Executive Summary

For corporate decision-makers navigating the next wave of retail transformation, the question is not whether Agentic Commerce will arrive—it’s already here. The question is whether your brand will be visible when autonomous AI agents execute purchases on behalf of consumers.

This report frames Agentic Commerce not as a technological novelty, but as a fundamental channel conversation—the same conversation brands had when deciding to sell on Amazon in 2010, or when questioning whether to build a DTC website in 2005. The emergence of AI agents that discover, evaluate, negotiate, and execute purchases represents a structural evolution comparable only to the rise of e-commerce itself.

The economic reality is stark: The “Marketplace Era” is approaching saturation. Customer Acquisition Costs (CAC) have doubled or tripled since Apple’s iOS 14 privacy changes in 2021. Amazon’s fee structure has evolved into a complex “pay-to-play” ecosystem where brands surrender 40-50% of revenue through referral fees, fulfillment costs, and advertising spend just to maintain visibility. The traditional DTC model—once profitable through cheap Facebook ads—now operates at contribution margin negative until the second or third purchase.

Agentic Commerce promises a correction: replacing probabilistic advertising (paying for impressions with uncertain outcomes) with deterministic commerce, where brands pay transaction fees only upon successful conversion. Early data suggests this could restore 15-20% of margin relative to marketplace sales. However, this efficiency arrives with significant trade-offs: potential commoditization of brand equity and loss of visual control as the “digital shelf” transforms into a “context window.”

This report synthesizes historical fee structures, post-iOS 14 CAC trends, and emerging protocols like the Agentic Commerce Protocol (ACP) to establish a comparative baseline. We argue that Agentic Commerce is best understood as “Digital Wholesale”—a new channel requiring fundamental re-architecture of how brands present data, pricing, and value.

The analysis reveals a critical insight: The brands that resisted selling on Amazon didn’t hurt Amazon—they hurt themselves. Similarly, brands invisible to AI agents won’t slow the agentic revolution; they’ll simply disappear from consumer consideration sets. The risk of invisibility in the Agentic era is the modern equivalent of the unauthorized seller problem.

The following sections detail the economics of previous channel migrations, the technical mechanics of agentic transactions, comparative unit economics analysis, visibility strategies for the “zero-click” era, consumer trust dynamics, and strategic recommendations for the dual challenge of optimizing for algorithms while preserving human brand equity.

Part I: The Channel Conversation—Historical Economics of Migration

To rationalize the investment in Agentic Commerce, we must first quantify the economic pressures that forced previous channel shifts. The narrative of the last two decades is one of seeking arbitrage in customer acquisition, only to see that arbitrage erode as channels mature and intermediaries exert pricing power.

1.1 The DTC Correction: The End of “Cheap” Growth

The Direct-to-Consumer boom (2010-2020) was predicated on a specific economic reality: the cost to acquire a customer via Facebook or Google advertising was significantly lower than the wholesale margin given up to a physical retailer (typically 50%). Brands like Warby Parker and Dollar Shave Club proved they could “rent” targeted audiences from social platforms more cheaply than they could rent physical shelf space.

This dynamic inverted sharply post-2020. The introduction of Apple’s App Tracking Transparency (ATT) framework in iOS 14 (April 2021) severed the deterministic feedback loop that made digital advertising hyper-efficient. The “signal loss” meant advertisers could no longer track user behavior across apps with precision, forcing ad platforms to rely on probabilistic modeling.

The result was a catastrophic rise in Customer Acquisition Cost. Data reveals that for many industries, CAC has doubled or tripled between 2020 and 2025, eroding the margin advantage that justified the DTC model.

Table 1: The Erosion of DTC Efficiency (2020–2025 Trends)

Metric	Pre-iOS 14 (2019-2020)	Post-iOS 14 & AI Era (2024-2025)	Strategic Implication
Signal Fidelity	High (Deterministic Tracking)	Low (Probabilistic/Modeled)	Ad spend is less efficient; “wasted” spend increases significantly as targeting precision drops.
CAC Trend	Stable / Decreasing	Rising (est. +60-100% in verticals like Beauty/Apparel)	DTC is no longer the “low cost” channel; it requires high LTV to sustain.
Search Volume	Growing Organic Traffic	Declining Organic Traffic (-25% projected by 2026)	Reliance on paid acquisition increases as organic reach fades due to AI summaries.
Unit Economics	Contribution Margin Positive on 1st Order	Contribution Margin Negative until 2nd/3rd Order	Brands must prioritize retention over acquisition; the “growth at all costs” model is dead.

