It is not defamation removal alone. It is not review suppression. It is not SEO with extra steps. It is not a one-time cleanup. It is a permanent operation, because the substrate is permanent.
ORM done well looks like a media company running inside a brand: producing primary content, maintaining entity infrastructure, monitoring the search and social surface, and engaging with the press, the platforms, and the review ecosystems where reputation actually lives.
The Three Eras of Reputation Management
The discipline has moved through three eras. Each one expanded the surface area without retiring the prior one.
Era 1, SERM (2003-2014): the search-result engineering era
Search Engine Reputation Management. The objective: control the first page of Google for a brand-name query. The tactics: own the top ten results through controlled domains, press placements, owned social profiles, and SEO discipline. The metric: who ranks for "[your name]."
SERM is still operative. The first page of Google for an executive's name remains a job interview, an investor diligence check, a journalist's first stop. It stopped being the whole job around 2014, but it never stopped mattering.
Era 2, ORM (2014-2023): the multi-surface era
Online Reputation Management expanded the surface. Glassdoor, Yelp, TripAdvisor, Trustpilot, Indeed, Reddit, LinkedIn, Twitter, YouTube, Instagram, TikTok. Each one indexed. Each one ranked. Each one capable of dominating a brand-name query.
The discipline became multi-surface management: reviews on review platforms, presence on social platforms, content on owned platforms, press in earned media, all operating in parallel. The firm landscape professionalized. Status Labs, Reputation.com, Terakeet, BrandYourself, ReputationDefender, the modern ORM firms were built or scaled in this window.
Era 3, AI Reputation Management (2023-present): the answer-engine era
The answer engines, ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews, now sit above search and social as the layer that mediates the first impression. The consolidated canonical for the AI era is AI Reputation Management: The Five Layers, the Six-Hour Clock, and Citation Share. The shared vocabulary of the field is collected in the AI reputation glossary and the communications and reputation vocabulary.
The Five Reputation Layers
A working ORM operation runs five layers in parallel. Each layer has its own substrate and its own mechanics.
Layer 1, Press. Earned coverage in publications the search and discovery systems treat as authoritative: the major national press, the trade press for the relevant vertical, the analyst layer. Press is the original reputation substrate and remains the most expensive to influence honestly. How the two disciplines divide the work is covered in PR vs Reputation Management, and how to score and pursue coverage is in the PR and SEO plan.
Layer 2, Social. LinkedIn, X, Instagram, TikTok, YouTube. The owned-presence layer. Empty or hostile social is a credibility tax against the rest of the operation.
Layer 3, Wikipedia. The single highest-weighted source for almost every external reputation read, journalists, investors, recruiters, partners. A thin entry, a hostile entry, or no entry at all is a structural problem. The Wikipedia sub-cluster on EPR opens at The Wikipedia Hub.
Layer 4, Owned media. The brand's own website, executive bios, founder blog, FAQ pages, press room. The pieces the brand fully controls. The page-level basics are in the search engine optimization guide.
Layer 5, Reviews and platforms. Glassdoor, Yelp, Trustpilot, TripAdvisor, Indeed, the App Store and Play Store for software, the platform-specific review ecosystems for restaurants, hotels, hospitals, schools, and professional services. How review platforms feed the answer engines is covered in Online Reviews and AI Citation Share, and the legal limits are in the FTC fake review rule.
The Wikipedia chokepoint
Wikipedia is the most-cited single source in external reputation reads. Journalists check it before they call. Investors check it during diligence. Recruiters check it before they reach out. Partners check it before they sign.
A company without a Wikipedia entry is harder for the outside world to talk about confidently. A company with a thin, dated, or adversarial Wikipedia entry has its weakest framing repeated wherever the brand is researched.
Wikipedia is its own sub-cluster on EPR, anchored at The Wikipedia Hub, because the platform is infrastructure, not encyclopedia. Editing it well is a discipline of its own: notability sourcing, neutral point-of-view drafting, citation engineering, and conflict-of-interest disclosure. Wikipedia penalizes undisclosed or poorly sourced edits, so the work follows its conflict-of-interest and sourcing rules. The full thesis, including the executive risks and the first-24-hours crisis protocol, is in Wikipedia Is the AI Reputation Chokepoint. The structured-data layer behind the prose entry, Wikidata, is covered at Why Wikidata Is the Quiet Infrastructure Behind AI Finance and Wikipedia, Wikidata, and the AI Citation Stack for Universities.
