$4B market opportunity (18 mo), who will lead it?

Today’s blog covers possible pricing models and market structures for agentic transactions, a new type of demand (purchase order with a payment instrument). Retailers may despise the idea of a new aggregator, but they can’t say “no” to a PO by their customer. US retailers spend over $400B on marketing ($90B of which is digital marketing). There is no CAC for an agentic transaction.. While the daily innovations of AI and Agentic is fascinating, it is the economics and structures for pricing value that will influence participation, value creation and market success.

Economics Structure Will Define Success

AI and Agentic represent a transformative shift in commerce. While the tech of agentic commerce evolves daily, the economic models and governance structures remain TBD. In my view, history clearly shows that market dominance is forged through robust economic structures (ie networks, markets, platforms) which incentivize broad participation and investment. The central challenge is in designing these economic frameworks to measure, monetize, and distribute value effectively. With governance (or market forces) balancing across the multitude of stakeholders.

As outlined previously, Agentic transactions introduce a novel form of demand: fully qualified consumers equipped with validated payment instruments and explicit purchase orders. This theoretically translates to a 100% conversion model (no abandonment). Agentic demand fundamentally alters the economic calculus compared to conventional e-commerce marketing funnels and a radical remaking of Customer Acquisition Cost (CAC). If an agent presents a purchase order with payments, the conventional marketing funnel is effectively bypassed for that specific transaction. Consequently, the economic imperative for sellers shifts from broad-spectrum marketing to influencing the agent’s decision-making. 

Beyond the daily improvements in Agentic efficacy is a strategic inflection point defined by economic structures. Early movers who successfully architect and implement a winning economic model will establish significant network effects and competitive moats and will be the ultimate determinant of success and failure.

Retailer Value

My Agentic retailer survey revealed apprehension regarding disintermediation and the potential for new aggregators to sever direct customer relationships. Retailers are keen to leverage own data to improve consumer experience within their own domain. However there is a pragmatic acknowledgment that retailers “must be where their customers ARE not where they want them to be,” compelling response to new channels that serve existing customers, much like DoorDash.

An agentic transaction, by its nature, may require no direct marketing expenditure by the retailer and, theoretically, incurs no incremental fulfillment cost beyond standard operations. In a scenario where the agentic platform levies no charge, the CAC for that specific transaction approaches zero. However, a critical trade-off emerges: a customer acquired via an agentic platform may exhibit lower Lifetime Value (LTV) for the retailer. This shift in LTV is pivotal, especially considering the common benchmark of a 3:1 LTV/CAC ratio for sustainable growth. The diverse economic profiles across retail segments—for instance, grocery with net margins around 1.5% and high loyalty, versus specialty retail with net margins near 3.5% and significantly higher CACs will dictate varying willingness to engage with different agentic economic models. For example Low-margin grocery sectors may find models with substantial per-transaction fees untenable.

The dynamic of customer loyalty shifting towards the aggregator implies that the agentic platform itself may become the primary “brand” or trusted entity for the consumer. This could catalyze intense competition among agentic platforms to cultivate user trust and loyalty. The basis of this competition will define a new frontier of brand development in commerce. In response, retailers possessing strong existing loyalty programs and rich first-party data may endeavor to launch proprietary “captive” agents or pursue deep integrations with broader agentic platforms to preserve a degree of LTV. This could foster a hybrid ecosystem where retailer-specific agents, offering preferential terms or access to exclusive inventory, coexist and interoperate with generalist agents.

As outlined, agentic platforms may expand the ability of manufacturers to engage in Direct-to-Consumer (DTC) sales, potentially bypassing traditional retail intermediaries. This is not merely a channel shift; it represents an opportunity for manufacturers to gain direct access to invaluable consumer data and demand signals, thereby diminishing their reliance on retailers for market intelligence and potentially restructuring retail supply chains.

Metric	Traditional Retail (Grocery)	Traditional Retail (Specialty)	Agentic Transaction (Platform-Neutral)	Agentic Transaction (Platform Fee Model)
Net Margin	~1.5%	~3.5%	Potentially higher (no direct marketing)	Variable, contingent on fee structure
Avg. CAC	Lower (loyalty-driven)	Higher (e.g., $129 Fashion)	$0 (user query, if no platform fee)	Platform fee (e.g., % of CAC saving)
Typical LTV	Higher	Moderate	Potentially Lower (aggregator loyalty)	Potentially Lower
LTV/CAC Ratio	High	Aims for ~3	Infinite (if CAC $0) / Undefined	Requires re-evaluation based on fee & LTV
Marketing Funnel Cost	Significant	Significant	Negligible for the transaction	Negligible for the transaction

Economic Frameworks for Agentic Commerce Transactions

Lets take a look at three different models: a publishing model, platform pricing of purchase orders, and demand-side auctions. Of course there are many other options, but a discussion of the trade offs here are useful to get a view on stakeholder reactions, and potential for disruptive value (and traction). 

