What Is Online Reputation Management (ORM)?

What is online reputation management?

Online reputation management, usually shortened to ORM, is the process of monitoring and influencing what people find when they search for a brand, a company or a person. That covers search engine results pages, review platforms, social media platforms, news coverage and forums, and now the AI answers assembled on top of all of them. The same discipline covers the reputation of an individual as readily as a company, because the surfaces are identical and only the assets differ.

The definition has a deliberate second half. Monitoring on its own is listening, and influencing on its own is publishing. Online reputation management is the loop between them: find what is being said, decide what needs answering, act, and then measure whether the search result, the review average or the AI answer actually moved. A brand with a monitoring dashboard nobody acts on doesn't have an ORM programme. Neither does one publishing positive content whose effect it never checks.

The scope is wider than your own website, and that's the single distinction separating online reputation management from most other marketing work. Your website is one result. The other nine on the first page of Google, the Trustpilot average, the Reddit thread, the Glassdoor rating and the Wikipedia article are read by the same buyer and, increasingly, by the same language model. ORM is the discipline that treats all of them as one surface.

A good reputation in that sense is not the brand image you publish. It is the online image that survives everything other people have written.

What does online reputation management cover?

Online reputation management covers five surfaces: the four of the PESO Model® (opens in a new tab), created by Gini Dietrich and used by public relations for a decade, plus review platforms, which have earned their own category because they output a number rather than a narrative. Between them they hold every piece of online content about you that a buyer might read, and the online mentions and reviews that make up your online presence are spread across all five rather than gathered in any one.

• Owned media: your website, blog, help centre, careers pages and brand-controlled social profiles. The only surface you can change directly, and much the cheapest to fix.
• Earned media: press coverage, analyst notes, podcast appearances and unpaid mentions. Earned media is what third parties say without being asked, which is why readers weight it heavily and why AI retrieval does too.
• Shared media: social media conversation, Reddit threads, community forums and user-generated content. You participate here; you don't control it.
• Paid media: sponsored posts, native advertising and paid placements. Useful for reach, close to useless for credibility, because readers discount them and assistants rarely cite them.
• Review platforms: Google Business Profile, Trustpilot, G2, Capterra, Yelp, Glassdoor and the sites specific to your category. A star rating is the one reputation signal a buyer or a model can act on without reading anything.

AI answers sit above all five rather than beside them. When somebody asks ChatGPT or Google AI Mode whether a company is any good, the assistant reads the review platforms, the forum threads, the news coverage and the brand's own pages, and returns one paragraph. Online reputation management now has to account for that paragraph as well as its inputs, because for a growing share of buyers the paragraph is the only thing they read.

Why does online reputation management matter?

Online reputation management matters because reputation has become the main input into decisions that used to turn on price and proximity. Online research now precedes almost every purchase of any size, and what it turns up is largely written by other people. In BrightLocal's Local Consumer Review Survey 2026, a representative panel of 1,002 US adults, 97 percent of consumers said they read online reviews for local businesses, 41 percent said they always do, up from 29 percent the year before, and 93 percent said they had made a purchase after reading reviews.

The effect scales well past small business. Weber Shandwick and KRC Research surveyed 2,227 executives across 22 markets for The State of Corporate Reputation in 2020 and found that, on average, those executives attributed 63 percent of their company's market value to its reputation. A Forbes Insights survey for Deloitte made the same point defensively: 87 percent of more than 300 executives rated reputation risk as more important or much more important than other strategic risks.

Reviews are the mechanism most of that runs through, and one online review can outweigh a page of marketing copy. BrightLocal's 2026 panel found 31 percent of consumers will only use a business rated 4.5 stars or higher, and Khoros reported that 83 percent of customers feel more loyal to brands that respond to and resolve their complaints. That second figure is the useful one. A negative review answered well is a different asset from a negative review ignored, rather than a slightly smaller liability.

Positive reviews and the occasional customer testimonial are the visible output of customer satisfaction, which is why the fastest route to more of them is more happy customers and not a better collection widget.

The revenue link has been measured properly at least once. Michael Luca's Harvard Business School study matched Yelp ratings against Washington State revenue records and found a one-star increase drives 5 to 9 percent more revenue, with a caveat almost everyone who quotes it drops: the effect held for independent restaurants and vanished for chains, whose brand already told the buyer what to expect. The scope matters as much as the number.

And the reason online reputation management has changed shape rather than simply grown sits in the same BrightLocal data. Use of ChatGPT for local business recommendations went from 6 percent to 45 percent in a single year, 82 percent read AI-generated review summaries, 40 percent say they trust AI platforms to recommend businesses, and 23 percent are willing to decide on the AI summary alone without opening a single review.

What is the difference between ORM and SEO?