The “CAC crisis” forced a strategic migration back to marketplaces. Brands that once shunned Amazon for its lack of data ownership returned to the platform because Amazon offered something DTC could no longer guarantee: high-intent traffic. However, this safety came with a steep price tag.

1.2 The Marketplace Tax: Anatomy of Amazon Fees (2015–2025)

The shift from DTC to Marketplaces exchanged a marketing cost (CAC) for a transaction cost (Referral + FBA fees). While this stabilized volume, it compressed margins through steady fee escalation. Historical analysis of Amazon’s fee structure demonstrates a clear trend: the platform captures an increasing percentage of every dollar generated by third-party sellers.

Between 2015 and 2025, Amazon’s fee structure evolved from a simple referral/fulfillment model into a complex “pay-to-play” ecosystem. The introduction of storage utilization surcharges, low-inventory fees, and the necessity of advertising to retain organic ranking effectively raised the “take rate” for sellers.

Table 2: Evolution of Marketplace Friction (Amazon FBA Focus)

Fee Component	2015-2017 Era	2024-2025 Era	Change Impact
Referral Fee	Flat % (typically 15%)	Variable (Lower for <$15 items, Higher for others)	Complexity increases; margin compression on mid-tier items while incentivizing cheap goods.
Fulfillment Fee	Simple weight-based ($2.50-$3.00 range)	Granular & Higher (Peak, SIPP discounts, Low-Inventory Surcharges)	Base fees rose ~45% for standard items. New surcharges penalize operational inefficiency.
Ad Spend (TACoS)	Low necessity (Organic reach viable)	High necessity (Pay-to-play visibility)	Effective “commission” rises from ~15% to ~30-40% when factoring in ad spend required to rank.
Storage	Seasonal spikes	Storage Utilization Ratio Surcharges	Penalizes slow-moving inventory, forcing clearance pricing and margin erosion.
2026 Outlook	N/A	+$0.08/unit average increase	Fees continue to track with or exceed inflation, ensuring the “Amazon Tax” never decreases.

The argument for Agentic Commerce is that the current marketplace model has become saturated and expensive. The “Amazon Tax” (fees + ads + logistics) has reached a tipping point where brands are paying 40–50% of their revenue to the platform. Agentic Commerce introduces a potentially lower-fee model, albeit with different risks regarding data and brand equity.

1.3 The Nike Case Study: The Limits of Independence

No case study better illustrates the tension between channel independence and marketplace necessity than Nike’s relationship with Amazon. In 2017, Nike entered a pilot program with Amazon to sell directly, aiming to control counterfeit goods and gain access to Amazon’s massive customer base. By 2019, Nike exited the partnership, citing a desire to focus on “direct, personal relationships” with consumers—a move that heralded the peak of the “DTC Pivot.”

However, the post-2020 reality forced a recalibration. While Nike successfully built its own digital ecosystem (SNKRS app), the broader market shifts—rising CAC and the persistence of unauthorized sellers on marketplaces—demonstrated that leaving a major channel creates a vacuum that competitors or gray-market sellers will fill.

Nike’s strategic oscillation highlights a critical lesson for the Agentic era: Leaving a high-volume channel often hurts the brand more than the channel. If Nike is not on Amazon, customers buy Adidas or counterfeit Nikes. Similarly, if a brand is not discoverable by an AI Agent, the Agent will simply recommend a competitor. The risk of “invisibility” in the Agentic era is the modern equivalent of the “unauthorized seller” problem.

Part II: Defining the New Channel—Agentic Commerce Mechanics

Agentic Commerce is not merely “better search.” It is a new form of “Digital Wholesale.” In this model, the retailer does not sell to a human via a visual interface; they expose their catalog to an AI agent via an Application Programming Interface (API). The agent acts as a procurement officer for the consumer, executing the transaction autonomously.

2.1 The Protocol Layer: How Agents “Buy”

The technological backbone of this shift is the Agentic Commerce Protocol (ACP), co-developed by entities like OpenAI and Stripe. This open standard solves the “trust gap” that previously prevented bots from buying. In a traditional e-commerce transaction, the human enters credit card details into a form. In an agentic transaction, this is replaced by Shared Payment Tokens (SPTs).

Mechanism: The consumer authorizes the agent (e.g., ChatGPT) to spend up to a certain limit. When the agent finds the product, it does not pass raw credit card data to the merchant. Instead, it passes a secure, single-use, or scoped token (SPT) that the merchant’s payment processor (e.g., Stripe) validates. This token is scoped to a specific merchant and a specific transaction amount, reducing fraud risk.

