The Maze: Social media is no longer just where products get noticed. In BCG’s Global Consumer Radar, it is the most influential listed media interaction on reported purchase decisions at 90%, tied with internet search and ahead of brand or retailer websites at 89%. LLM search reaches 85%. The headline needs a footnote in bold: these are conditional influence measures among journeys where the touchpoint was part of the consumer journey. They do not mean 90% of all shoppers use social, or that an LLM response causes 85% of all purchases. They mean that once a consumer brings either touchpoint into the decision, it sits unusually close to the final choice.
Social has the edge, but LLM search is already in the same decision neighborhood. Across the full ranking, social media is 90%, LLM search 85%, online ads 84%, print ads 78%, TV/radio ads 77% and billboards 76%. The spread is narrower than a conventional “new versus old media” story suggests. What changes is the type of work done: social can make a product feel relevant, while an LLM can compress comparison, specifications, alternatives and objections into one interaction. Both operate when shoppers are trying to reduce uncertainty, which is why the right comparison is not just impressions or reach.
The two channels travel across categories rather than living in a single digital niche. Social/LLM influence is 93%/87% in beauty care, 92%/86% in food and beverages and 91%/85% for mobile services plans. The smallest visible gaps are household appliances at 90%/89% and travel at 89%/88%. Even automotive is 86%/85%. That pattern matters because it covers habitual categories, complex services and infrequent high-consideration purchases. Social is higher in every displayed category; LLM search is consistently close enough that treating it as an experimental side project is increasingly a choice to leave a comparison surface unmanaged.
Apparel reveals the largest displayed separation—and the actual operating tension. Social media reaches 88% influence in apparel, while LLM search reaches 81%, a seven-point difference. Fashion buying often requires desire, identity and visual proof alongside functional comparison. That is social’s natural advantage. But an 81% conditional influence score still makes LLM search part of the decision environment, especially when buyers ask about fit, alternatives, price, materials or retailer options. The job is therefore two-sided: make content emotionally legible in social contexts and factually legible in product data, reviews and retailer pages.
Presence is not the same thing as influence. BCG also reports that social media is used for information gathering in 22% of research-led journeys and GenAI tools in 13%. Those penetration figures should not be blended with Exhibit 3’s conditional influence percentages. A sensible channel plan needs both views: how many relevant buyers reach a touchpoint, and how much that touchpoint shapes choice when they do. Conflating the two is how teams celebrate a large “influence” number while missing a discovery or distribution problem upstream.
Why it matters: The practical response is not “post more” or “game the model.” It is to make useful content and product information travel together. Social needs clear creative proof, creators and timely answers. LLM-facing surfaces need accurate, structured and comprehensive facts. Both need a source of truth that commercial, content and product teams maintain together. In a research-led journey, a stale detail is no longer a website flaw. It is a quiet competitor recommendation.
Sources: BCG, *The Research-Led Consumer Journey Is Changing the Rules of Marketing Influence*, published June 17, 2026; BCG Global Consumer Radar survey, November 2025, N = 12,308.