The difference between SEO and ORM is scope and goal. Search engine optimisation works on one website and tries to win traffic for the searches that describe a product or a problem. Online reputation management works across every result a branded search returns, most of them on domains you don't own, and tries to change what a person concludes rather than where they click.

They aren't alternatives, and treating them as a choice is the most common mistake in the category. Online reputation management uses search engine optimisation as one of its tools: to put a positive asset on the first page you have to rank it, and ranking it is SEO. Reputation X puts it neatly, that if ORM were a car then SEO would be the engine. What differs is the target. SEO asks which page should rank for a keyword. ORM asks which ten results should exist for a name.

The practical consequence is a different keyword list. An SEO keyword list describes the problem the product solves. An ORM keyword list is the brand name, the product names and the founders' names, each of them followed in turn by reviews, complaints, scam, lawsuit, alternatives and refund. Those are the searches a buyer runs last, immediately before deciding, and almost nobody tracks them.

Search engine reputation management is the name for the overlap, and it's where most of the budget actually goes: publishing and ranking assets until negative search results are pushed off page one. You can't delete a result you don't own, so the work is to outrank it, and the honest word for that is suppress. Suppress negative content by giving the search engine ten better answers to the same query, and the online brand reputation a buyer sees changes without anything being removed.

How is ORM different from PR and brand reputation management?

Online reputation management differs from public relations in where the work lands. Public relations tries to change what publications write. ORM tries to change what a search result, a review page or an AI answer says, which may or may not involve a journalist at all. The two meet in digital PR, where coverage is earned specifically because a mention on an authoritative domain does reputational work inside search and inside AI retrieval.

Press releases, influencer partnerships and the rest of the digital marketing toolkit all feed the same surfaces, which is why ORM is best treated as a layer across the marketing function rather than a line item inside it.

Brand reputation management focuses on the broader picture and takes in offline perception too: staff behaviour, product quality, sponsorships, what customers tell each other in person. Online reputation management is its digital subset, and it's the subset that scales, because everything on it is written down, indexed, and quotable by a machine.

How do AI assistants decide what to say about your brand?

AI assistants build a brand's reputation mostly out of other people's writing. Kevin Indig's citation dataset found brands are roughly 6.5 times more likely to enter an AI answer through a third-party source, a review site, a forum or a publisher, than through their own domain, and the skew widens as commercial intent rises. That one finding rearranges the ORM budget: on the questions where money is decided, your own website is the minority contributor to what the assistant says about you.

Which third parties carry the weight is measurable. Peec AI analysed 30 million cited sources across ChatGPT, Google AI Mode, Gemini, Perplexity and AI Overviews, reported in March 2026, and found Reddit the most-cited domain overall, followed by YouTube, LinkedIn, Wikipedia and Forbes. The per-engine split matters more than the ranking: ChatGPT favoured Wikipedia, Reddit and editorial sites such as Forbes, Google's surfaces leaned towards Facebook and Yelp, and Perplexity emphasised Reddit, LinkedIn and G2 on B2B questions. Review platforms recurred throughout recommendation queries on every engine.

Being read isn't the same as being named, and the gap is wide enough to have earned its own name. A Semrush study published in June 2026, covering 3,981 domain appearances across 115 prompts, 14 countries and four AI engines, found 62 percent of citations were ghost citations: the assistant linked the domain as a source and never said the brand's name in the answer. Domains were cited in 74.9 percent of appearances and mentioned in only 38.3 percent, and just 13.2 percent of appearances produced both.

Engine behaviour is almost inverted. ChatGPT cited in 87 percent of appearances and named the brand in 20.7 percent, while Gemini named the brand in 83.7 percent and produced a link only 21.4 percent of the time.

For online reputation management, that ghost-citation number is the entire argument for tracking unlinked brand mentions rather than backlinks. Across 75,000 brands, Ahrefs measured plain web mentions correlating with AI visibility at a Spearman coefficient of 0.664 against 0.218 for backlinks, roughly three times stronger. A brand mentioned widely and linked rarely is in better shape inside AI answers than the reverse, which is close to the opposite of what a link-building budget assumes.

None of this is visible from a rankings report. Seeing it takes a panel of prompts run repeatedly against each assistant with the answers stored, which is what AI visibility tracking does, and what an AI brand sentiment score is calculated from. Until somebody samples it, how an assistant describes your brand online is simply unknown.

What happens when an AI gets your reputation wrong?

When an AI assistant states something false about a brand, the damage to the company's reputation arrives before anybody can correct it, and the correction has nowhere obvious to go. The clearest documented case is Wolf River Electric, a Minnesota solar installer that sued Google in March 2025 after Google's AI Overview told searchers the company faced a lawsuit from the state Attorney General. The Attorney General had sued four solar-lending companies. Wolf River Electric wasn't among them.

Sales staff began noticing cancelled contracts in late 2024, with customers citing what they had read in Google, and the company's disclosures put damages between 110 and 210 million dollars. The case was remanded to Ramsey County District Court in January 2026.

