, a credible report on your conduct. The route is accountability: acknowledge, explain what changed, and let accurate new material build over time. We do not help hide accurate, lawful information about wrongdoing, and any provider who offers to is selling you a second crisis.

Write a one-page fact sheet before you talk to anyone outside your closest circle. List each claim in the coverage or posts as a separate line, mark it false, incomplete or true, and next to each write the evidence you hold and the evidence you lack. That page becomes the base of your statement, your lawyer's brief and your correction requests. It also protects you from the most dangerous error in a crisis: defending a claim publicly that you later have to concede.

### The questions that settle the sorting

Three questions do most of the work. Is there a document, recording or record that shows the claim to be true? Would a neutral person who read the original source reach the same conclusion as the headline? Is anyone identifiable harmed, and would you want the same treatment applied to someone else? If you cannot answer the first one with confidence, say less, not more, until you can.

## The first hour: five decisions you cannot easily undo

The opening hours of a reputation crisis are a sequence of small choices that feel urgent and are hard to reverse. Slow down on exactly these five.

<figure class="fig">
<figcaption><b>Five decisions to slow down in the first hour</b></figcaption>
<ol class="fig-steps">
<li><b>Do not delete</b>Deleting your own posts, comments or accounts feels like tidying up and reads as concealment. Screenshots exist by then, and the deletion becomes its own headline. Archive and restrict instead, and keep a private copy of everything you remove for your lawyer. If a post is genuinely unlawful or harmful to others, take it down and say plainly that you did.</li>
<li><b>Do not reply in the thread</b>A reply under the original post gives it more reach and places your first words in the worst possible setting. Reply, if you reply, from a prepared statement placed where people look for it.</li>
<li><b>Do not contact the reporter or the poster to argue</b>A respectful, factual note to a journalist with documents is normal. A phone call that sounds like pressure is not, and it is the kind of message that ends up quoted. The same goes for people who made complaints about you.</li>
<li><b>Do not open new accounts, ask friends to defend you, or request reviews</b>Coordinated defenders can be discovered, and the discovery becomes the story. This is a hard line for us, not a risk trade-off.</li>
<li><b>Do not sign anything without counsel</b>If the matter involves a possible legal claim, an employer, a regulator or a contract. A rushed apology or settlement can create liability that the original facts never did.</li>
</ol>
</figure>

What you should do in the first hour is dull: stop posting, preserve evidence, name one decision maker, tell the few people who must hear it from you, and start the fact sheet. If intimate images or private data are involved, report the content under the platform's rules the same day, before anything else.

### Preserving evidence the right way

Capture each page with its full address and a visible date, save the original file where you can, and note the time you captured it. Keep a simple log of who said what, when. Do not edit the files. Unaltered, dated captures are far easier for a lawyer, a platform or an editor to rely on than cropped images forwarded in a chat.

## The three tracks: say, fix, find

A crisis is managed on three tracks that run in parallel, each with a different owner and a different clock. Treating them as one stream is why people end up with a good statement and a terrible first page of search results, or a perfect legal letter and no public position.

| Track | Question it answers | Typical owner | Clock |
| --- | --- | --- | --- |
| <strong>Say</strong> | What do we state publicly, and who speaks? | Principal with a communications lead | Hours to days |
| <strong>Fix</strong> | What is wrong, and which route corrects it? | Counsel, publisher contacts, platform reports | Days to months |
| <strong>Find</strong> | What do people and AI assistants see when they look? | Search and content team | Weeks to months |

**Say** covers the holding statement, the full statement if one is warranted, answers to the three questions you will be asked, and silence where silence is the right answer. **Fix** covers verifying facts, requesting corrections or updates from publishers, reporting content that breaks platform rules, sending legal letters, and filing search engine removal requests where a policy applies. **Find** covers monitoring what appears, publishing accurate owned material, and checking how search engines and AI assistants summarize the matter.

The tracks constrain each other. What you say must match what your lawyer has cleared. What you ask a publisher to correct must match what you said. What you publish for search must match both. One person should hold the master version of the facts and approve every outgoing sentence, even in a family or a two-person team.

## Choosing a response by type of crisis: a decision table

Crises differ in speed, source and what can legitimately be done. This table sorts the common types by the response that usually fits first. It is a starting point, not a script; real cases mix types.

