Why Brand Reputation Matters More in the Age of AI Search
Industry and Competitive Context
For two decades, brand discovery online ran on a simple exchange. A consumer typed a query, a search engine returned ranked links, and brands competed to be clicked. Search marketing was built on that exchange, and visibility and traffic were treated as nearly the same thing.
That exchange is changing because search engines and AI assistants now generate answers rather than only list links. Google introduced AI-generated summaries in Search, called AI Overviews, in 2024. On Alphabet's second-quarter 2025 earnings call, CEO Sundar Pichai said AI Overviews had more than 2 billion monthly users, up from 1.5 billion in May 2025. He also said AI Mode, Google's more conversational search experience, had over 100 million monthly active users in the United States and India. On the third-quarter 2026 call, published on 28 October 2026, Pichai said the Gemini app had reached one billion users. He also said Google's year-over-year query growth was the highest in five years, and that AI Mode queries average about three times the length of traditional searches.
Independent measurement shows the scale of the behavioural change. The Pew Research Center analysed the browsing activity of 900 U.S. adults in March 2025. About 18% of the Google searches in its sample produced an AI summary. When a summary appeared, users clicked a traditional result link in 8% of visits, compared with 15% of visits without a summary. Only 1% of visits with a summary included a click on a link inside the summary. Users ended their browsing session after 26% of pages with a summary, against 16% of pages without one.
Google disputed these findings, and its position belongs in the record. A Google spokesperson told MediaPost that the study "uses a flawed methodology and skewed query-set that is not representative of Search traffic." Google also argued that the study did not account for overall traffic growth and counted only a limited set of links in the summary. The honest reading is that the click-through effect is documented by one independent study and contested by the platform itself.
Forecasts add to the uncertainty. In February 2024, Gartner predicted that traditional search engine volume would drop 25% by 2026 as AI chatbots and other virtual agents take share.
Early commercial data shows where consumer attention is heading. Adobe Analytics reported that traffic from generative AI sources to U.S. retail sites rose 1,300% year over year during the November to December 2024 holiday season. It also reported a 1,200% increase between July 2024 and February 2025. Adobe noted that this traffic remains modest compared with channels such as paid search and email. In an Adobe survey of more than 5,000 U.S. consumers, 39% said they had used generative AI for online shopping and 53% planned to do so that year. Research (55%) and product recommendations (47%) were the leading uses.

Brand Situation Prior to the Shift
Because this case concerns a category-wide change, the relevant "prior situation" is the operating model most brands relied on. Brands invested in being found through ranked results, in owning the click, and in directing visitors to their own web properties. In that model, a brand controlled much of the journey from query to landing page.
What can be documented is the condition of trust in the surrounding environment. The Reuters Institute's Digital News Report 2025 covered 48 markets and found that trust in news was stable at 40% for the third consecutive year. It also found that trusted news brands were the most frequently named first place people go to check whether something is true or false online, at 38%, ahead of official sources at 35%, fact-checkers at 25% and social media at 14%. Before AI search scaled, then, brand names were already one of the main tools consumers used to verify information.
Strategic Objective
No single company has published an objective for this shift, so none is attributed here. The case instead frames the objective that any brand faces: remaining accurately represented, and favourably selected, in answers that a machine composes on the consumer's behalf.
This differs from the old objective of ranking highly. In a link list, a brand needs to appear. In a generated answer, a brand needs to be included, described correctly, and trusted by a system that has chosen which sources to draw on. The unit of competition moves from the page to the claim, and from the click to the mention.
Campaign Architecture and Execution
What is documented is how the answer-generating systems themselves have behaved and been corrected, which shows how much of the execution sits outside brand control. In a post dated 30 May 2024, Google's head of Search, Elizabeth Reid, acknowledged that "some odd, inaccurate or unhelpful AI Overviews certainly did show up." Google said the problems came partly from nonsensical queries and satirical content, and partly from what it called "data voids," meaning topics where there is "a limited amount of high quality content." It also noted that forum content can sometimes give unhelpful advice. Google announced more than a dozen technical improvements, including limits on satire and humour content and on user-generated content in responses that could give misleading advice. It said it found policy violations on fewer than one in every 7 million unique queries where AI Overviews appeared.
