Online shopping changing general merchandise retail
This research explores how general merchandise retail is changing due to online shopping. It will examine shifts in shopping behavior, retail operations, and competitive dynamics driven by e-commerce.
Last update Jun 5, 2026, 1:01 PM EST
Intelligence Brief
The current state and what matters now
Actors
The field is still led by Amazon, Walmart, Target, Costco, and a long tail of marketplace sellers and private-label operators. The actor set is widening further: AI shopping assistants, agentic-commerce protocol vendors, retail media networks, commerce infrastructure providers, and membership ecosystems now shape discovery and conversion. Signals suggest the interface layer is becoming more contested and more operational:
- Google is pushing Universal Cart and shopping agents across Search, Gemini, YouTube, and Gmail, making shopping a cross-surface layer rather than a single checkout flow.
- Amazon is moving from assistant-led shopping toward delegated execution and cross-merchant routing, while also licensing shopping AI to other retailers.
- Walmart is using marketplace assortment, unified app experiences, and 30-minute delivery to extend general merchandise online.
- Target is leaning on membership, same-day delivery, creator commerce, timed drops, and marketplace/media monetization as an integrated operating layer.
- Best Buy, Kohl’s, and other general-merchandise players continue to lean on marketplace and ads as platform-style revenue lines.
Moves
- Delegated shopping is moving from novelty to utility: assistants can recommend products, compare options, track prices, auto-add to cart, and increasingly complete purchases.
- Cross-merchant routing is becoming explicit, with AI surfaces surfacing products outside native catalogs and sending shoppers to merchant sites or assisted checkout flows.
- Universal carting is emerging as a shared layer, with Google’s Universal Cart linking items across Search, Gemini, YouTube, and Gmail before checkout.
- Agent-led discovery is accelerating as retailers test natural-language search, conversational assistants, and AI-guided decision support.
- Structured product feeds are becoming core infrastructure for AI shopping because assistants depend on clean metadata, pricing, and inventory.
- Marketplace expansion is absorbing more general merchandise assortment, especially where long-tail selection and seller depth matter.
- Marketplace monetization is widening beyond take rate into ads and platform revenue, especially where core demand is uneven.
- Store fulfillment is expanding, with stores increasingly serving shipped online orders, same-day delivery, pickup, and returns demand.
- Rapid delivery is extending into electronics, household supplies, pet care, and other general merchandise, not just food and convenience.
- Mission-based baskets are becoming more visible, with shoppers combining essentials and discretionary items in the same digital trip.
- Timed demand shaping is emerging through platform-led events, seasonal pull-forward, and scarcity-style drops.
Leverage
Advantage now comes from controlling the full commerce loop: discovery, trust, assortment, fulfillment, and monetization. The strongest players combine traffic, first-party data, inventory density, and delivery reliability. Physical stores still matter when they reduce last-mile cost, support returns, and improve immediacy. The newest leverage point is AI-mediated intent capture: whoever influences the assistant, ranking, or recommendation layer can shape demand before the shopper reaches a retailer’s website. Control over product feeds, identity linking, catalog quality, and inventory accuracy is now a source of bargaining power. Retailers that expose clean inventory and pricing data across channels gain an edge because digital convenience is now a core battleground even in value-oriented general merchandise. Cross-service identity graphs, such as linking shopping to media behavior, are becoming a new personalization asset. A further advantage is emerging in interface control: retailers that can separate discovery from checkout without losing the customer can monetize routing, not just transactions. Signals also suggest that being visible inside AI search and shopping surfaces is becoming a new form of shelf space.
Constraints
- Thin margins still limit how much price competition and free shipping can be absorbed.
- Fulfillment costs remain structurally high for bulky, low-value, or high-return general merchandise.
- Product-data quality is now a hard constraint: if catalogs, attributes, pricing, or availability data are wrong, AI search and marketplace conversion degrade quickly.
- AI adoption is outpacing execution, creating a gap between strategic intent and operational readiness.
- Consumers appear more willing to use AI for shopping help than for letting AI decide, so delegated buying still faces trust friction.
- Merchant defenses against bots remain a friction point, because retailers must distinguish helpful agents from malicious automation.
- Marketplace abuse is rising, including counterfeit listings, scam goods, and unauthorized sellers.
- Attribution conflicts are intensifying as media teams, creators, merchants, and store operators optimize for different outcomes.
- Checkout fragmentation persists: shoppers may discover products in one interface, compare in another, and complete payment on a different surface.
- Platform control disputes are rising as retailers block or limit shopping bots that threaten traffic, pricing power, or customer ownership.
- Demand orchestration risk is increasing as timed drops and event-led promotions can amplify stockouts, frustration, and uneven conversion.
