Brand Reputation Management

Nearly half of what people search in Google is a brand name.

Ahrefs studied roughly 150 million U.S. keywords and found that 36.9% of distinct search queries contain a brand, product, or trademarked term. Weighted by how often those queries actually get run, branded searches account for 45.7% of all Google searches (Ahrefs, May 2025). Rand Fishkin prompted the second calculation by pointing out that counting distinct keywords and counting real searches are different measurements. The gap between 36.9% and 45.7% is where the argument lives: branded queries get run more often than the average query.

Brand reputation management is the discipline of governing what those searches return.

Construction and burial are different jobs

Most reputation budgets get approved after something has already gone wrong. A story runs, a review thread gains traction, a former employee posts, and the organization looks for someone to make it stop. That work is suppression, and it is the reactive half of this discipline.

The proactive half runs on a different clock. It asks what page one for the company name ought to contain, builds the assets that occupy those positions, and does the work before there is anything to bury. Both halves share most of their tactics. They do not share timing, and timing is what sets the price. Displacing an established result takes months and money. Occupying the position first takes neither.

The sequence is not optional either. A suppression campaign launched without an existing foundation of owned assets, claimed profiles, and indexed content does not underperform. It has nothing to amplify. That is the most common reason suppression work fails, and it is a construction problem wearing a displacement label.

This is why the proactive half stays underfunded. Nothing visibly breaks when a company skips it. The cost shows up later, denominated in suppression retainers.

The branded SERP is an assembled document

No single party writes the page that appears when someone searches a company name. It is assembled at query time from sources with different owners, different incentives, and different rates of change:

Owned properties: the homepage, product pages, careers pages, support documentation, and the pages that answer specific questions about the company.

Controlled third-party profiles: social accounts, business listings, and directory entries the company can claim but does not host.

Contributed and adversarial sources: review platforms, Reddit threads, Wikipedia, news coverage, and employee commentary.

Generated surfaces: autosuggest predictions, People Also Ask panels, and AI summaries that compress the rest into a paragraph.

Governing that page means influencing four categories of source with four different sets of tools. Companies that treat it as a website problem control one row of a page they do not otherwise touch.

Most branded search is not the brand name

The same Ahrefs study found that queries of three or more words account for the largest share of branded search and concluded that most branded search is not someone typing a company name and hitting enter. It is someone who already knows the brand, looking for something specific about it.

Ahrefs published two of its own failures to illustrate this. Users searched to find out whether the company ran an affiliate program, and because no page answered that question, Google assembled a result from whatever content happened to contain both words. Separately, users searching for recent product changes received an inconsistent set of results until a changelog was available to satisfy the query. Neither case involved a critic, a competitor, or a crisis. Both were unanswered questions, resulting in bad search results.

The modifier set is where reputation actually lives: reviews, complaints, scam, lawsuit, alternatives, layoffs, cancel subscription, and the platform-specific ones like Reddit and Glassdoor. A brand that has never audited its own modifier set does not know what its buyers are reading. Those same modifiers surface in autosuggest, which means the prediction and the result page reinforce each other.

AI is reaching branded queries later than everywhere else

The reflexive claim is that AI is consuming branded search. The data is more specific than that, and it moves fast enough to be worth dating precisely. Ahrefs analyzed 56 million AI Overviews and found that they appear less often than average on branded queries, as well as on local and short ones (Ahrefs, June 2025). Google has been slower to generate a summary for a company name than for an informational question.

That gap is closing quickly. Semrush analyzed more than 10 million keywords using clickstream data from Datos and found that the share of navigational queries triggering an AI Overview rose from 0.84% to 10.33% between October 2024 and October 2025, while commercial queries climbed from 8.15% to 18.57% over the same period (Semrush, December 2025). Semrush reads that trajectory as branded traffic becoming exposed. Overall coverage stayed volatile throughout, peaking near 24.61% of keywords in July 2025 before settling to 15.69% in November, so the direction matters more than any single month's figure.

One assumption in that same study deserves correcting, because it circulates in vendor decks unchallenged. Tracking identical keywords before and after an AI Overview appeared, Semrush found the zero-click rate fell from 33.75% to 31.53%. Summaries did not automatically cost those pages their clicks. Anyone selling AI panic by the click is selling something the measurement does not support.

Two features sit beside nearly every AI Overview: related searches at 95.32% and People Also Ask at 90.03%. Both are generated from aggregate query behavior, which makes them the same machinery as autosuggest rendered in a different position. Autosuggest predicts the modifier before the search runs. Related searches and People Also Ask display it afterward. A company name paired with a damaging modifier can therefore appear three times on a single page: once in the box on the way in, and twice more on the way out, whether or not the brand has ever answered the question. Discussion and forum blocks are prominent in the same SERPs, which is the structural reason community threads keep surfacing against company names.

Chat assistants are a different surface again. Someone asking an assistant whether a company is legitimate, or how a product compares to a competitor, is not running a Google query at all. That answer gets composed from sources the company may never have audited. Semrush analyzed more than 150,000 AI citations in June 2025 and found Reddit cited most often at 40.1%, ahead of Wikipedia at 26.3% and YouTube at 23.5%. The share has since proven unstable: a three-month follow-up recorded ChatGPT citing Reddit in close to 60% of prompt responses in early August 2025, then in close to 10% by mid-September (Semrush, October 2025). Treat the position as durable and the percentage as a snapshot.

The practical consequence is that the modifier set now has two audiences. AEO and GEO cover how those systems select and cite sources. The point for brand reputation management is narrower: the content that answers a trailing-modifier query well is the same content an assistant reaches for when asked the same question in conversation.

What the discipline is not

It is also not brand marketing wearing different clothes. Awareness work creates the searches. Brand reputation management determines what those searches find. A campaign that succeeds at generating demand while page one remains unmanaged has increased the number of people encountering an unmanaged result.

The Board's View

Reputation risk is discussed at the board level in the language of crisis: what happens when something goes wrong, who speaks, and how fast. Branded search sits upstream of that conversation and rarely reaches it. The asset either exists before a crisis or it does not, and it cannot be built during one. By the time suppression is on the agenda, the cheap window has closed.

The relevant board question is not whether the company monitors its reputation. It is whether anyone owns the branded SERP as a deliverable with a budget attached. In most organizations, the answer is that marketing owns the homepage, communications owns the press, support owns the help center, and no one owns the page that assembles all three. That is the structural gap the Chief Reputation Officer role exists to close.

Treated as an asset, branded search presence behaves like one. It compounds, it depreciates when neglected, and it is far cheaper to build than to repair.

Board Questions to Ask

• "Show me page one for our company name, and for our name plus reviews, complaints, and alternatives." The second set is the one that matters. A team that has only looked at the first has audited the easiest surface.
• "Which results on that page do we control, and which do we only influence?" A candidate who cannot separate the two categories will overpromise on both.
• "What questions are people asking about us that we have never answered on our own site?" Unanswered questions do not go away. They get answered by whoever publishes something.
• "What does an AI assistant say when asked whether we are worth buying from, and what does it cite?" The citations matter more than the answer, because the citations are what can be influenced.
• "What would it cost to displace the worst result on page one, and what would it have cost to prevent it?" The ratio between those two numbers is the argument for funding this work at all.

Related pillars: SEO and reputation, autosuggest, review management, and AEO and GEO.