By: Matt Schiffman, VP Pricing Strategy & North America | Quicklizard
Google’s Universal Commerce Protocol announcement at NRF brought Shopify, Walmart, Target, and others into an alliance where Gemini can negotiate, apply loyalty rewards, and complete checkout inside a conversation. Within days, pricing teams were mapping what a UCP channel would require in their infrastructure. The protocol is no longer theoretical.
The discourse has centered on discovery: how brands surface when an agent shops. This framing, while intuitive, obscures what is actually happening.
Aggregation and Physical Commerce
The defining dynamic of the internet era has been aggregation: services that scale to all users attract the most users, which attract the most suppliers, which improves the service further. The result is winner-take-all economics where platforms capture demand and commoditize supply.
Physical commerce has resisted full aggregation for a structural reason: buying things involves friction that digital goods do not. Evaluation takes effort. Trust must be established. Comparison requires attention. These transaction costs preserved space for differentiated retailers.
Agentic commerce eliminates that friction on the demand side.
This distinction matters. Google isn’t solving logistics. Inventory management, shipping networks, returns processing: these remain someone else’s problem, and someone else’s potential margin. What Google dissolves is the cognitive work: evaluation, comparison, trust assessment. The customer-facing friction that previously justified the retailer’s role in the relationship.
The Funnel as Integration
Consider what the purchase funnel actually represented:
[Customer] → [Awareness] → [Consideration] → [Purchase] → [Loyalty]
Each stage was an integration point. Brand advertising drove awareness. Stores and websites owned consideration. Loyalty programs created switching costs. The customer relationship was the asset; transactions were monetization.
Agentic commerce restructures this:
[Customer] → [Agent] → [Protocol] → [Fulfillment]
The integration point moves from retailer to agent. A customer says “buy me running shoes” and the agent handles everything that previously constituted the retailer’s value proposition in the consideration phase.
The analogy is the shift from newspapers to the news feed. The information still exists. The sources still produce it. But the integration point, where attention gets allocated and value captured, moved to a new layer. Commerce is undergoing the same structural change.
The Agent Layer
The protocol announcement is Google’s, but the threat is architectural. An intermediary layer has emerged between customer intent and transaction. Whether Google, OpenAI, Apple, or others win that layer matters less than the fact that retailers don’t control it regardless of outcome.
This is the key point: the threat is structural, not competitive. Retailers aren’t losing to Google specifically. They’re losing to the existence of an agent abstraction they sit beneath.
What Breaks
Three foundations of retail economics become unstable.
Preference becomes machine-legible or invisible. The concern isn’t that agents only see “quantifiable” attributes. They’ll read Reddit sentiment, review velocity, return rates, social signals. The shift is from human-mediated discovery (I browse, I consider, I weigh intangibles) to algorithmically-mediated discovery (it surfaces based on signals, I confirm). Brand equity that cannot be expressed in machine-readable form, whether structured data or parseable reputation, loses its purchase on the decision.
Behavioral pricing erodes. Retail pricing strategy exploits predictable irrationality: anchoring on original prices, charm pricing, manufactured urgency. These techniques work because humans make decisions inefficiently. Agents optimize. This isn’t incremental pressure; at most retailers behavioral tactics contribute 200-400 basis points of margin. That’s survival margin, not optimization upside.
Customer data flows to the agent. Retailers have invested billions in first-party data on the thesis that superior customer understanding yields superior service. But if the agent intermediates the relationship, the agent accumulates the behavioral data. Retailers receive fulfillment records.
The Strategic Response
The obvious conclusion, universal price competition and margin compression everywhere, is probably wrong, or at least incomplete.
Agents arbitrage products with substitutes. But several forms of defensibility remain:
Products without comparisons. Private labels, exclusive assortments, curated bundles, genuine long-tail inventory. An agent cannot arbitrage a product against alternatives that don’t exist. This is medium-term defensibility. Agents will learn functional equivalences over time. But the medium-term is worth more than none.
Relationships that predate delegation. Customers who choose to buy directly, because they want the brand or the experience, belong to the retailer rather than the agent. These become disproportionately valuable. But loyalty is more fragile than balance sheets suggest. When friction disappears, so does the switching cost it was masking. Investment in owned channels, brand preference, community: the economics improve relative to agent-mediated demand, but only for relationships that are genuine rather than inertial.
Physical necessity. Products requiring bodies in space: apparel fit, sensory evaluation, complex consultation. These cannot be fully intermediated. The showroom, the fitting room, the expert consultation justify trips that commodity purchases no longer do. Stores shift from distribution infrastructure to customer acquisition infrastructure.
Speed of commercial response. An agent request requires millisecond response integrating inventory, pricing, and margin targets simultaneously. Most organizations route pricing decisions through weekly committees. The gap between these two realities is not closable with faster software. It requires different organizational architecture. The capacity to respond at machine speed becomes a moat because competitors lack it.
Information asymmetry. Agents optimize on available information. Proprietary knowledge, including customer history not shared with platforms, product performance data beyond specs, and local context, creates decision advantage. The catch: this erodes with integration. The more you participate in protocols that normalize your data, the less differentiated your knowledge becomes.
The strategic shape is a portfolio. No single moat suffices. Defensible retailers stack them: differentiated assortment for margin, direct relationships for ownership, physical presence for categories that require it, organizational speed for competition at machine pace, information asymmetry for decision advantage.
What Becomes Untenable
The corollary: certain positions stop working.
Undifferentiated assortment competing on national brands. Dependence on agent-mediated traffic with no owned customer base. Physical presence without experiential justification. Slow commercial decision-making requiring committee approval. Competing on data you’ve already shared with the platform.
The middle, adequate on all dimensions and excellent on none, faces the most pressure. The agent doesn’t need “good enough.” Second place in the optimization gets nothing.
The Platform Question
Retailers joining the protocol gain demand access. The agent layer gains the relationship and the data that follows from controlling the integration point.
The question every company facing platform dynamics must answer: are you building the platform, or building on it? Protocol participation is, by definition, building on someone else’s platform. That can be a viable business. It is not the business these retailers believed they operated.
The retailers who navigate this will treat protocol participation as a channel, not a strategy. They will invest in the moats that don’t depend on the agent layer: owned relationships, differentiated assortment, physical experiences that resist intermediation, commercial decision systems that operate at machine speed. They will recognize that the funnel they built their business around has collapsed, and defensibility now comes from different sources.
The retailers who frame this as a discovery problem, how to surface in the agent’s recommendations, will optimize for a game that has already moved on.