Merchant of Record (MoR): Crucially, the brand remains the Merchant of Record. This distinguishes Agentic Commerce from a wholesale relationship where the platform (e.g., Amazon) buys the inventory. The brand retains the liability, the tax obligation, and the fulfillment responsibility, but the transaction origin is the agent. This allows the brand to keep the transactional data, even if they lose the pre-purchase clickstream data.

2.2 The “Instant Checkout” Model

The first live iteration of this is OpenAI’s “Instant Checkout,” currently active with partners like Etsy and rolling out to over one million Shopify merchants.

User Experience: A user asks ChatGPT for “a gift for a runner under $50.” The AI suggests a specific pair of socks. The user clicks “Buy” within the chat interface.

Backend: The order is injected directly into the merchant’s Shopify/Etsy backend via ACP. The user never visits the brand’s website.

Economics: The merchant pays a “small fee” on successful transactions. While specific percentages are guarded, industry comparisons suggest this will function like an affiliate fee combined with payment processing, likely in the 2-5% range plus standard processing fees.

This fundamentally changes the “channel conversation.” It converts the website from a storefront (designed for human browsing) into a fulfillment node (designed for agent API calls).

2.3 The Shift from Visual Commerce to Data Commerce

In traditional e-commerce, the “digital shelf” is visual. Product discovery happens through images, color palettes, lifestyle photography, and layout. The brand invests heavily in User Experience (UX) design to guide the human eye through a journey of discovery and conversion.

In Agentic Commerce, the “shelf” is a JSON file. The agent doesn’t “see” the hero image or read the emotional copy about “embracing your journey.” It parses:

{
  "product_id": "12345",
  "name": "Trail Runner Socks",
  "price": 18.99,
  "in_stock": true,
  "rating": 4.7,
  "review_count": 3421,
  "shipping_days": 2,
  "material": "merino wool blend",
  "sustainability": "certified organic"
}

This data structure becomes the new battleground. Brands that cannot translate their value proposition into machine-readable attributes will be filtered out before a human ever sees them.

Part III: Economic Analysis—The “Real Numbers” of Agentic Commerce

To evaluate this decision, brands must compare the unit economics of an Agentic sale against a Website sale and a Marketplace sale. The central question is whether the “Agent Fee” is lower than the combined cost of CAC (on DTC) or the “Amazon Tax” (on Marketplaces).

3.1 The Cost of Discovery vs. The Cost of Conversion

In the DTC model, brands pay for Discovery (CPM/CPC). They pay Meta or Google to show an ad to 1,000 people, hoping 10 click and 0.2 buy. The risk of non-conversion sits with the brand.

In the Agentic model, the paradigm shifts to paying for Conversion. The AI agent does the discovery work. The brand only pays (via the transaction fee/commission) when the sale occurs.

Table 3: Comparative Unit Economics Analysis

Cost Component	DTC (Website)	Marketplace (Amazon)	Agentic (AI Checkout)
Customer Acquisition	High (CPA via Ads) - Rising due to signal loss	High (Ads + Fees) - “Pay to Play” environment	Low/Zero (Organic Relevance) - Initially free, likely to evolve
Platform/Commission Fee	~2.9% + 30¢ (Payment Proc.)	~15% (Referral Fee)	Est. ~2-5% + Processing (ACP)
Logistics Cost	Variable (3PL rates)	High (FBA fees) - subject to surcharges	Variable (Merchant Fulfillment)
Data Ownership	100% Owned	Restricted / Masked (Amazon owns customer)	Shared / Masked (Brand is MoR but lacks clickstream)
Pricing Power	High	Low (Price Parity Rules enforce floors)	Low (Agent compares globally for best price)

3.2 The Margin Implication: A Case Study

The data suggests that Agentic Commerce could be margin-accretive relative to Amazon FBA, provided the “Agent Fee” remains lower than the combined Amazon Referral + Ad Spend burden.

Scenario A (Amazon): Selling a $50 item.

Amazon Referral (15%): -$7.50
FBA Fee (Standard): -$6.00
Advertising (TACoS 10%): -$5.00
Total Channel Cost: -$18.50
Net to Seller: $31.50

Scenario B (Agentic): Selling a $50 item.

