• Writing
  • Work
  • Activity
  • Contact

Mike Garite

ok@mikeg.meResume
  • GitHub (opens in a new tab)
  • LinkedIn (opens in a new tab)
Writing/Product

How Ecommerce Is Changing

Online retail keeps taking share, but the interesting shift is operational: real-time decisions, AI in the payment path, and reverse logistics treated as a product surface rather than a cost center.

October 26, 20263 min read

Ecommerce is no longer the new channel. It is a large, still-growing share of retail that is being rebuilt around faster decisions and messier after-purchase work.

The size of the channel#

The U.S. Census Bureau's Quarterly Retail E-Commerce Sales report for the second quarter of 2026 put seasonally adjusted e-commerce sales at $340.2 billion. That was up 3.8% from the first quarter and 12.2% from the second quarter of 2025. E-commerce accounted for 17.1% of total retail sales. Total retail itself was about $1.99 trillion for the quarter.

Those figures are useful because they keep the conversation honest. Online is large enough that small percentage improvements in conversion, fraud, or returns move real money. It is also still a minority of retail, which means omnichannel operations remain the rule rather than a slogan.

From pilots to operating systems#

NRF's retail technology framing for Nexus 2026 described a shift from implementation theater to execution. Retailers already bought commerce platforms, data infrastructure, and omnichannel tools. The work now is making those investments operate together: pricing, inventory, fraud, service, and merchandising decisions made on current data rather than last week's summaries.

That matches what payments teams are saying. In Adyen's NRF 2026 pulse check, the recurring priorities were data, loyalty, fraud, and AI as connected levers. One concrete claim from Adyen's commercial leadership: AI applied to transaction routing and optimization has shown up to a 6% uplift in conversion in their experience. Whether any one vendor's figure generalizes, the direction is clear. AI is being judged on authorization rates, fraud loss, reconciliation, and conversion, not on demo charm.

Payments are becoming agent-aware#

The Merchant Risk Council's 2026 Global eCommerce Payments and Fraud Report, based on 1,278 merchant respondents, published a few markers worth keeping:

  • 43% of merchants now accept real-time payments.
  • 63% are exploring or plan to implement agentic AI payments.
  • 3.2% of total annual e-commerce revenue is lost to payment fraud globally.
  • 45% see real-time payment fraud as a coming major attack vector.

Shoppers are not the only actors in the funnel anymore. Agents that compare, repurchase, and check out will push merchants to decide what an automated buyer is allowed to do, how it authenticates, and how disputes are evidenced later.

Returns changed the product#

NRF and Happy Returns estimated that in 2025 retailers would see 15.8% of annual sales returned, or about $849.9 billion. The online return rate was 19.3%. Free returns mattered to 82% of consumers when choosing where to buy. Instant refund or exchange options mattered to 76%.

That is not only a logistics story. It is assortment, sizing, product data, and policy design. If one in five online purchases comes back, the post-purchase path is part of the product. NRF's reverse-logistics writing makes the same point: centralized return hubs, inspection, routing, and resale are becoming revenue-recovery systems, not only cost sinks.

What I would build for next#

If I were prioritizing an ecommerce roadmap from these signals, I would keep it short:

  1. One commerce data plane for inventory, price, customer, and order state that pricing and service can both trust.
  2. Payment optimization with measurable conversion and fraud outcomes, not a generic AI badge.
  3. Return and refund flows with the same UX care as checkout, because customers already judge the brand there.
  4. Evidence trails for disputes and agent-initiated purchases, before those volumes get large.

Ecommerce is changing less as a storefront problem and more as an operations and decision-systems problem. The winners will treat data quality, payments, and reverse logistics as product surfaces.

Tagged

  • Business
  • Product
  • Data
  • AI
← PreviousAgent UI Matters as Much as the SchemaNext →Completed Work, Not More Code

Related

  1. November 1, 2026·Product

    What to Do with Adjustments→

    Returns, refunds, chargebacks, and price adjustments are not edge cases. They are a large share of ecommerce economics. Here is a practical way to handle them without training customers to abuse the policy or punishing honest ones.

    Business · Product · Data · AI3 min read

  2. November 2, 2026·Product

    When Serving Agents Costs More Than They Earn→

    Grok Bot hit a unit-economics wall in early November. Usage limits tightened because serving costs outran revenue. The lesson is broader than one product: agent systems fail first as businesses, not as demos.

    Agents · AI · Product · Business3 min read

  3. October 30, 2026·Engineering

    Vendor Matching Is Where Due Diligence Starts Cascading→

    A weak vendor or customer match does not stay local. It multiplies through finance, risk, support, and AI agents. Entity resolution is not housekeeping. It is the control point for everything that follows.

    Data · Architecture · Business · Product3 min read