<div class="tablewrap"><table><thead><tr><th>Type</th><th>First move</th><th>What usually works</th><th>What usually fails</th></tr></thead><tbody>
<tr><td><strong>Viral post, claim is false</strong></td><td>Evidence file, short factual correction</td><td>One clear correction with proof, platform report for rule breaks</td><td>Arguing in replies, deleting your own history</td></tr>
<tr><td><strong>News article, facts wrong</strong></td><td>Written correction request to the editor with documents</td><td>Specific errors, sources attached, calm tone</td><td>Threats, vague "it is unfair" complaints</td></tr>
<tr><td><strong>News article, facts right</strong></td><td>Counsel review, short acknowledgement if warranted</td><td>Accountability, later accurate material</td><td>Removal pressure, which tends to produce a second story</td></tr>
<tr><td><strong>Leaked private material</strong></td><td>Safety and platform reports the same day</td><td>Rapid reports under privacy and intimate imagery rules</td><td>Public commentary on the content itself</td></tr>
<tr><td><strong>Accusation, unresolved</strong></td><td>Legal advice first, minimal holding line</td><td>Respect for the process, no attacks on accusers</td><td>Detailed rebuttals before the facts are established</td></tr>
<tr><td><strong>Impersonation or scam in your name</strong></td><td>Platform reports, public notice on real channels</td><td>One verified page stating what is official</td><td>Chasing each copy by hand without a central notice</td></tr>
</tbody></table></div>

Notice what is absent: no row says "make it go away". Where content is false, unlawful or breaks a platform rule, there are routes. Where it is accurate and of public interest, the realistic goal is context, completeness and time.

## The holding statement: what to say while you do not know enough

A holding statement is a short public position that buys time without lying. It is the single most useful document in the first day, and it should be written before you need the full one. It has four parts.

First, acknowledgement that you are aware of the matter, in plain words, without adjectives. Second, one sentence on what you are doing, such as reviewing the claims or speaking with advisers. Third, a commitment about what comes next and when, only if you can keep it. Fourth, a place where people can find updates, which is usually one page you own.

What it leaves out matters as much: no accusation of the person who spoke, no adjectives about the media, no denial broader than your evidence, no apology for things you have not yet established, and no promise to "get to the bottom of it" unless you will publish what you find.

### When silence is the right answer

Not every crisis needs a statement. A claim seen by few people that is easily disproved may be better handled privately with the publisher or platform, because a public reply spreads it. The test is audience and trajectory: is it moving beyond the community where it began, are journalists asking, and is it appearing in search results for your name? If the answer to all three is no, prepare the statement, hold it, and monitor. Silence is a decision with a trigger, not a lack of one.

### Apologies

If you are in the true-and-damaging bucket, an apology may be right, but it has to be specific, short, free of "if anyone was offended", and cleared with counsel where liability is possible. An apology tells people what you did, who it affected, what you have changed, and what you will not do. It does not ask for forgiveness and does not explain at length. A defensive apology is usually worse than none.

### Who speaks

Choose one voice, the principal or a named representative with access to the facts. Everyone else is briefed to point to the owned page. Many crises are lengthened by well-meaning relatives, colleagues and agents each saying a slightly different thing.

## Fixing the record: corrections, platform reports, legal routes and search requests

The fix track is where most of the technical work sits, and where false hope is most common. Four routes exist, and each has limits that you should know before you start.

### Corrections and updates from publishers

A correction request to an editor works when it is specific, documented and respectful of the publisher's job. Quote the sentence, state what is wrong, attach the evidence, and ask for what you actually want: a correction, an update, a clarification, or, for old coverage that reflects an outdated situation, a note on the current status. Editors have standards for correcting errors and little patience for requests to remove accurate reporting. An aggressive note is more likely to produce a follow-up story than a correction.

### Platform reports

Platforms act on their own rules, not on whether a post is fair. A report that cites the exact policy, such as impersonation, private information or non-consensual intimate imagery, and attaches the link and proof, has a chance. A report that says "this is false and hurts me" without a rule usually does not. Platform policies change, so read the live page for the platform concerned before filing.