Agent Commission (Est. 5%): -$2.50
Payment Processing (Stripe 2.9% + 30¢): -$1.75
Shipping (Merchant Rate): -$6.00
Total Channel Cost: -$10.25
Net to Seller: $39.75

This differential (+$8.25 per unit) is the “real number” argument for adopting Agentic channels. It represents the efficiency gain of removing the “ad auction” from the middle of the transaction.

However, this is only true if the agent discovers the product organically. If Agentic platforms introduce “Sponsored Recommendations”—which is inevitable as the channel matures—the economics will likely regress toward the Amazon model.

3.3 The Hidden Economics: Return Rates and Inventory Efficiency

Beyond the direct transaction costs, Agentic Commerce impacts several second-order economic factors:

Reduced Return Rates: AI agents that utilize precise sizing data, virtual try-on integration, and “best fit” logic can drastically reduce return rates. By ensuring the product matches the user’s specific needs before purchase, the agent acts as a quality assurance filter. For apparel and footwear categories where return rates often exceed 30%, this represents significant margin recovery.

Inventory Optimization: By enabling Just-in-Time (JIT) purchasing and predicting demand with higher accuracy, agents help retailers avoid overstocking and obsolescence. This reduces working capital requirements and storage costs, directly impacting the bottom line.

Higher Conversion Probability: Traffic driven by AI agents is reported to be significantly higher intent. While direct traffic volume may decrease, the quality of that traffic increases. The “browsing” and filtering have already been done by the agent before the interaction reaches the brand’s systems.

Part IV: The Visibility Crisis—From SEO to GEO

The most significant risk in moving to Agentic Commerce is the “Zero-Click” phenomenon. Gartner predicts that by 2026, traditional search volume will drop by 25% as users shift to AI chatbots. This creates a “Walled Garden 2.0” where the brand’s website becomes invisible, rendering traditional SEO less effective.

4.1 The “Invisible” Brand Problem

In a traditional Google search, a user sees 10 blue links. Even if they don’t click the top one, they scan the brand names, creating impression value. In an Agentic interaction, the AI acts as a curator. It might analyze 50 options but only present one recommendation to the user: “I found the best running socks for you; here they are.”

This represents the commoditization of the brand. The AI evaluates products based on structured data (price, specs, shipping speed, verified reviews), effectively stripping away the “brand halo” created by expensive marketing campaigns.

Third-Order Insight: This shifts the marketing battleground from persuasion (emotional ads) to data hygiene (structured feeds). Brands that rely on “vibe” or “lifestyle” marketing without superior product specifications may be filtered out by the algorithm before the consumer ever sees them. The “Share of Shelf” metric is being replaced by “Share of Context.”

4.2 Optimizing for the Machine: GEO and AEO

To survive, brands must pivot from Search Engine Optimization (SEO) to Generative Engine Optimization (GEO) or Answer Engine Optimization (AEO).

Mechanism: This involves ensuring product data is machine-readable, utilizing Schema markup extensively, and maintaining real-time inventory feeds.

New Metrics: Success is no longer measured in “clicks” or “visits,” but in citations within AI answers.

The New Content Strategy: Brands must create content not for human readers, but for the training data of future models. High-authority, fact-dense content is more likely to be retrieved by an agent than “fluff” marketing copy. Data from Seer Interactive shows that organic CTRs for top positions have already dropped from 28% to 19% when AI Overviews are present.

Table 4: The Shift in Discovery Metrics

Traditional SEO Metric	Agentic Era Equivalent	Why It Matters
Organic Clicks	Agent Citations	How often does the AI reference your product in its answer?
Bounce Rate	Conversion Rate (Direct)	High-intent agent traffic converts or exits; no “browsing” middle ground.
Time on Site	N/A (Obsolete)	Users never visit site; transaction happens in chat interface.
Impressions	API Query Volume	How many times do agents query your catalog data?
Keyword Ranking	Attribute Relevance	Does your structured data match the agent’s search parameters?

4.3 The Bias of the Machine

It is critical to acknowledge that AI Agents are not neutral arbiters. Research indicates that LLMs often exhibit bias toward established global brands and may hallucinate product availability. Furthermore, unless explicitly guided, agents may favor generic products with lower prices over branded products with higher margins, creating a “race to the bottom” for commoditized goods.

Brands must therefore actively monitor their representation in AI outputs, much like they monitor brand sentiment on social media. “Agent SEO Audits” will become a standard practice, where brands systematically query major AI platforms to verify their products appear in relevant recommendations.

Part V: The Trust Equation—Consumer Adoption Dynamics

The economic model of Agentic Commerce relies entirely on consumer adoption. While the technology is ready (via ACP and similar protocols), historical precedent regarding technology adoption suggests a “Trust Lag.”