The visual layer
Reputation is not only text. AI engines now read images natively, generate them, and cite them in answers, which has moved image reputation work from Google Image Search suppression to engine-layer source shaping. Deepfakes, hallucinated visual answers, and Citation Share for visual queries are an emerging part of the same discipline.
How an ORM engagement runs, step by step
A competent online reputation management engagement runs in six stages, and the order matters because each stage depends on the one before it. Firms that skip straight to content production are guessing, and clients that skip the baseline cannot tell whether the work moved anything.
Stage 1, audit and baseline. The work starts by documenting what exists. That means a fixed query set (the brand name, the executive names, and each name paired with words like reviews, complaints, lawsuit, and scam), the first two pages of results for each, the state of the Wikipedia entry, every review profile, and the answers the major AI engines give to the same prompts. The baseline is dated and saved, because every later claim of progress is measured against it.
Stage 2, source mapping. Every unwanted result has a source, and the source determines what can be done. A negative review on a platform the client can respond to is a different problem from a news article that is accurate, and different again from a court record. Mapping which sources are influenceable (owned media, earned media, review responses, corrections) and which are not (archives, public records) sets realistic expectations before any money is spent.
Stage 3, triage. Findings sort into four buckets: factual errors that can be corrected at the source, outdated material that can be updated, hostile but accurate material that can be contextualized and displaced over time, and potentially defamatory material that belongs with counsel. Mixing the buckets is the most common way engagements go wrong.
Stage 4, build. The build phase produces the assets that will own the surface: executive and company bios with consistent entity data, a newsroom and FAQ content the engines can quote, earned coverage, a sourced Wikipedia presence where notability supports one, and a review program that earns ratings honestly. Consistency matters more than volume, because the engines reconcile across sources and punish contradictions.
Stage 5, monitoring. The same query set is rerun on a fixed cadence, weekly for acute situations and monthly otherwise, with alerts on new negative results and on changes in AI answers. Monitoring is also what catches the next problem before it hardens.
Stage 6, reporting on outcomes. Reports should show movement against the Stage 1 baseline: which results changed, which errors were corrected, how ratings and review volume moved, and how AI answers shifted. The crisis-time version of this sequence, compressed into the first hours, is laid out in the Crisis Management Guidebook, and the 12-tip checklist for individuals and small teams follows below.
Why does every reputation event run on two clocks?
Every reputation event runs on two clocks: a news clock and an answer clock. The news clock covers statements, briefings and corrections, and it usually resolves within days. The answer clock starts when search engines, news archives and AI engines begin indexing the story, and it keeps running for months or years.
A reporter or buyer who looks up what happened to a brand years later meets the version of events that won the long record. A response built only for the news cycle can end the headlines and still leave the long record to other people, which is why the first week sets the baseline that the rest of this guide measures against. The hour-by-hour sequence for the first hours and the first week is in the Crisis Management Guidebook.
What are the 12 online reputation management tips?
The 12 tips run from a baseline audit to knowing when to hire a specialist, and the order matters because each step builds on the one before it.
| # | Tip | What to do |
|---|
| 1 | Audit what is there | Search your name or brand on Google and each AI engine, and save the top 10 results and the full AI answers as your baseline |
| 2 | Own your name | Register your domain and claim handles on LinkedIn, X, Instagram, YouTube, TikTok, Facebook and Threads to prevent impersonation |
| 3 | Build one canonical site | Publish a single site on your own domain with bio, history, work samples, press and contact details, linked from your profiles |
| 4 | Get Wikipedia right | If you have an entry, audit it for accuracy and sourcing. Do not edit your own entry, because Wikipedia's conflict-of-interest rules discourage it |
| 5 | Build authority where engines look | Earn coverage in established and trade press, and publish on LinkedIn, YouTube, podcasts with transcripts and reputable directories |
| 6 | Publish under your own byline | Write op-eds, bylined articles and expert commentary to build primary sources over time |
| 7 | Outweigh negative results | Publish authoritative content so negative items fall in prominence, instead of making legal threats |
| 8 | Manage reviews on purpose | Respond professionally, ask satisfied customers for reviews, and never buy reviews |
| 9 | Keep entity data consistent | Use identical names, titles, dates and facts across LinkedIn, your site, press bios, Wikipedia and IMDb |
| 10 | Re-audit AI answers | Rerun the baseline queries quarterly, or monthly in an active situation |
| 11 | Plan recovery in years | Treat a serious reputation event as a multi-year project |
| 12 | Know when to hire | Bring in a specialist for executive, public-figure or crisis situations |
Why do outside sources matter most?