For marketers, the strategic point in this episode is the "data void." Google itself named thin high-quality content as a condition that produces poor answers. That is a documented statement about how generated answers fail, and it implies that the quality and availability of authoritative material about a topic shapes what the system says.
Independent testing shows the scale of the accuracy problem across assistants. In October 2025, 22 public service media organisations from 18 countries, working in 14 languages and coordinated through the European Broadcasting Union and the BBC, published an evaluation of four AI assistants: ChatGPT, Copilot, Gemini and Perplexity. Professional journalists assessed more than 3,000 responses. They found that 45% contained at least one significant issue. Sourcing problems appeared in 31% of responses and accuracy problems in 20%.
Positioning and Consumer Insight
The documented consumer insight is an ambivalence about AI. People use it and are interested in it, but they do not fully trust what it produces. The Reuters Institute found that 7% of people across its markets used AI chatbots for news each week, rising to 15% among under-25s. Respondents expected AI to make news cheaper and more up to date, but also less transparent, less accurate and less trustworthy. The net expectation for trustworthiness was minus 18.
Adobe's survey points the same way from the shopper's side. Among respondents who had used generative AI for shopping, 92% said it enhanced their experience, and 87% said they were more likely to use it for larger or more complex purchases. Consumers value the convenience of AI-assisted research, yet the wider evidence shows they remain cautious about its reliability.
The analytical consequence is a tension that brand reputation can resolve. When a consumer cannot easily judge whether a machine-generated answer is correct, a recognised brand name becomes a shortcut for judging credibility. This interpretation is the author's, but it rests on the Reuters Institute finding that trusted brands are people's most common first stop for verification.
Reputation also operates upstream, on the machine's side. Muck Rack's May 2026 report analysed more than 25 million links cited by ChatGPT, Claude and Gemini. It found that about 27% of citations were journalistic and about 84% came from earned media. Non-paid sources accounted for about 99% of citations. Among journalism citations with known publication dates, 57% were published within the preceding 12 months. Muck Rack's report also shows that each assistant leans on different sources. Wikipedia and YouTube appear among top domains for ChatGPT and Gemini, while Gemini's list also includes Reddit and Quora. Pew's earlier work likewise found Wikipedia, YouTube and Reddit together accounted for 15% of the sources listed in Google AI summaries.
Two cautions apply. First, Muck Rack sells media-intelligence software, so its findings concern a field in which it has a commercial interest. Second, the report shows which kinds of sources are cited, not that a brand's reputation causes it to be cited. No verified public information is available establishing a causal link between a brand's reputation score or sentiment and the frequency or favourability of its mention in AI-generated answers.
Media and Channel Strategy
The documented channel evidence is about where AI systems draw their material. Earned media dominates the citations in the Muck Rack data. That matters because earned media, meaning coverage and mentions that a brand does not buy, is the channel a brand influences least directly and over the longest horizon. Muck Rack also reports that citations peak in the first month after publication and decline through about month six before forming a long tail. That pattern suggests freshness matters, though the report does not say how long a brand's coverage continues to influence answers.
Gartner offered related advice to marketers when it published its forecast. In the same release, it said marketing leaders whose brands rely on SEO should consider allocating resources to testing other channels. This is a research firm's recommendation, not evidence that any company has done so.
Paid search is not displaced in the documented record. Alphabet reported Search and other ads revenue growth of 12% year over year in the third quarter of 2026. Google has also said that Search is being built as "one seamless experience across AI Overviews, AI Mode, and traditional Search." Brands are therefore operating across both ranked links and generated answers at once, not choosing between them.