Success Metrics
Success is increasingly measured by profitable digital penetration, not just online sales growth. Key metrics include gross margin after fulfillment, repeat purchase rate, order frequency, basket size, conversion rate, and customer lifetime value. Retailers also track on-time delivery, pickup adoption, return rates, inventory turns, and retail media revenue. In the AI era, new metrics matter too: assisted conversion rate, recommendation accuracy, search-to-purchase time, the share of traffic influenced by agents, the percentage of orders completed through conversational or delegated surfaces, and the share of purchases triggered by price alerts or auto-buy rules. For membership-led commerce, retailers are watching subscription attach rate, partner usage, and retention lift. For marketplaces, seller quality, take rate, API integration depth, and trust signals remain central. For platform-heavy retailers, ad revenue and marketplace mix are becoming important offsets when consumer demand is choppy. A newer signal is whether AI-driven traffic converts better than traditional referral traffic. Retail media may also be measured less by impressions and more by cost per recommendation or cost per outcome.
Underlying Shift
The deeper shift is from a store-centric distribution model to a data- and AI-orchestrated commerce system. Online shopping has already changed general merchandise retail by making assortment, pricing, logistics, and media continuously adjustable. The new phase goes further: shopping is becoming mediated by assistants, recommendation engines, membership ecosystems, creator networks, and unified operating layers that blur the line between browsing, buying, and fulfillment. The retailer is less a shelf owner and more a platform operator coordinating demand across digital interfaces, stores, and delivery networks. In this model, the store is a node, the app is a control surface, and AI is becoming the front door and, increasingly, the checkout layer. Commerce is also becoming more interoperable, with merchant data and transaction flows exposed to agents through shared protocols rather than only proprietary retailer stacks. The interface is shifting from keyword search to intent interpretation, and now also from direct purchase to routed purchase and delegated execution. The latest signals suggest the system is moving from AI-assisted shopping toward AI-mediated commerce operations, with product data, agent compatibility, and demand timing becoming part of the operating core.
Current Phase
The market is in a late adoption, early transformation phase. Omnichannel is no longer novel; it is table stakes. The next competitive wave is about who can operationalize unified commerce, agentic shopping, and marketplace-led assortment without destroying margin or trust. The profit pool is still contested among retailers, marketplaces, brands, creators, and retail media businesses, but the battle is shifting toward who owns the customer interface, the AI layer, and the transaction rules. The winners will be those that turn complexity into a simpler customer experience while also reducing internal friction and improving inventory truth. The evidence now suggests online shopping is not just supplementing general merchandise retail; it is increasingly setting the operating logic for it. The newest phase is a contest over who owns the first question, not just the final click. Attention appears to be shifting from whether AI matters to where AI sits in the commerce stack, from whether marketplaces help to how they become the default growth engine, and from seasonal promotion to platform-controlled demand timing.
What to Watch
- Agentic and conversational commerce adoption, especially whether AI assistants become a meaningful source of traffic and conversion.
- Universal cart and cross-surface checkout, especially whether carts can persist across search, video, email, chat, and marketplace portals.
- Merchant acceptance of AI checkout, including whether retailers monetize assistant traffic or keep blocking it as bot risk.
- Marketplace enforcement, especially whether fraud and counterfeit controls become a standard operating layer.
- Auto-buy and price-alert usage, including whether shoppers trust delegated purchase execution for repeatable categories.
- Marketplace operating models, especially whether marketplace becomes a core assortment and growth system rather than a side channel.
- Unified commerce execution, especially whether retailers can truly collapse channel silos for customers and operations.
- Retail media integration, including whether on-site, in-store, and offsite media become one measurable system.
- Inventory and catalog accuracy, since AI ordering and availability tools only work if product data stays clean.
- AI visibility and feed quality, including whether structured product data becomes a gating factor for demand capture.
- Demand timing control, including whether retailers increasingly front-load, bundle, or drop merchandise through platform-led events.
What's new
Latest brief updates
What’s new: Signals have strengthened around AI-mediated shopping becoming operational rather than experimental: Google’s Universal Cart and shopping agents, Amazon’s cross-merchant Buy for Me and auto-buy features, and Target’s Gemini checkout all point to a more brokered interface layer. Momentum also increased around marketplace-led growth, with Walmart, Target, and others using third-party assortment, ads, and membership to monetize general merchandise beyond core product sales. Rapid delivery and mission-based baskets also intensified, while the main constraint remains trust, data quality, and margin pressure.