Outside sources shape AI answers more than a person's own pages do. Muck Rack's May 2026 What Is AI Reading? report analyzed more than 25 million links cited by ChatGPT, Claude and Gemini and attributed about 84% of citations to earned media, a broad category that includes journalism, academic research, government sources, Wikipedia and similar sites, social platforms and third-party corporate content. Journalism alone was about 27%. Muck Rack sells media-monitoring and AI-visibility software, so read the figure as vendor research.
What ORM can and cannot do
Online reputation management can shape the inputs that search engines and AI engines read, but it cannot rewrite the record or guarantee an outcome. Being clear about that boundary is the fastest way to judge a firm's honesty.
What it can do. It can correct factual errors at their source, build authoritative assets that earn the top of the results page for a name, add context around accurate but damaging coverage, earn and respond to reviews, maintain a sourced Wikipedia entry, and keep the entity data consistent across every profile so that engines describe the subject accurately. It can also monitor continuously and shorten the time between a problem appearing and a response.
What it cannot do. It cannot erase accurate reporting, court records, or regulatory filings. It cannot force a platform to remove content that complies with the platform's rules. It cannot edit what an AI engine says directly; it can only change the sources the engine retrieves from. And it cannot deliver instant results: most observable shifts take months of sustained work, with six to twelve months typical for the engine layer, as covered in the AI Reputation Management canon.
The removal question. Removal routes exist but are narrow. Platforms and search engines offer policy-based removal for specific categories such as certain personal information, and courts can order removal of content found to be defamatory. These are the exception, not the method, and they usually run through counsel. The legal boundaries of the more aggressive tactics are examined in Is Search Result Suppression Legal? A firm that guarantees removal of accurate coverage is selling something it cannot deliver.
Where tactics become liabilities. Fake reviews, impersonation, and undisclosed paid editing on Wikipedia are not gray areas. They violate platform rules, and in the case of fake reviews they now carry regulatory exposure as well. The Federal Trade Commission's final rule on consumer reviews and testimonials, announced August 14, 2024, lets the agency seek civil penalties of up to $51,744 per violation at the 2024 level, a figure adjusted annually for inflation, as explained in the FTC fake review rule. A reputation built on shortcuts is a reputation one exposure away from collapse.
When the damage is already done
Recovery is slower than the original damage, and it follows a different logic from prevention. The concept of accumulated exposure, reputation debt, and realistic timelines are set out in Reputation Recovery Timelines, the point at which crisis response should give way to repair is covered in Crisis Response to Reputation Repair, and the longer arc of rebuilding a public image is traced in The Reputation Reinvention Playbook.
The Terakeet investigation
In May 2026, the New York Times investigation into Terakeet, a major ORM firm that has been quietly running reputation operations for Fortune 500 boards, including a multi-year engagement for Goldman Sachs in the aftermath of the Epstein revelations, opened the industry to public view in a way it had not previously been.
What the investigation revealed: the largest ORM operators are running coordinated, multi-surface, search-result-displacement campaigns at the seven-to-eight-figure-per-year price point. The legality remains contested. The EPR investigation series opens at Terakeet, The $100M Firm From Syracuse and spans five pieces.
The firm landscape
The ORM firm landscape sorts into four tiers. The full framework lives at Best Reputation Management Companies: 31 Ranked, which also collects every firm profile and the full Terakeet series in one place.
Tier 1, Dedicated specialists. Firms that do ORM as the core product. Status Labs, Terakeet, Five Blocks, and a layer of smaller boutiques. Highest per-engagement spend. Most willing to run aggressive search-result displacement.