5.1 The Trust Gap

Current surveys indicate significant hesitation to delegate financial autonomy to AI. While 72% of U.S. consumers have used AI in some form, only roughly 24% to 34% are currently comfortable letting an AI agent make a purchase autonomously on their behalf.

Nuance: Trust is significantly higher for retailer-specific agents than for general AI. Consumers are more likely to trust a “Walmart Shopping Bot” than a generic “OpenAI Bot” with their credit card data, due to the established transactional relationship.

This creates a strategic opportunity for retailers to deploy “Owned Agents” that keep the customer interaction within their ecosystem while providing the efficiency benefits of AI-mediated discovery.

5.2 The “Agent-ish” Bridge

We are currently in a transitional phase described by Forrester as “Agent-ish” commerce. In this phase, Agents recommend products and pre-fill the shopping cart, but humans click the button to confirm the transaction.

Prediction: This behavior mirrors the adoption curve of mobile payments. Apple Pay was viewed with skepticism until the friction reduction (ease of use) outweighed the security anxiety. Once the friction of Agentic checkout proves to be lower than web checkout—eliminating the need to create new accounts or type addresses—adoption will follow a logistical, rather than emotional, curve.

The shift from “Search → Click → Buy” to “Ask → Approve” is the path of least resistance. The question is not if consumers will adopt this behavior, but when the friction reduction becomes compelling enough to overcome the trust barrier.

5.3 The Generational Divide

Survey data reveals a significant generational split in attitudes toward AI-mediated purchasing:

Gen Z and Millennials: Higher comfort with AI assistance for research and deal-finding, but significant resistance to fully autonomous purchasing, particularly for emotionally significant categories like gifts.
Gen X and Boomers: Lower initial trust, but higher conversion once trust is established through retailer-specific agents.

This suggests that the initial adoption vector will be through low-stakes, high-frequency purchases (household replenishment, basic apparel) rather than high-emotion categories (luxury goods, gifts, complex electronics).

Part VI: Strategic Imperatives—Framing the Decision

To speak to both sides of this decision—the Pro-Innovation faction and the Risk-Averse faction—we must frame Agentic Commerce not as a replacement for DTC, but as a high-efficiency acquisition channel that requires a new operational posture.

6.1 The Case for Immediate Integration

The Argument: “We are currently paying a premium for human attention on Meta and Google. That attention is shrinking due to zero-click searches. Agents offer ‘high-intent’ distribution where we pay for success, not exposure. By integrating now, we secure ‘first-mover advantage’ in the training data of these agents, training them to prefer our brand before the channel becomes pay-to-play.”

The Data:

Utilize the pro-forma margin analysis (Table 3) to show that an Agentic sale contributes more to the bottom line than a CAC-loaded DTC sale.
Highlight the Gartner statistic (25% search drop) as a quantifiable risk of inaction—if we don’t sell to the agent, we don’t sell to the user.
Point to early movers (Etsy, Shopify merchants) already capturing agent-mediated revenue.

6.2 The Case for Cautious Evaluation

The Argument: “Integrating with agents cedes the customer relationship to a third party (OpenAI/Google/Anthropic). We risk becoming a white-label supplier to an algorithm. If the agent decides a generic competitor is ‘better’ based on price, we lose the sale instantly. We also lose the ability to upsell via our site experience and cross-sell high-margin items.”

The Mitigation:

Maintain the Merchant of Record status (ensured by protocols like ACP) to keep the transaction data.
Use Agentic channels for acquisition of new customers (selling the ‘hero’ product), but use email/loyalty programs (owned channels) for retention and upselling.
Monitor the “Sponsored Recommendation” landscape carefully; as soon as agentic platforms introduce pay-to-play mechanics, re-evaluate the margin advantage.

6.3 The Synthesis: The Hybrid Strategy

The prudent decision is not binary. Brands should view their e-commerce architecture as API-first, preparing for the agentic future while maintaining the human-facing brand experience.

Technical Readiness

Implement the Agentic Commerce Protocol (or similar standards) via platforms like Shopify or Stripe. This is a low-risk infrastructure play that makes the brand “buyable” by bots.

Data Defense: Audit product feeds. If an AI cannot read your specs, price, and stock level in milliseconds, you do not exist in this channel. Invest in structured data hygiene:

Complete Schema.org markup
Real-time inventory APIs
Consistent product attribute taxonomy
Machine-readable sustainability/sourcing data

Portfolio Segmentation

Use Agentic channels for “utility” SKUs (replenishables, standard items) where the AI’s logic dominates. These are commoditized products where brand loyalty is low and price comparison is high.