Agentic in a Publishing Model

As discussed previously, This model extends established digital advertising mechanics and operates today in search engine optimization (SEO) and DSPs / SSPs exchanges where advertising networks adapt to facilitate placement. Affiliate revenue models, where agent developers earn commissions on sales they facilitate, also align with this approach.

Economically, this model provides a direct and immediate monetization pathway with existing ad markets, exchanges and stakeholders. However, it introduces the risk of “seller bias,” within Agentic results, as the quality and impartiality of results given that agents may prioritize sponsored recommendations over the user’s optimal outcome. Such bias can lead to information asymmetry and principal-agent conflicts. Sustained exposure to biased results may erode consumer trust in agentic platforms, thereby impeding broader adoption. 

Platform Pricing 

Under this framework, the agentic platform (e.g., Google’s Gemini) would furnish unbiased AI-driven results, but the fulfilling supplier would remit a fee, ostensibly representing the savings in CAC.  This fee could be negotiated based upon average CAC for that retailer/segment.  Rather than complain about a 3% payment processing cost, Google agentic transactions could run at a 7%, a cost  that the retailer is glad to pay as it is a PO set for fulfillment. Note that this model also supports a direct to consumer (DTC) end run by the manufacturer directly (think Samsung or Apple).

The operational complexity of determining “average CAC for that retailer/segment” and negotiating fees on a granular basis could be substantial, necessitating robust data analytics and benchmarking capabilities. At scale, retailers fears could be realized as the new “aggregator” would wield considerable pricing power as it steers demand. 

A more sophisticated pricing approach could entail dynamic fee structures, varying not just by retail segment but by individual retailer’s historical CAC, prevailing market conditions, or specific product categories. This requires significant understanding of the retailer and the market, but allows the platform to function as a highly effective price-discriminating source for these purchase orders.

Demand-Side Auctions

This model envisions an agentic adaptation of contemporary demand-driven advertising networks, where qualified demand (the purchase order) is published, and suppliers competitively bid to fulfill it. This establishes a real-time, auction-based market for individual agentic POs.

The auction structure holds the potential for maximum consumer benefit through lower prices driven by competitive bidding. It could also foster highly efficient price discovery and resource allocation. However, auctions would also create significant resistance from retailers, who would be compelled to participate in a market that could commoditize their fulfillment services and shift pricing power substantially towards the buyer (acting through their agent) thus creating market-making challenges.

For auctions to scale effectively, a degree of standardization in how agents articulate product or service requests is paramount. SKU data is NOT standardized. Thus, common ontologies or descriptive standards for products, usable by agents, would be a prerequisite, posing a significant data governance and interoperability challenge. MCP provides the protocol.. But the mess of comparing non-standard goods and price is complex.

Zero Cost is Not an Option

Agentic platforms cannot operate at zero cost and constructing a scalable economic model, with willing participants, is a challenge. Beyond transaction-centric models, platforms hold the advantage in diversified revenue streams such as transaction fees, lead generation, payment, risk data and affiliate marketing. There is also the prospect of retailer agents collaborating with Agentic platform agents to create peer-to-peer sharing economies or enable users to barter either identity, or anonymized data or services for tangible benefits.

The monetization of anonymized shopping data, where users might consent to share such data in exchange for discounts or exclusive offers, introduces a novel asset class. This aggregated intent data, supplied by retailer and curated by the agentic platform, becomes key to quality of results. The platforms that control this data flow, will have created shared economics with the permissioned holders of this information and gain significant advantages, which can be monetized through insights provided to retailers, CPG companies..

The “cost to serve” for agentic platforms, particularly those offering sophisticated, unbiased advice or managing complex negotiations as envisaged in an auction model, can be substantial. If platforms are unable to adequately monetize through primary transaction models they may be compelled to pivot towards data monetization or advertising-supported models. Such a shift could, paradoxically, reintroduce the very biases or privacy concerns that a platform might have initially sought to avoid by championing unbiased results or fee-for-service models. This highlights an inherent tension in platform business models: the initial choice of economic architecture can create feedback loops that influence long-term platform behavior and, ultimately, the market’s structure.

