The Maze: Holiday shopping discovery is spreading across more doors, but the evidence does not show shoppers abandoning one for another. In Bain’s U.S. holiday surveys, Amazon remained the most named starting point in August 2026, at 71%. Retailer and brand websites climbed to 60%; generative-AI platforms reached 24%. The strategic signal is a wider set of places retailers must be ready to meet customers. The statistical one is narrower: separate samples, self-reported intent, and answers that can overlap.
The non-Amazon doors are gaining ground. The share naming a retailer or brand website as a place to start rose from 51% in October 2025 to 60% in August 2026, a 9-percentage-point increase. GenAI platforms rose from 17% to 24%, up 7 points. Both gains are large enough to matter to channel planning. They do not tell us whether shoppers started with a brand, Amazon, or an AI assistant on any individual trip. These are stated entry points, not clickstream paths or attributed orders. The nine-point rise is a change in the share naming an owned destination, not a nine-point gain in retailer traffic. Teams should treat it as a reason to inspect direct and referral discovery, then test whether those visits convert.
Amazon still sets the pace. It was named by 71% of respondents in August 2026, versus 60% for retailer or brand websites and 24% for GenAI platforms. Its figure was 73% in October 2025, a 2-point dip. So the current evidence describes a leader with a slight decline alongside alternatives with stronger reported momentum. Calling this a handoff would put a sharper edge on the data than it can support. Amazon remains the largest starting point in the comparison. Its two-point movement is small beside its 71% level, and the data provide no uncertainty range or significance test. The clean takeaway is rank and reported direction, not a claim that Amazon’s position is collapsing.
The bars can overlap. Bain says respondents could select multiple answers. A shopper could name Amazon, a retailer’s site and an AI platform in the same survey. The percentages are not pieces of a 100% pie, and the change in one channel does not identify the origin of growth in another. Bain used different survey samples: 950 U.S. consumers in October 2025 and 1,105 in August 2026, in partnership with ROI Rocket. These are two snapshots, not a matched panel tracking the same people.
“AI” is a broad door, not a winner’s podium. The GenAI category names ChatGPT, Claude and Google Gemini as examples, but gives no separate share for each. It also does not fold retailer-owned assistants into the three-bar comparison. Bain discusses a separate measure: 13% of online shoppers expected to use a retailer’s own AI shopping agent. That measure belongs in a different conversation; combining it with the 24% would mix categories and risk double counting.
Why it matters: Retailers have more discovery surfaces to earn. Owned sites gained nine points in stated starting intent; GenAI gained seven. Amazon still led by 11 points over retailer/brand sites and 47 over GenAI in August 2026. Operators need marketplace reach and useful owned destinations. For AI, track brand visibility, product facts in answers, qualified referrals and conversion. Bain measures intent, not those outcomes. Entry-point behavior will show whether discovery shifts create demand. Until then, this is a planning signal, not proof of channel substitution.
Sources: Bain & Company, 2026 Holiday Shopping Outlook; Bain’s holiday discovery exhibit.