Tier 2, Major PR firms with ORM practices. Edelman, Weber Shandwick, Burson, Golin Ketchum, FleishmanHillard, and 5W. Reputation work runs inside a broader communications operation, press, digital, and crisis, and the ORM component is built on earned-media access rather than search engineering alone. Everything-PR News, LLC is owned by a shared interest group that includes the founder of 5W, and 5W is listed here on the same basis as every other firm.
Tier 3, Crisis communicators. FGS Global (which now includes Sard Verbinnen), Joele Frank, Brunswick, Kekst CNC. Engaged when reputation is acute, litigation, M&A, shareholder activism, public scandal. ORM is a downstream consequence of the crisis work, not the product.
Tier 4, SEO agencies with ORM offerings. The long tail. Variable quality. Capable of producing the search engineering layer cheaply, less equipped to handle the press, Wikipedia, or platform layers.
How to choose an ORM firm
The right ORM firm is the one whose type matches the problem, because the tiers above solve different problems and are priced accordingly. A founder with a search-results problem, a hospital network with a review problem, and a public company in a crisis need three different kinds of firm.
Match the problem to the type first. Individuals and small businesses dealing mainly with search results and review profiles are served by the consumer and SMB firms profiled in the directory. Multi-location businesses managing ratings at scale are served by the enterprise platforms. Executives and high-profile individuals with press and Wikipedia exposure need a specialist or a PR firm with a reputation practice. Acute events tied to litigation or transactions call for a crisis communicator.
Then test the firm. Ask which tactics it will not use, and how it handles Wikipedia and conflict-of-interest disclosure. Ask who does the work, who reports to you, and what a monthly report contains. Ask for references from engagements that lasted more than a year. Treat any guarantee of specific outcomes, any promise to remove accurate coverage, and any pitch that leans on purchased reviews as disqualifying. Pricing varies enormously, and the full range is analyzed in Reputation Management Cost in 2026.
How to measure ORM
The measures that matter are the ones tied to the baseline: whether the surfaces a buyer actually reads now say what they should. A serious program tracks six things on a fixed cadence.
- Brand-query ownership. The share of the first two pages for the name and its risk-modified variants that are accurate or favorable.
- Review health. Average rating, review volume, and response time on each platform that matters to the audience.
- Wikipedia and entity accuracy. Whether the entry, the structured data, and the major profiles are accurate, sourced, and consistent.
- AI answer accuracy and Citation Share. Whether the major engines describe the subject correctly and how often they cite it, as defined in the AI Reputation Management canon. The audit method is in The AI Visibility Audit.
- Time to correct. How long a factual error stays live between detection and fix.
- Business outcomes. Win rates, recruiting conversion, and diligence friction, which are the reason the work is funded in the first place.
Sentiment scores alone hide where the damage lives, which is why the seven-layer recovery dashboard measures dimensions separately. Report movement over quarters rather than weeks, because the surfaces that matter most refresh slowly.
Common ORM mistakes
Waiting for a crisis. The cheapest time to build reputation infrastructure is before there is a problem. Operations that start during an incident are building assets under pressure and usually lose the first window.
Suppressing instead of correcting. Pushing a result down leaves the underlying error in place, and answer engines read the sources, not the ranking. Where an error can be fixed at its source, that fix outlasts any displacement.
Ignoring Wikipedia and the AI layer. Programs built entirely around Google results miss the surfaces that now shape the first impression. A clean first page with a hostile Wikipedia entry is still a hostile reputation.
Inconsistent entity data. Different titles, dates, and descriptions across the website, LinkedIn, Crunchbase, and press releases force the engines to average them, and the average is rarely flattering.
Buying reviews or shortcuts. Purchased reviews, fake profiles, and undisclosed editing are the fastest route from a manageable problem to an unmanageable one.
Measuring activity instead of outcomes. Reports full of articles published and impressions delivered say nothing about whether the surfaces a buyer reads have changed.
Hiring on a guarantee. No firm controls third-party platforms or AI engines. A guaranteed outcome is a sign the firm will tell you what you want to hear.
Vertical applications
ORM is not one playbook. It is a discipline applied to a substrate, and the substrate varies by vertical.
- Executive and personal. Founders, athletes, politicians, family-office principals. Covered at Personal Reputation Management for Founders, Athletes, and Politicians and Executive Reputation Management.