Pricing and Governance 

The development of agentic commerce will not occur in a vacuum; existing incumbents, particularly in payments and technology, are poised to play a significant role in shaping its foundational infrastructure and economic pathways.

Payment networks like Visa and Mastercard have gone through most of this in their own structure and just as Visa has adapted to agentic transactions. This proactive engagement allows payment networks to define new bespoke pricing models for agentic transactions, analogous to Visa’s Flex Credentials where Affirm and a Merchant can set their own pricing for a BNPL transaction. Card networks are not merely about enabling payments but expanding their role in the trust and security layers for the entire ecosystem. By setting these standards early, they ensure their continued relevance and the #1 area for VAS growth (IMHO).

Google holds a big strategic advantage with existing service agreements that span the globe (retailers and advertisers), coupled with its best-in-class Google Analytics measurement platform, retailer data, and a payment method that just killed the checkout page GPay. Google is embedding AI into every stage of the purchasing process, from browsing to payment, and its agentic AI checkout feature automates data entry and completes transactions via Google Pay. I see a strong potential for Google to mandate GPay in its agentic commerce offerings (e.g., via Gemini) as they create a monetization opportunity at the end of the funnel, regardless of how it was sourced. Google Analytics further solidifies value creation by offering retailers a proven demonstration of ROI, making it challenging for competing measurement or agentic platforms to gain traction.

Every model discussed requires a service agreements with both the seller and the buyer. The only entities that have this in place today are card networks and Google.  

Market Sizing

The U.S. retail sales surpassed $7.26 trillion in 2024.  Of this, e-commerce accounted for approximately $1.337 T, representing about 22.56% of in-person retail spending. U.S. retail marketing expenditure (all marketing) is projected to reach approximately $317B in 2025, with retail digital ad spend estimated between $83B and $90B  in 2024. 

Assuming agentic transactions achieve a conservative 5% penetration of e-commerce, this would translate to nearly $66.85B in agentic Gross Merchandise Volume (GMV), near-term pricing opportunity for agentic commerce platforms from this GMV would be ~$4B (1-2% of marketing spend)  a pool largely reallocated from existing retail marketing and customer acquisition expenditures.

The $4 billion agentic  “pricing opportunity” represents a conservative 18mo estimate, depending on which segments participate and which customer segments are active. It is TBD on HOW the value will be distributed: how much accrues to agentic platforms, how much is retained by retailers as tangible CAC savings, and how much is passed through to consumers in the form of lower prices. The relative bargaining power of platforms, retailers, and potentially consumers (represented by their agents) will determine this distribution. 

A shift of even 5% of e-commerce to agentic models will impact ad tech and entail a significant reallocation of advertising budgets, disrupting the ad-tech ecosystem, and  diminishing revenues for platforms not strategically positioned in as a leader or stakeholder in Agentic.

Wrap Up

Agentic commerce will be defined by its economic architecture. While technological advancements have been the primary focus, it is the underlying economic structure (how value is created, measured, monetized, and distributed) that will determine market leadership and the ultimate configuration of this new commercial landscape. 

Google’s strategic advantages, including its vast network of existing retailer and advertiser agreements, position it as the clear leader in the early stages of agentic commerce. Payment networks like Visa and Mastercard also play a crucial role by establishing essential infrastructure for trust and security, thereby facilitating broader adoption with a potential to provide the governance, pricing and infrastructure for every other agentic platform. While the outcome is not settled, this is how I see the world today.

For economists and senior executives, the imperative is to engage in rigorous economic modeling and value-price assessment. A new type of demand is here and the “death of the checkout page,” signifies a fundamental reordering of the commerce value chain. If agentic platforms effectively shift the point of conversion upstream, value capture will migrate from optimizing checkout processes to influencing or controlling the agent’s decision-making calculus much earlier in the consumer journey. Future competitive differentiation in retail will hinge on AI sophistication, the ability to leverage data for profound personalization, and, critically, the authenticating the customer and understanding their needs is necessary.

Note that the ultimate “winner” in agentic commerce may not be a singular entity but rather a dominant economic model that could entail collaboration based upon value. Should an OPEN transparent, efficient, and economic model achieve critical mass, it could compel even the largest players to adapt or offer comparable structures. The ensuing network effects could establish such a model as the de facto standard, shaping the market according to its core principles. The focus, therefore, must be on architecting these economic foundations.

Oh.. BTW my friends in Europe will just depend on the government to define all this for them and try to prevent US Big Tech from gaining any traction (zing).