The Maze: The most trusted recommendation engine in fashion and sports is not celebrity. It is proximity. In a BoF Insights EU5 consumer survey, friends and family sit at 73%, customer reviews at 68%, and brand-owned content at 40%. Professional athletes lead the public-figure pack at 29%. Macro-influencers and celebrities both land at 17%.
Reach is a distribution metric; trust is a conversion input. A 56-point gap separates friends and family from macro-influencers. That does not mean a brand can buy word of mouth. It means the brand has to earn it: dependable product, credible service, and a customer experience worth repeating. The 68% score for reviews and ratings makes the commercial mechanism plain. The customer is not looking for the loudest recommendation. They are looking for one that feels costly to fake.
Athletes beat celebrity reach because relevance beats familiarity. Professional athletes or sportspeople score 29%, twelve points above both celebrities and macro-influencers. For fashion and sports products, performance can make an endorsement feel closer to product use than a general lifestyle appearance. That is still a survey-reported trust ranking, not proof that an athlete partnership produces a better return on investment. But it is a sharp warning against buying audience size as if it were persuasion.
Micro-influencers have a narrow but real advantage. They reach 19%, two points above the two broad-reach categories. The report footer defines them as smaller creators with niche audiences. That small gap matters because creator cost, audience fit and creative credibility do not move together. A lower-cost creator with a specific reason to recommend a product can be more useful than a larger account whose relationship to the category is decorative.
Brand content still has a job—just not the leading role. At 40%, brand-owned content outranks every individual endorser in the exhibit. The implication is not “publish more.” It is to build content that helps a shopper judge the product: fit, use case, proof, comparison and service terms. Useful owned content supports the recommendation path. Polished campaign material that avoids the hard questions does not.
AI is already in the middle of the trust ladder. AI-powered recommendations score 22%, above micro-influencers and below athletes. The visible material does not define the product types or question wording behind that number, so it cannot establish whether shoppers trust an AI assistant, a retailer recommendation tool, or both. Still, 22% is enough to make the operating question practical: if an AI layer recommends a product, what evidence will it surface—reviews, product detail, creator proof, or paid brand copy?
Why it matters: The first money in a recommendation plan should go to evidence that compounds: product quality, review volume, clear owned content and credible category fit. Athlete and creator partnerships can amplify that base. They cannot replace it. The visible survey result applies to fashion and sports-product recommendations in an EU5 study, and it does not reveal sample design, buying behaviour or campaign economics. Treat it as a direction for testing, not a media-buying formula.