### Intimate images and the Take It Down Act

If the crisis involves intimate images, including AI-generated or altered ones, a US federal law adds a route. The FTC says Section 3 of the Take It Down Act has been enforced since May 19, 2026, and requires covered platforms, which include social media, messaging, image and video sharing and gaming services, to offer a clear and conspicuous removal process. After a valid request, the platform must remove the content and known identical copies within 48 hours and make reasonable efforts to find those copies, so you do not have to report each duplicate. The FTC enforces the rule and takes reports of platforms that fail to act at TakeItDown.ftc.gov. Which sites count as covered and what a valid request must contain are set by the statute, which could not be opened for this guide, so confirm both with counsel. Source: [FTC, Complying with the Take It Down Act](https://www.ftc.gov/business-guidance/resources/complying-take-it-down-act).

### Search engine removal requests

This is where expectations most often outrun reality. Google's own help page on removing information says the process covers specific categories: personal contact information, confidential government identification numbers, financial details such as bank account or card numbers, images of a signature or ID, private records such as medical records, confidential credentials, and doxxing content, which it defines by two conditions, either personal information shared with explicit or implicit threats, or a significant amount of aggregated personal information without a legitimate purpose. Content covered by its sexual-content policies and nude or sexual images of people under 18 are also covered, and content outside its policies can be raised through a legal removal request, where local laws may apply.

The same page is clear about limits. Google only removes results from Search and does not remove content from the web itself. For that, you must contact the website owner. It says it may keep newsworthy or important content up even when it concerns you, and that removed pages may still be reached through links, social media or other search engines. If a removal is granted it can be full, or partial, meaning the page does not appear for searches containing your name or identifier but may appear for others. Source: [Google Search Help, removing information](https://support.google.com/websearch/answer/9673730).

Two further tools matter in a crisis. If you control the site that hosts the material, Google's Search Console Removals tool can hide a page from results quickly, but its documentation says requests last about six months, the page stays live on your site, and other search engines and visitors can still reach it; the durable options are to remove or update the content, or protect it. Source: [Google Search Central, remove information from Google](https://developers.google.com/search/docs/crawling-indexing/remove-information). And if a page has been corrected or deleted but Search still shows old text, Google's refresh tool exists for exactly that, and it says you do not need to file if you have already updated a live page. Source: [Google Search Help, refresh outdated content](https://support.google.com/websearch/answer/6349986).

<figure class="keyfacts">
<div class="keyfacts-grid">
<div><b>About 6 months</b><span>how long a Search Console removal request lasts, with the page still live</span></div>
<div><b>Two conditions</b><span>how Google defines doxxing content eligible for removal</span></div>
<div><b>Search only</b><span>Google removes results, not the page at its source</span></div>
</div>
<p class="fig-src">From Google Search Central's page on removing information and Google Search Help's page on removing information from Google, both read for this guide.</p>
</figure>

### Legal routes

A lawyer can assess defamation, privacy, harassment, data protection or copyright claims, send a letter, or apply to a court. The tests differ sharply by country, and limitation periods can be short. We do not summarize them here because they would be wrong in half the places it is read; take advice in the jurisdiction where the publisher, the platform or you are based. Two practical points hold almost everywhere. A letter is a serious step that can be published by the recipient, so decide in advance whether the story can survive that. And a lawsuit over a true statement will fail while keeping the story in the news for the length of the case.

<aside class="callout">
<span class="callout-label">Rule</span>
<h4>Never ask a route to do what it cannot</h4>
<p>A copyright notice does not remove a true statement about you, a privacy request does not reverse a public record, and a defamation letter does not work on an accurate report. Match the claim to the route before you file, or you create the second article.</p>
</aside>

## What people and AI assistants see while the story is live

The find track starts the first day and does not end when the story fades. Open a private browser window and run searches for your full name, the name plus the topic of the crisis, and the name plus words people would naturally add, such as "lawsuit", "scandal" or "complaint". Record the first two pages, including news results, videos, images, forums and suggested searches. Then ask the main AI assistants the questions a stranger would ask about you, and save their answers with the date. That baseline lets you tell whether things are improving or whether an unflattering detail has been picked up as the headline fact.

During a news event, the first page of results is usually dominated by coverage and discussion within days. You cannot prevent that, and you should not try to. You can make sure the accurate material you control is present, findable and consistent, so that when interest cools, the page is not made only of the worst week.

### What to publish, and when

During the peak, publish only what you would stand behind in a year: a single owned page with your official statement, verified facts, a way for journalists to reach you, and links to your confirmed profiles. Keep it plain and dated. Use the same wording as your statement, and update the page when facts change rather than adding new pages. Do not publish a long rebuttal during the peak; it extends the news cycle and cannot be easily withdrawn.

