RelayMag
ReportNo. 56

Where Brand Perception Actually Forms

RelayMagJune 20266 min read
Key takeaways

Most of what people believe about a brand gets decided in places the brand has no control over. Marketers know this in the abstract, yet the budget and the attention keep flowing to the owned surfaces, the website, the paid ads, the carefully scheduled social posts. Those surfaces are easy to measure and easy to change, so they feel like where the work is. The harder truth is that a buyer forms most of an opinion before they ever land on a brand's homepage, and they form it on reviews, in communities, in conversations, and now inside AI answers that pull from all three.

The pattern is consistent across channels. The sources a buyer trusts most are usually the ones a brand can shape least. A company's own messaging gets discounted precisely because it is self-interested. Everyone knows the brand is rooting for itself, so the brand's claims carry less weight than a stranger's offhand comment. That asymmetry is the whole problem in one sentence. The most persuasive evidence about you is written by people you cannot direct.

Online reviews and ratings

Reviews are the closest thing to a public verdict, and they sit right at the point of decision. Consumer survey after consumer survey finds that the large majority of shoppers read reviews before buying, and that a strong rating can move a purchase about as much as a personal recommendation. The intent is high, the trust is high, and the content is visible to everyone including search engines.

The limits are real, though. Reviews skew toward the extremes, because people who are delighted or furious are the ones motivated to write. The quiet middle stays quiet. Ratings are also gameable through incentives, selective solicitation, and outright fakes, which is why platforms keep tightening their fraud rules and why a perfect score can read as suspicious. A brand can ask for reviews and respond to them, but it cannot author them, and that is exactly why they work.

Reddit, niche communities, and forums

When someone wants the unvarnished version, they increasingly go to a community rather than a brand. Reddit, specialist forums, Discord servers, and subject-specific groups are where people compare notes without a sales motive in the room. The tone is skeptical, the participants often know the category cold, and a thread can surface the failure modes a marketing page will never mention.

Two things have raised the stakes here. First, this content ranks. People now append a community name to their searches to escape promotional results, and search engines have responded by surfacing those threads more prominently. Second, AI answer systems lean on the same discussions, so a single well-argued comment can echo into thousands of synthesized responses. The downside is representativeness. A loud subforum is not the whole market, and a brand that overreacts to one heated thread can misread the room. Communities also resist anything that smells like marketing, so a brand's options are mostly to show up honestly, fix what gets raised, and otherwise stay out of the way.

Social media

Social platforms give reach and speed that nothing else matches. A complaint or a moment of praise can travel far in hours, and the public nature of it means the audience is watching how a brand responds as much as what was said. For awareness and for catching problems early, social is hard to beat.

It is also noisy and unrepresentative. The people posting are not a clean sample of buyers, algorithms amplify whatever provokes a reaction, and sentiment tools struggle with sarcasm, memes, and context. A viral spike can look like a crisis and amount to nothing, or look like nothing and quietly harden into a reputation. Social tells you what is moving right now. It is a weaker guide to what most customers actually think.

Word of mouth and private channels

The most trusted source of all is a person you know telling you something directly. Research on word of mouth consistently finds it influences a large share of purchase decisions, and that a recommendation from a friend or family member outweighs almost any form of advertising. People act on it because there is no angle behind it.

The catch is that almost none of this is visible. It happens in text threads, group chats, direct messages, phone calls, and hallway conversations, the so-called dark social that no dashboard can see. A brand can earn it by being good and by giving people something worth repeating, but it cannot track it, cannot target it, and cannot prove it moved a sale. The channel with the most influence is the one with the least measurement, which is part of why it gets underfunded. It does not show up neatly in a report.

Earned media and press

Coverage from a credible outlet works slowly, but it lends a kind of authority the other channels do not. A review in a respected publication, a journalist's analysis, an inclusion in a roundup, these carry the weight of an outside party staking its own name on a judgment. That borrowed credibility is the point.

It is slow and uncertain by nature. You can pitch, but you cannot dictate the angle, the timing, or the verdict, and a single critical piece can stick for years because it ranks and gets cited long after publication. Earned media also compounds with everything else. AI systems and search results treat reputable coverage as a strong signal, so one solid article keeps paying out quietly in places the brand never sees.

AI-generated answers

The newest layer does not create opinion so much as compress it. When a buyer asks an assistant whether a product is any good, the answer is a synthesis stitched together from reviews, community threads, press, and whatever else the system has read. The buyer gets a confident paragraph instead of ten blue links, and that paragraph often arrives before they have formed any view of their own.

This is where the earlier points converge. Because these systems read the unowned channels and repeat them, perception now compounds. A pattern of complaints in a forum or a run of mediocre reviews does not just sit there anymore. It gets absorbed, summarized, and handed to the next buyer as a tidy verdict, sometimes with the rough edges sanded off and sometimes with an error baked in. The brand is frequently the last to know what the synthesized impression even says, and correcting it means changing the underlying sources rather than the answer itself. Treat this as a channel, not as a separate optimization project. It mostly reflects the health of everything upstream.

What this means for where the work goes

The uncomfortable read is that the channels a brand can fully control are the ones buyers trust least, and the channels buyers trust most are the ones a brand can only influence. Polishing the owned surfaces still matters, but it has a ceiling, because a buyer already discounts whatever the brand says about itself.

The more durable work is earning perception on ground you do not own. That looks like giving customers genuine reasons to leave honest reviews, showing up straight in the communities where your category gets discussed, fixing the problems that keep surfacing instead of arguing with them, and earning coverage that an outside party will vouch for. None of it is as tidy as a campaign you can schedule, and most of it resists clean attribution. It is also where the opinion is actually being formed, and increasingly where the machines doing the summarizing are looking.

Sources
R
RelayMag is an independent publication on marketing, search, and how companies get found.