Q2 2026 e-commerce hit $340.2B as AI assistants move toward placing orders

U.S. retail e-commerce sales totaled $340.2 billion in Q2 2026, up 3.8% from Q1 and up 12.2% from Q2 2025, according to the U.S. Census Bureau. Seasonally adjusted, e-commerce represented 17.1% of total retail sales for the quarter.

That’s the macro line. The operator’s line is sharper: AI assistants are starting to behave like a new checkout surface, not a novelty layer on top of search. And some sellers are now putting numbers on it.

The new metric that’s showing up in earnings: assistant-driven revenue

Digital Commerce 360 reported Aug. 28 that Williams-Sonoma said revenue attributed to its AI assistant rose 620% in Q2. The key point is not the size of the increase but what it implies about measurement: a large seller is already tagging, attributing, and reporting assistant-influenced sales in a way other operators can replicate internally.

Once that attribution exists, commerce teams can stop debating “AI strategy” in the abstract and start managing an assistant like any other channel. The hard work shifts to funnel instrumentation, offer governance, and the downstream systems that decide whether the assistant’s promise is true.

In housing and manufacturing, AI is being wired into product selection workflows

Digital Commerce 360 also reported Aug. 26 that Builders FirstSource unveiled an AI platform partnership with Digs. In building products, where specs, options, and substitutions can slow purchases, an assistant-led process is intended to move faster from a project need to an accurate order.

In manufacturing procurement, Digital Commerce 360 reported Aug. 25 that Xometry posted record Q2 revenue and rolled out new AI tools. Xometry’s marketplace relies on matching a buyer’s requirements with a supplier that can deliver on schedule. In that context, the focus of AI is operational, including quoting, finding the right sourcing fit, and reducing cycle time, rather than only generating marketing content.

What the macro data suggests: ecommerce is still growing faster than retail

Census data shows e-commerce outpacing total retail growth in Q2 2026: e-commerce rose 12.2% year over year while total retail sales increased 6.7% over the same period, according to the U.S. Census Bureau. That gap is where teams tend to test new buying interfaces as more volume shifts online and conversion remains a priority.

Why this lands on operations teams: agent UX is only as good as the systems of record

Assistants that recommend products, propose substitutes, or guide a buyer through configuration inevitably pull on the same operational truths that already break ecommerce programs: incomplete attributes, stale inventory, inconsistent lead times, and price and entitlement rules that live in too many places.

If the assistant is allowed to “act”, not just “answer”, governance becomes practical. What approvals does it respect. What substitutions can it make. When it’s wrong, where does the exception land, customer service, inside sales, the branch, the supplier. Those aren’t AI questions. They’re order-management and policy questions wearing an AI mask.

Removed: The referenced SHRM page is paywalled and does not provide usable support for employment-outlook or HR budgeting claims, so no SHRM-based budgeting statement is included here.

How this fits into 2027 planning: treat the assistant as a channel and strengthen the plumbing

• Stand up analytics for assistant-led activity now, including sessions, carts created, orders placed, substitutions suggested, and escalations to humans. Williams-Sonoma’s 620% figure indicates this kind of tagging and attribution is already happening (Digital Commerce 360).
• Expand scope to inventory and promise accuracy. If an assistant will suggest alternates, enforce rules based on real-time availability and approved equivalents rather than one-off similarity.
• Define the policies the agent must follow: contract pricing, approvals, shipping methods, restricted items, and returns. Document what happens when the agent cannot comply and where the handoff goes in the OMS or CRM.
• For marketplaces or distributed supply networks, stress-test quoting and lead-time logic. Xometry’s AI tooling rollout highlights how quickly assistant interactions turn into sourcing and capacity decisions (Digital Commerce 360).