After the peak, build outward with accurate material that deserves to rank: updated professional profiles, a biography, interviews and genuine contributions to your field. We do not use fake sites, paid links or invented content for this, and we tell clients that those tactics are detected, penalized by search engines and often exposed in a second story. Moving a result down by adding better pages is a matter of weeks to months and is never certain. For a live file, our [reputation monitoring](/reputation-monitoring) page describes how monthly tracking of search and AI answers works.

### AI answers

AI assistants summarize what they find, and during a crisis they can repeat a false claim in confident prose, merge you with a person of the same name, or mention an old matter as current. Check at least weekly during the peak. Where an answer is wrong, the most durable fix is to correct the source it relies on, and to make sure the accurate version exists on pages the assistant is likely to read. The route for AI-specific problems is covered by [AI search reputation](/ai-search-reputation), and each provider has its own way to report inaccurate statements about a person; check the current page of the provider concerned before relying on it.

## A worked scenario: a ninety-day plan for a false-claim crisis

The following is an invented scenario with generic parties, built to show the order of operations. The people are fictional and no real case is described.

A founder of a small consultancy wakes to a social media thread claiming she misled a former client about the results of a project. The thread is spreading and a trade journalist has asked for comment. Facts, as she knows them: the project result was reported accurately, the claimant was unhappy with a delay that was agreed in writing, and she holds the contract amendment and the email chain.

<div class="tablewrap"><table><thead><tr><th>When</th><th>Say</th><th>Fix</th><th>Find</th></tr></thead><tbody>
<tr><td><strong>Hours 0 to 4</strong></td><td>No public reply. Holding line prepared for the journalist.</td><td>Evidence file: contract, amendment, emails, thread captures. Counsel briefed.</td><td>Baseline of search results and AI answers saved.</td></tr>
<tr><td><strong>Day 1</strong></td><td>Short holding line sent to the journalist with an offer to share documents.</td><td>Fact sheet with each claim marked. Platform report only if a rule is broken.</td><td>Owned statement page drafted, not yet public.</td></tr>
<tr><td><strong>Days 2 to 3</strong></td><td>Factual statement published once, with the documents the client agreed may be shared.</td><td>Correction request to the outlet if it prints an error.</td><td>Statement page live and linked from official profiles.</td></tr>
<tr><td><strong>Week 2</strong></td><td>Silence unless new facts. Staff and clients briefed.</td><td>Counsel decides whether a letter helps or merely extends the story.</td><td>Weekly checks of results and AI answers.</td></tr>
<tr><td><strong>Weeks 3 to 6</strong></td><td>Normal publishing resumes on professional topics.</td><td>Any outstanding correction or update follows up.</td><td>New accurate profiles, interviews and case material added.</td></tr>
<tr><td><strong>Days 60 to 90</strong></td><td>Short note to clients on what was learned and changed.</td><td>Contract and communication process updated.</td><td>Results reviewed against the baseline; monthly check set.</td></tr>
</tbody></table></div>

Assumptions: the claim is false or misleading but not defamatory in a way that justifies litigation, the documents may lawfully be shared, and nothing is under regulatory review. Change any of those and the plan changes. If the claimant had a point, the statement would acknowledge it and the fix track would be a remedy, not a rebuttal.

### Where the plan could fail

The founder could post a long thread at hour two, which would give the story its first quotable material. She could ask the outlet to remove the article instead of correcting an error, which tends to turn a one-day piece into a feature. The week-2 line, silence unless new facts, is the hardest to hold.

## Mistakes that extend a crisis, and how to spot them early

Experience points to a short list of errors that repeat across cases. Knowing them in advance lets you spot them in your own team.

**Over-denial.** A broad "none of this is true" that is later contradicted by one fact invalidates everything else you said. Deny only what you can prove, and say "we dispute" when you can prove less.

**Changing the story.** Each revised explanation is read as proof that the first was untrue. Settle the facts first, then speak, even if it takes longer.

**Attacking the source.** Criticism of the person who spoke, the reporter or the platform shifts attention from the claim to your tone. Complaints about process belong with counsel or with the publisher's standards contact.

**Over-lawyering the public voice.** Statements written purely for liability often sound evasive. Counsel should clear facts, not choose adjectives alone.

**Treating removal as the goal.** Time and money go to takedowns that cannot succeed while the facts, the response and the record go unaddressed. A route that fails visibly can also create a new item for search results.

