There is a version of marketing history where brands once held a direct line to the consumer. Television gave them thirty uninterrupted seconds in the family living room, radio occupied the car, and print sat on a kitchen table for days. The relationship was not equal, but it was structurally direct. What little stood in the way was largely controllable—the media buy, the creative and messaging, and the placement.
That directness began eroding when consumption moved to screens, and it has not stopped. Cable, now streaming, fragmented the living room, the internet transformed consumer attention from a predictable commodity into active competition, and social infrastructure divided what remained across an infinite number of sources where brands are, at best, peripheral. As tech iterated, it installed more and more layers of mediation between the brand and its buyers. The advertising industry’s consistent response has been to update tactics and tools, to reallocate budgets and proceed with the assumption that relationship with brand and buyer was still, at a fundamental level, intact.
In reality, modern marketing has simply been renting proximity from a rotating cast of third-party platforms that control the point of access. That is very different from having a relationship.
The Socialization of Trust
When creator culture became a marketing strategy, the typical framing was that an influencer could provide authenticity to brands who could not easily produce it themselves. The narrative was about storytelling, nativeness and trust. What was actually happening is that brands had already lost their direct access to buyers and were paying people who had it to carry a product into that relationship.
Research data from Edelman’s annual Trust Barometer highlights a consistent behavioral pattern that purchase intent is seldom driven by an isolated influencer endorsement. Instead, it is the highly visible community response to the content—comments, liking, sharing—outperformed in driving downstream behavior. The creator acts as an introduction while the community’s accumulated reaction to the content is what actually moves people.
Generative AI has absorbed this same social logic and has operationalized it at scale, making the creator economy look like a beta test.
Architecture of Generative Consensus
When someone queries an AI tool regarding a product category or potential brand choice, the model does not immediately go to fetch different brands’ websites for content. It synthesizes an immense corpus of open web data from editorial coverage, historical forum discussions, rating and review platforms, community commentary, affiliate content, etc, and produces a type of cohesive, cocksure narrative. Essentially, ~80% of all AI-generated products are sourced from assets outside what is created, owned or managed by a brand.
People will trust an output not because they can trace its sources for verifiability, but because of the linguistic projection of consensus. What AI delivers feels more like a definitive conclusion an informed group would arrive at after doing online research. It is not unlike sorting a Reddit thread by “Best” and trusting the top result.
A study published in the PNAS (Jakesch et al.) demonstrated that AI-generated text is consistently rated as more credible than human-written text under identical conditions. It sounds certain, and in a particular way. This stems from things such as a lack of hedging, structural uncertainty and a unique type of personal register, not the accuracy of data it sources from. The absence of doubt is interpreted by users as confidence and without perspective, it is interpreted as objective. This triggers automation bias—the tendency to defer to automated systems even when reasons exist to question them. It also gives people an illusion of competence, but many others have covered this phenomenon.
This dynamic gets compounded by the personalization layer in these tools. Models increasingly calibrate their responses to ongoing conversation context, stated user preferences, and inferred demographic characteristics. The consensus a model delivers is individually assembled, making the hyper-relevance of it feel trustworthy, while making it nearly impossible for a brand to audit what is actually being said about them across the user base.
The sources cited in answers and outputs are also weighted by the volume of community reaction and engagement to what has been said. Social teams know this as sentiment, PR people know this as reputation. A critical product review from 2019 does not exist in isolation within a model training environment. The model ingests it alongside every other reply, secondary article that referenced it, the forum threads that posted it (truly, the internet is forever). Where legacy search retrieved independent web pages, generative engines are delivering something closer to a summary of opinions of the original content.
Structurally Relocating Trust
There is a pattern in how consumer trust relocates and the advertising industry has been slow to articulate this relocation, largely because doing so would require acknowledging the limitations of paid media and what marketing is actually capable of controlling.
Brands are not losing reach or top-of-funnel awareness. Both remain highly purchasable commodities. What is dissolving is the consumer’s perception of the brand as an authoritative and objective participation in the consumer’s own decision making process. Their trust has progressively migrated to parasocial creator relationships, closed online communities, and collected opinions of strangers and peers who document their experiences in searchable, permanent and publicly visible forms. Brands are spoken about, not to, and characterized by the weight of external opinion rather than shaped by a brand’s own messaging.
