Streaming & Video
Why AI shoppers convert better but almost nobody uses them yet
AI referral traffic is tiny for retailers like Etsy, but it buys more per visit. That gap tells you where the money is, and where the risk hides.
Why it matters
For merchants and marketing teams, the signal here is that AI-referred shoppers are worth more per head than average, even though there are very few of them. That means the early economics favor building for these buyers now, while acquisition through AI platforms is still cheap and uncrowded. Anyone budgeting ad spend or SEO for 2026 has to decide whether to chase a channel that is under 1% of traffic but converts at a premium.
What changes next
Watch whether that sub-1% share of AI-referred traffic grows across two or three more earnings quarters, and whether conversion stays above baseline as volume rises. If both hold, expect retailers to start disclosing AI channel metrics separately, and to test giving agent platforms direct feed access. If conversion falls as volume climbs, the premium was just early-adopter bias.
If you run a store online, the number worth staring at this quarter is not how many people came from an AI tool, but how much each of them spent. According to PYMNTS' reporting on second-quarter earnings calls, Etsy CEO Kruti Patel Goyal told analysts that traffic arriving from AI agent platforms sits under 1% of the site's total, yet those visitors convert at a higher rate and place larger orders. That combination, a tiny channel with unusually good economics, is more useful to a merchant than a headline about AI adoption ever could be.
A high-value channel that barely exists yet
Two things are true at once here, and both matter. The volume is trivial: under 1% of visits means AI referrals are not moving anyone's revenue line today. But the per-visit value is not trivial, and that is the tell. When a channel that small already outperforms the average on conversion and basket size, it usually means the people using it arrive with clearer intent. They have described what they want to an assistant, the assistant has done some filtering, and the shopper lands on a product page closer to buying than a casual browser from search or social.
For a merchant, that changes the math on where to spend attention. Acquiring these buyers is cheap right now because almost nobody is competing for them. The playbooks, feed formats and structured product data that let an AI agent recommend your listing accurately are not yet standard practice. Getting them right early is the kind of advantage that gets more expensive to win the moment the channel scales.
The catch is that quality may not survive volume
The honest reading has a real caveat: high conversion in a tiny sample often reflects who the early users are, not what the channel does. The people routing purchases through AI agents in early 2026 are enthusiasts and power users, exactly the sort who buy decisively. As these tools reach ordinary shoppers, the average intent behind an AI referral will drift toward the mean. The premium could shrink or vanish.
There is also a control problem. When a shopper comes through search, the retailer owns most of the relationship. When an AI agent sits between the buyer and the store, the platform controls discovery, ranking and increasingly the checkout flow. A channel that converts well but that you do not control is a channel that can reprice access to you later. That risk is why the low volume today is not purely bad news. It gives merchants time to build presence on these platforms before the terms harden.
What a merchant should actually do with this
The practical move is to treat AI referrals as a channel worth measuring separately, even at under 1%. Tag the traffic, track its conversion and basket size on your own, and watch the trend rather than the absolute number. If the premium holds as volume rises over the next few quarters, invest in clean structured product data and accurate feeds so agents represent your catalog well. If the premium erodes as the channel broadens, you will have learned that early, and cheaply, instead of overbuilding for a buyer who turned out to be ordinary all along.
Source: PYMNTS.com