Dominant Themes
High-density signal formations
Loading cluster map
Aggregating signals by recency and strength
Fastest-Rising Themes
Themes showing the strongest momentum
Loading cluster history
Reading snapshot progress over time
Analysis
Interpretation of what’s changing
Retail is becoming a machine for delegated buying
Full analysis summary: What looks like convenience is starting to behave like delegation. Amazon’s Buy for Me, Alexa price-triggered auto-buy, and Google’s push for a Universal Commerce Protocol all point to the same shift: the shopper is no longer the only actor in the transaction. The interface is becoming a broker, and in some cases an agent. That matters because the scarce resource is changing. In the old model, retailers competed to win attention and then convert a human browser. In the emerging model, they need to be legible to systems that can search, compare, wait, and execute. The winning layer is not the prettiest storefront; it is the one that can reliably translate intent into purchase across fragmented catalogs, merchant sites, and checkout flows. Think of it less like a mall and more like an air-traffic control tower. The tower does not own the planes, but it decides what lands, when, and under what rules. That is where the power is moving: toward the protocols, trust controls, and data standards that let an agent act safely on a shopper’s behalf. Rithum’s focus on governance is a clue that this is not just a UX experiment; it is an operational problem. The implication is uncomfortable for retailers that still rely on browsing friction. If an AI can monitor prices, assemble a basket, and buy at a threshold, then “loyalty” may increasingly mean compatibility with an agent’s rules, not emotional attachment to a brand. That could compress the value of traditional merchandising theater while rewarding merchants that expose clean data, fast checkout, and dependable fulfillment. There is a limit, though: delegated commerce only works where trust is high enough for users to hand over control. Categories with ambiguity, high consideration, or emotional weight may resist automation longer than commodity replenishment. So this is not the end of human shopping. It is the start of a split market, where some purchases are still browsed and others are simply authorized.
Retail Is Splitting Into the Store and the Machine Around It
Full analysis summary: The important shift is not that retailers are adding more services around shopping. It’s that the margin is moving out of the product transaction and into the machinery that surrounds it. Target’s non-merchandise growth, Walmart’s marketplace and ad momentum, and the broader NRF talk track all point to the same decomposition: the item on the shelf is becoming the traffic source, while the real profit pools sit in media, membership, seller access, measurement, and fulfillment. Think of it less like a store expanding and more like a retail engine being disassembled into parts that can each be priced separately. That changes how retailers behave. Once the customer journey is monetizable in layers, the retailer no longer needs to win only on assortment or price. It can win by owning the ad slot, the subscription gate, the marketplace take rate, or the delivery promise. A purchase becomes a stack of toll booths. The signals from Amazon and Target make the mechanism even clearer. Personalized deal guides, auto-buy thresholds, and paid early access are not just convenience features; they are ways to pre-shape demand before the shopper reaches an open market. They compress decision time and route buyers into controlled paths, where the retailer can steer timing, inventory, and margin capture. There is an implication here that investors sometimes miss: headline merchandise growth may understate the real business change. A retailer can look flat in product economics while quietly improving the economics of the system around the product. The caveat is that this split is not complete or uniform. Some categories still depend on pure assortment and price, and some monetization layers are still immature or noisy. Measurement, especially in physical retail media, is improving but not yet fully standardized. So the direction is clear, even if the accounting is still catching up.
Retail Is Becoming Machine-Readable Before It Becomes Human-Persuasive
Full analysis summary: General merchandise is starting to look less like a store aisle and more like a protocol. The shopper still exists, but increasingly as a set of rules: target price, preferred delivery speed, trusted seller, auto-buy threshold. Amazon’s price history, Alexa shopping, and “Buy for Me” are all nudging demand away from browsing and toward delegated execution. That matters because the competitive unit changes. If an AI assistant is deciding whether to buy, the winning merchant is not the one with the best slogan or the prettiest homepage. It is the one whose catalog is clean, prices are legible, inventory is current, and trust signals are machine-readable. In other words, the shelf is being replaced by a scorecard. This is why the Walmart-Google-Gemini flow is more important than a simple partnership announcement. It moves discovery, carting, and checkout into an interface that is not owned by the retailer. The retailer still supplies the goods, but the front door is being rented from someone else. That creates a new kind of dependency: brands and merchants may have to optimize for agent compatibility the way they once optimized for search rankings. The implication is uncomfortable for anyone built around traditional merchandising. Human persuasion still matters for high-consideration or emotional purchases, but routine general merchandise is becoming more like auto-renew software than a shopping trip. The merchant’s job shifts from “convince” to “be eligible.” There is a catch. Delegated shopping only works when the underlying data is trustworthy and the economics are stable. Price history and auto-buy make volatility more visible, but they do not eliminate it. If fulfillment is unreliable, if assortment is messy, or if consumers do not trust the assistant to spend on their behalf, the machine-readable layer stalls. So the real race is not just for attention. It is for the rules engine underneath attention.