The broadcast era operated on a different psychological framework. Robert Zajonc’s 1968 research on mere exposure effect established that repeated exposure to stimulus increases positive evaluation independent of any other variable. Familiarity reliably manufactured preference and this was used, quite successfully, to build bThe indsutrurand relationships.
Screens, especially the smaller personal ones we can’t get enough of, redistributed that. The parasocial relationship migrated from television personalities to digital media then intensified as social platforms removed the distance between content creator and consumer. A creator interacting within comment sections and publishing real-life relatable moments engineered a felt closeness that repeated ad exposure of any media spend could not match nor replicate. The creator can be real in ways the brand never quite can.
Sequential Feedback Loop
The creator economy widened the gap between brand and buyer, and the industry mostly chose to treat creators as a novel media placement. As consumers transferred trust to individual creators, brands purchased access to those relationships and agencies were more than happy to leverage creators as a tactical workaround for establishing a brand’s standing.
Social platforms are staging the same dynamic on a separate layer. TikTok and Instagram now function as a dominant space for discovery at scale, where people encounter new products through filtered experiences of someone whose taste they appreciate or align to. In this representation of genuine consumer proximity, the brand is not the trusted source but a passive passenger.
A deeper, less explored challenge is how social platforms connect to an AI citation environment. Reddit accounts for roughly 44% of all citations in AI overviews, and YouTube’s current 32% is growing (note: each platform prioritizes these differently). The commentary and conversations happening on these platforms are the material models train on in effort to formulate various characterizations of categories, topics and brands.
A TikTok that drives initial consumer discovery and a Reddit thread that shapes an AI overview six months later about whether a product is trustworthy are not separate marketing moments. They are consecutive points in a single consumer decision, each governed by different trust conditions that require a different brand presence. A high-performing creator post can generate discovery, but it does not naturally convert into the permanent and indexable text that shapes a model’s downstream characterization. But, the sentiment and conversations on that post absolutely do.
Reputation As An Input Layer
Brand strategy treats brand reputation as a measurement category that serves as a lagging performance indicator. You monitor sentiment index trends, track it over time, and flag anomalies to appropriate communication teams. This process places reputation as being a downstream byproduct of brand activity.
In generative search, this flows in reverse. The compiled body of public opinion, aka sentiment, about a brand—what has been written, where it lives, how much community engagement it generated, and how long ago—is what determines whether a brand achieves visibility within a generative answer. Reputation is not downstream of marketing; it sits upstream of visibility, dictating discoverability. A brand pairing strong paid media budgets with poor earned strategy is now operating with a structural disconnect.
This reframes what PR, communications and community management mandates are. Previously, they operated defensively to protect brand equity. In a generative search environment, they become instrumental in building a brand’s foundation for a visibility strategy through creating and nurturing verifiable, third-party opinion and sentiment.
Most agencies and brands are not organized to deliver against this new reality. The teams, briefs, metrics and account relationships are often separate with PR often sitting outside of the agency of record relationship. Paid media is no longer the reliable way to have proximity to buyers or to build a brand’s characterization.
Our New Intermediary
Brands began to shed their consumers to screens first, then surrendered more to creators, then to peer communities, and now to large language models that are hungry and eating entire historical footprints who are decreeing who a brand is to the consumer.
At each juncture, the industry updated tactics, persevering on the belief that the right message, delivered at the right time, would break through to the right consumer. This reliable path is not wrong, but it is insufficient for the portion of the purchase journey occurring well beyond what any brand publishes or controls.
Brands that navigate this well will treat their collective third-party digital footprint as a foundational strategic asset rather than an insular communications output. This is not a mandate for a set of GEO optimization tactics, but recognition that the consumer has been forming that opinion in environments with no brand controls.
Generative search is not disrupting the brand’s relationship to buyers but highlighting that the proximity that brand’s thought they had to them isn’t real.