Business and Brand Outcomes
Only documented outcomes are reported here, and they describe the market rather than the results of a particular brand programme.
At the platform level, Alphabet reported the usage figures and the 12% Search and other ads growth described above. At the click level, Pew recorded the drop from 15% to 8% in traditional link clicks when a summary appears, a finding Google disputes. At the traffic level, Adobe recorded rapid growth in generative-AI referrals to retail sites from a small base. Adobe also reported that these visitors showed 8% higher engagement than visitors from other sources, viewed 12% more pages per visit and had a 23% lower bounce rate. Their conversion rate was 9% lower than that of other traffic sources, a gap that had narrowed from 43% in July 2024.
At the quality level, the EBU and BBC study found that 45% of the answers it assessed had a significant problem. At the trust level, the Reuters Institute found that consumers expect AI to reduce trustworthiness in news.
Strategic Implications
The following is analysis built on the documented evidence above, not additional fact.
Reputation becomes an input to the distribution system. Under the link-list model, reputation influenced whether a consumer chose to click. In generated answers, the evidence suggests reputation also influences what the system retrieves and cites, because assistants draw heavily on earned and journalistic sources. Brand building and search visibility, usually run by different teams, converge on the same asset: credible third-party coverage. The causal strength of that link remains undocumented.
Less traffic can coexist with more influence. If fewer visits reach brand websites, as Pew's data suggests for the sample studied, then a brand's message is increasingly encountered inside someone else's interface. Measurement built on clicks will then understate where persuasion is happening. At the same time, the high engagement Adobe found among AI-referred visitors suggests that the visitors who do arrive may be well-informed. The two findings do not yet resolve into a single conclusion about value.
Misrepresentation is now a managed risk. The EBU and BBC finding that 45% of tested answers had a significant issue, and Google's acknowledgement of data voids, show that a brand can be described inaccurately by a system it does not control. Reputation management extends from monitoring what people say to monitoring what machines say. The documented remedy implied by Google's own explanation is more high-quality, authoritative content on the topics where a brand has expertise, although no verified evidence shows how effective this is for any individual brand.
Trust is the scarce resource. Consumers use AI for its convenience but expect it to be less trustworthy, while they still turn to known brands to verify claims. A brand that is recognised and credible therefore holds an advantage at the moment of verification, whichever interface delivers the first answer. This conclusion reflects the Reuters Institute data on news brands and should not be assumed to hold equally in other categories.
Uncertainty argues for flexibility. The evidence is contested and still developing. Gartner's forecast is unconfirmed, Google disputes Pew's methodology, and Alphabet reports rising query growth. A sound strategy neither abandons search nor assumes its decline, and it invests in assets that work in both worlds: credibility, authoritative content and earned coverage.
Discussion Questions
Pew found that clicks on traditional results fell from 15% to 8% when an AI summary appeared, while Google called the study flawed and Alphabet reported record-level query growth. How should a CMO weigh conflicting evidence like this when deciding whether to cut or maintain search investment?
Muck Rack's data shows that most AI citations come from earned media, but it does not prove that reputation causes citation. What additional evidence would a brand need to justify shifting budget from paid to earned media, and how could it gather that evidence ethically and reliably?
The EBU and BBC study found significant issues in 45% of the AI answers assessed. If an AI assistant misdescribes your brand, who bears the reputational cost, and what practical and governance responses are available to a brand that does not control the system?
Adobe reported that AI-referred shoppers showed higher engagement but lower conversion than other traffic, with the gap narrowing over time. How should a firm value a channel whose traffic is small, fast-growing and different in quality, and which metrics would you use given that many internal metrics are not publicly available?
The Reuters Institute found that consumers go first to trusted news brands to verify information while expecting AI to be less trustworthy. Does this finding generalise to non-news categories such as retail, finance or healthcare? What would you need to know before treating brand trust as a competitive moat in an AI-mediated market?



Comments