**Ignoring the people closest to the story.** Employees, clients, family members and partners often hear about the crisis from the media. A short, direct note from you before they read it elsewhere does more for trust than any public statement.

**Stopping at the peak.** The story fades, you stop checking, and a small error left in the record becomes the lasting version. Monthly checks for at least a quarter after the peak are cheap and decisive.

### Choosing a pace for the legal track

Counsel should recommend, and you should ask for, a clear view on three things: which claims are actionable, what the realistic outcome is, and what the step would do to the story. A letter that settles a matter privately in a week can be right. A letter that leads to publication of the letter is not. Ask for that second view explicitly, because it is rarely offered unprompted.

## What recovery looks like after the peak

Recovery is slower and quieter than the crisis. It is not a campaign; it is a period in which accurate, useful material accumulates and the proportion of the record taken up by the worst week shrinks. For most files the visible change in what people find is measured in months, and for a serious matter that is accurate and well sourced, the original coverage will remain findable for a long time. The goal is a complete record, not an empty one.

Three habits make recovery faster and more credible. Keep doing the work you are known for, in public, on a regular schedule. Keep the owned statement page accurate and dated, and add a short note when something is resolved, such as a correction by a publisher or the end of a process. And keep a monthly review of search results, news, images and AI answers for your name so a new item is seen in days rather than months.

### When to bring in outside help

Outside help is reasonable when the story is moving faster than you can sort it, when journalists are calling regularly, when private material is involved, or when the matter involves several jurisdictions. Look for a provider who starts by asking what is true, explains which routes apply and which will fail, puts the fix and the statement under the control of you and your counsel, and does not promise outcomes. Be wary of anyone who offers to make accurate coverage disappear, who proposes fake reviews or accounts, or who asks for payment before assessing the facts. Our own approach, as we describe on our [ethics](/ethics) page, rests on the same boundaries.

## Next step: the first two hours

<div class="box hot"><b>If this is happening to you now</b>
<p>Stop posting, save evidence without editing it, and write the one-page fact sheet with every claim marked false, incomplete or true. Then speak to counsel if there is any legal dimension, and decide who will be the single voice. If you want a specialist to read the situation with you and set the three tracks, our <a href="/crisis-management">crisis desk</a> mobilizes within two hours.</p></div>

The first call with a crisis desk follows the order of this guide: sort the facts, protect anyone at risk, fix the record where a route exists, and plan the search and AI picture for the weeks ahead. Reputation Partners does not promise removal of accurate coverage, and an honest advisor will tell you early which parts of your file are removable, which can be corrected and which can only be answered.

## Questions people ask

### How fast should I respond to a reputation crisis online?

Acknowledge quickly only when journalists or a large audience are asking, and keep the first message short and factual. Inside the first day, preserve evidence and sort the claims. A prepared holding line can go out within hours; a full account should wait until you can prove each part. Speed matters less than accuracy, because a wrong first statement is hard to retract.

### Can a crisis be removed from search results?

Only parts of it, and only through specific routes. Search engines remove results under defined policies, such as certain private information or intimate images, and under legal orders. Accurate, newsworthy coverage is usually kept. Removal of a result also does not delete the page. Most files combine corrections, context and new accurate material, rather than depending on removal alone.

### Should I delete my own posts when a crisis starts?

Usually not without a record. Deleted posts are often already captured, and deletion tends to read as concealment. Save copies for your lawyer first. Take down content that is unlawful or harms other people, and say you did. Restricting visibility is a milder step than deletion and can be reversed if you later need to show what was posted.

### Do I need a lawyer and a communications adviser?

If there is any risk of a legal claim, a regulatory step or an employment consequence, yes, a lawyer first. A communications adviser helps most when the story has a public audience. The two should work from the same fact sheet. A lawyer alone often produces cautious statements, and an adviser alone may produce statements that create liability.

### How long does a reputation crisis last?

Most have a peak of a few days, a tail of several weeks, and a recovery that takes months. The length depends on whether the claim is true, how newsworthy it is, whether new developments keep arriving, and how you respond. Accurate coverage of serious matters can stay visible for years even when interest has faded.

### How do I know whether AI assistants are repeating the story?

Ask the main assistants the questions a stranger would, using your full name and the topic, in a fresh session, and save the dated answers. Repeat weekly during the peak and monthly after. If an answer is wrong, find the source it draws on and correct that, and report the inaccuracy to the provider through its own channel.