The mechanism is ordinary AI hallucination rather than anything exotic. An assistant holding several conflicting sources about what a company does, or one that has confused two similarly named organisations, produces a fluent sentence with no flag on it. The reader can't separate a retrieved fact from a generated one, because the answer arrives as a single paragraph in a single voice.

What makes this an ORM problem rather than an engineering one is that every fix is upstream. There's no edit button on a language model. Correcting what assistants say means changing what they read: publishing an unambiguous, entity-dense description of the company on its own site, getting the correct version onto the third-party sources these engines actually cite, and keeping the brand name, description and category identical everywhere, including the knowledge panel and every directory, so that nothing is left to confuse.

Where a specific false claim is indexed, the older instruments still apply, including delisting requests and, in the EU and UK, an Article 17 erasure request under the GDPR that the 2014 Google Spain ruling established for search results. Be clear about what delisting does: it removes a result from a search engine, it doesn't remove the page from the web, and it retracts nothing a model has already trained on.

How do you build an online reputation management strategy?

An online reputation management strategy is a monitoring loop with an escalation path attached, not a content calendar. To manage your online reputation you first have to know what it currently says, so a reputation management plan that opens with content creation has skipped the diagnosis.

Steps one and two below are what we call the two-surface audit: the branded first page is surface one, the AI answer is surface two, and each takes an afternoon. Everything before step four tells you what's actually wrong, and most reputation budget is wasted by starting at step seven. The output of the whole management plan is a positive online presence that holds up under scrutiny, not a folder of assets nobody checked.

1. Audit the branded first page. Search the brand name, the product names and the founders' names in a private window and record the top ten results and whether each reads positive, neutral or negative. Repeat with reviews, complaints, alternatives and scam appended. This is the baseline, and surface one of the two-surface audit.
2. Audit the AI answer. Ask ChatGPT, Gemini, Claude, Perplexity and Google AI Mode what the company does, whether it is any good, and who its alternatives are. Save the answers, check every factual claim, and note who else gets named. Surface two, and the one most brands have never looked at.
3. Claim and complete every profile. Google Business Profile, the review platforms your category uses, LinkedIn, Crunchbase, Wikidata. One canonical brand name spelled identically, one description, everywhere. Inconsistent naming is the most common reason an assistant describes the wrong company.
4. Fix whatever the complaints are actually about. Reputation work applied to a genuine product or service failure is expensive and temporary. Read the negative reviews for the pattern before writing a single positive asset.
5. Build review volume deliberately. Ask every satisfied customer at the moment the value lands, through a process rather than as a favour. Volume and recency both matter more than the average, because 4.6 stars from 900 reviews reads as real and 5.0 from 11 reads as staged.
6. Respond to reviews. A bad review answered in public is worth more than a good one nobody replied to, so reply quickly, without arguing, with a route to resolution off the platform. Responding to reviews is done for the next reader, not for the complainant.
7. Publish and rank owned and earned assets on the terms that matter. This is the content strategy half of the job: an about page that answers who, what and why in plain sentences, a pricing page with real prices in the HTML, founder profiles, and earned coverage on the domains the engines cite. This is where SEO does the work for ORM, and where a positive online reputation is actually built.
8. Prepare the crisis path before you need it. Crisis management is cheap to plan and ruinous to improvise: decide who decides, who writes, who can publish inside an hour, and what the holding statement says.

How do you measure online reputation management?

Effective online reputation management is measured on the surfaces themselves, not on traffic, and a reputation strategy nobody can measure is a hope with a budget attached. Four numbers are worth reporting: the sentiment mix of the branded first page, the review average and volume per platform with their trend, share of voice against named competitors, and the rate at which AI assistants name the brand together with what they say when they do.

That last number carries a measurement trap. SparkToro and Gumshoe ran 2,961 repeated prompts and found under a 1 in 100 chance that two runs of the same prompt return the same list of brands. A single check of what ChatGPT says about you is not a measurement. It is one sample drawn from a distribution. What holds up is aggregate share across a panel of dozens of prompts, run repeatedly, reported per platform, and read as a trend rather than as a reading. Setting that up is its own exercise: how to track brand mentions in AI search covers the prompt set, the cadence and the states a tracker has to distinguish.

We ran this on ourselves before recommending it to anyone. RankX AI tracked its own brand across ChatGPT, Claude, Gemini, Grok and Perplexity for 30 days: 218 analysed answers, a 20.2 percent overall mention rate, a spread from 11.4 to 23.3 percent by platform, and Google AI Overviews appearing on 59 of 60 checks without citing the site once. The full 30 day baseline is published, zero citations included, because a reputation metric you wouldn't publish about yourself isn't a metric worth selling.