- Corporate. Governance, stakeholders and metrics at the institutional level. Corporate Reputation Management.
- B2B and enterprise. Enterprise procurement committees research vendors on review sites such as G2, Gartner Peer Insights and TrustRadius, in Gartner and Forrester reports, and increasingly in AI answers. The work centers on four things: case studies published openly, because AI engines generally cannot read content behind a login; review depth on those platforms; analyst relations; and executive visibility through interviews, talks and bylined writing.
- Casino and gaming. Regulatory, social, host-community reputation. Casino Reputation Management.
- Law firm. Competitive reputation, partner mobility, representation-choice scrutiny. Covered in Public Relations and Law Firms.
- FDA-regulated brands. Warning letters, recalls, regulator-facing reputation. FDA Reputation Management: From Canada Dry to JUUL.
- Celebrity and talent. Scandal cycles, awards positioning, legacy management. Celebrity Reputation Management and what the MSG data leak means for celebrity reputation.
- Restaurant and hospitality. Review platforms as primary substrate. Covered at Restaurant PR.
What ORM costs
The pricing band is the widest in professional services. A few hundred dollars a month for a small business buying a monitoring tool and a content package. Several million dollars a year for a Fortune 100 board running a multi-surface, multi-firm reputation operation. The full analysis lives at Reputation Management Cost in 2026.
The buyers writing checks at the high end are increasingly interested in outcome-tied engagements rather than activity-based retainers, measurable reputation outcomes on the surfaces where the audience actually researches. What those executives and boards are asking for is covered in the next section.
What are CEOs and boards asking about reputation?
CEOs and boards now ask five practical questions about reputation, and the common thread is that the answer lives in what search and AI engines retrieve, not only in what the company publishes. This is an editorial reading of public executive commentary, not a survey of CEOs, so treat it as a directional guide to the conversation.
- What do the engines say about us? The starting point is a dated baseline of what ChatGPT, Claude, Gemini, Perplexity and Google AI Overviews say when a customer, recruiter, partner or journalist asks about the company.
- Who controls the answer? No single party does. The engines draw on Wikipedia, press archives, filings, research, reviews and reference content, so the answer shifts when the sources shift.
- Why doesn't our existing communications work solve this? Press releases, owned content and paid media still matter, and they work best as inputs that feed the sources the engines read.
- What gets us into the answer? Original research others cite, a sourced Wikipedia presence, consistent entity data, and named executives who publish under their own byline.
- What is our exposure if a crisis surfaces? Negative framing can outlast the news cycle in search and AI answers, so the useful habit is to monitor what comes back for the company name paired with words like lawsuit and investigation.
Boards tend to turn the same questions into governance. They ask for a measured baseline against peers, a named owner for the work, regular reporting on risk, and a view on how visible the CEO and other officers are as authorities. Companies that put an owner and a baseline in place early have the easiest path, because every later quarter of reporting builds on the first one.
Which reputation cases show the discipline in action?
Four widely reported cases show the mechanics of reputation management, from a press-era recall to a values-led ownership change.
| Case | What happened | Lesson |
|---|
| Johnson & Johnson Tylenol, 1982 | After the cyanide-tampering deaths, Johnson & Johnson recalled about 31 million bottles, put its CEO in front of the press and introduced tamper-evident packaging. Full case: the Tylenol crisis of 1982 | Speed, transparency and structural action beat defensive messaging |
| Domino's, 2009 to 2010 | After a viral contamination video, Domino's responded publicly, then acknowledged that customers disliked its pizza and reformulated it in a 2010 campaign | Use criticism to fix the product, not only the message |
| Old Spice, 2010 | Procter & Gamble's "The Man Your Man Could Smell Like" campaign reset how the brand was perceived | A brand's meaning can change without a new product |
| Patagonia, 2022 | The Chouinard family transferred ownership of the company to a trust and a nonprofit focused on the climate | Actions that match stated values compound into credibility |
For longer recovery arcs, see The Reputation Reinvention Playbook.
The Reputation Management Cluster, Full Directory
AI-era canonical
Discipline and framework
Operational playbooks
Wikipedia sub-cluster
Terakeet investigation series
Firm profiles and landscape
Vertical applications
Pricing