The value of the AI half isn't hypothetical either. Seer Interactive tracked 53 brands across 5.47 million queries and 2.43 billion organic impressions from January 2025 to February 2026, and found that being cited in an AI Overview returns 120 percent more organic clicks per impression than sitting uncited under the same Overview. Being named in the answer moves the numbers on every other channel, which is why an AI mention belongs on the reputation report beside the star rating.

Read that with the figure nobody quotes beside it, from the same dataset. Per million informational impressions Seer measured about 33,500 clicks with no Overview, 20,743 cited, and 9,445 uncited. Citation is worth roughly 2.2 times an uncited appearance and still lands about 38 percent below no Overview at all. On those queries it is damage control, not upside.

The tooling splits along the same line as the work. Reputation monitoring divides into pieces: social listening tools and media monitoring tools cover mentions and sentiment across social and news, a review management tool covers the online review platforms, and online reputation management software bundles the two together with reporting. None of those monitoring tools read AI answers, because an AI answer is generated per request rather than published, so it has to be sampled rather than crawled.

RankX AI measures the AI half of this directly. AI Visibility runs a prompt panel across ChatGPT, Claude, Gemini, Grok and Perplexity on a schedule and reports mention rate and sentiment as an aggregate rather than from a single run, and AI Overviews records whether Google shows an Overview on your tracked keywords and whether it cites you. The free AI Overview checker does the one-off version of the second, and how to measure AI search visibility covers the method in full.

How much does online reputation management cost?

Reputation management companies price the work in a wide band. Published online reputation management pricing runs from roughly 500 to 2,000 dollars a month for a small business, 2,500 to 7,500 for mid-market and executive work, and 3,000 to 15,000 or more a month for enterprise programmes, with one-off suppression projects quoted between about 1,500 and 15,000. Read those as agency list prices gathered from published pricing pages rather than as market data. Reputation management agency pricing is unusually opaque even by the standards of marketing services.

Two variables drive the real number more than anything on a rate card. The first is whether the negative content is true, because a false claim can sometimes be removed while a true one can only be outranked, and outranking is a months-long content and links programme. The second is how much authority the negative result carries. A national newspaper sitting on a low-volume brand name is a different problem from a forum thread. Ask any agency which of those two situations you are in before you ask what it costs.

The monitoring and review half is viable in-house, and an online reputation management tool for monitoring starts in the low hundreds of dollars a month. What doesn't scale in-house is the publishing programme, because that part is SEO and it needs the same skills and the same patience. That, rather than the monitoring, is what reputation management services are really charging for.

How do you choose reputation management software or an agency?

The market splits into software you run and services you hire, and the selection test is the same for both: can it see every surface your reputation actually lives on? Most tools were built for one or two of the five surfaces and stretched afterwards, so ask these five questions before comparing prices.

• Coverage. Which of the five surfaces does it read: reviews, search results, social conversation, news mentions, AI answers? Most cover two. The gap it can't see is the gap you'll be surprised by.
• AI sampling method. If it reports what ChatGPT says about you from a single run, it's selling noise; the SparkToro data above shows why. Aggregate share over repeated prompts, per platform, is the only defensible read.
• Workflow, not just watching. Monitoring without a route to respond, escalate and resolve is a dashboard, not a management tool. Ask who gets alerted, in what time, with what next step.
• Checkable reporting. Every number should carry a source, a sample size and a date, for the same reason every number in this article does: so your finance director can verify it.
• For an agency, one diagnostic question. Ask whether they treat true and false negative content differently. The honest answer is a completely different plan for each; a pitch that doesn't distinguish them is selling a package, not a fix.

Ranked vendor lists date in months and read as adverts, which is why there isn't one here. The questions age better than any shortlist.

Which reputation management statistics should you not trust?

The most-quoted statistic in online reputation management should be treated as folklore. The claim that one negative result on the first page of Google costs a business 22 percent of prospective customers, rising to 44 percent for two, 59 percent for three and around 70 percent at four or more, appears on hundreds of agency pages and in almost every guide to online reputation management, including guides published this year. It is usually attributed to Moz, or to research Moz conducted for Google.

The primary source can't be located. No paper, no dataset, no methodology and no date travel with it anywhere it appears.

Directionally it's almost certainly right, which is exactly why it survives. But a number with no method behind it can't size a budget, and quoting it to a finance director who checks is worse than quoting nothing at all. The same caution applies to market-size figures for the category: published 2026 estimates of the online reputation management market range from about 340 million dollars to about 6.9 billion depending on which vendor drew the boundary, a span too wide to carry meaning.

The Yelp study earlier in this article shows what the credible version looks like: a named researcher, a public dataset, a stated method and a scope the effect holds inside. The statistics worth using carry a sample size, a date and a method. Every figure in this article carries its source and date inline, in the sentence that quotes it, so the arithmetic can be checked rather than trusted.