Reserve the full DTC web experience for “discovery” SKUs (high emotion, complex storytelling) where human browsing is still preferred. Invest in experiential commerce for categories where the purchase journey is part of the value proposition.

The “Owned Agent” Strategy

Deploy retailer-specific AI shopping assistants to capture the efficiency benefits while maintaining customer relationship control:

Case Study: Walmart’s “Sparky” - Customer-facing agent that assists shoppers with discovery and decision-making, effectively keeping the query within Walmart’s ecosystem.
Case Study: Sephora’s Virtual Artist - AI-driven beauty consultant that provides “best fit” recommendations based on skin profiles and past purchases, creating loyalty through personalization.
Case Study: Nike’s Virtual Try-On - Solves the “sizing” problem through AI-driven measurement and product matching, reducing returns while building direct customer relationships.

6.4 New KPIs for the Agentic Era

Retailers must adopt new Key Performance Indicators (KPIs) to measure success in this environment:

Agent Conversion Rate: The success rate of agent-driven interactions. How effectively does the brand convert an agent’s query into a sale?
Citation Frequency: How often do AI platforms reference your products in recommendations?
API Latency / Uptime: Technical metrics that now directly correlate with sales visibility.
Structured Data Completeness: Percentage of catalog with full machine-readable attributes.
Return on Algorithm (ROA): Measuring the efficiency of investments in data structuring and API optimization, replacing traditional ROAS.

Conclusion: The Inevitable Wholesale

Agentic Commerce is the “Digital Wholesale” of the 2020s. Just as brands accepted that they must sell on Amazon to reach the Prime customer, they must accept that they will sell through Agents to reach the AI-assisted consumer.

The historical lesson from the transition to Online and Marketplaces is clear: Resistance to a new high-efficiency channel results in obsolescence, but total capitulation results in margin death. The winners of this new era will be the brands that treat AI Agents as a distinct customer persona—optimizing their pricing, data, and logistics to satisfy the algorithm—while fiercely protecting their direct relationship with the human on the other end.

The Bifurcation Thesis

Retail is bifurcating into two distinct modalities:

The Commodity Layer: Hyper-efficient, agent-mediated transactions for low-emotion, high-frequency purchases. This is where Agentic Commerce will dominate, stripping out marketing overhead and optimizing for price and logistics.
The Experience Layer: Hyper-experiential, human-centric discovery for high-emotion, identity-driven purchases. This is where brand storytelling, visual merchandising, and the “soft attributes” of shopping remain valuable.

Brands that attempt to sell commodities with storytelling, or luxury goods with bare-bones data, will find themselves invisible to the agent and irrelevant to the human.

The Real Numbers

The decision matrix is clear:

Current State: Paying 40-50% to Amazon or burning 30-40% on DTC CAC with declining returns.
Agentic State: Paying 5-8% on successful conversions with higher-intent traffic.
Risk: Loss of brand visibility and commoditization pressure.
Mitigation: Hybrid strategy with portfolio segmentation and owned-agent deployment.

The Path Forward

The decision is not if to engage, but how to structure the engagement to ensure the brand remains the destination, even if the journey is automated.

Immediate Actions:

Audit your data infrastructure. Can an AI agent read your entire catalog? Is your inventory API real-time?
Implement ACP or equivalent protocols. Make your brand “buyable” by bots through Shopify/Stripe integrations.
Segment your portfolio. Identify which SKUs should be optimized for agents (utility) and which should remain human-centric (discovery).
Launch pilot programs. Test agentic channels with high-volume, low-emotion products to gather baseline conversion and margin data.
Monitor the training data. Systematically query major AI platforms to verify your products appear in relevant recommendations.

The “real numbers” of rising CAC and Amazon fees leave little room for hesitation. The fourth channel is open, and the first movers are already training the agents to prefer their brands.

Key Data Summary Table

Metric	DTC (Post-iOS14)	Amazon FBA (2024/25)	Agentic Commerce (Early)
Primary Cost Driver	Ads (CAC)	Fees + Ads + Logistics	Transaction Fee
Est. “Take Rate”	~30-40% (Ad Spend)	~40-50% (Total Fees)	~5-8% (Est.)
Traffic Trend	Declining (Organic)	Stable / High	Growing (Exponential)
Data Transparency	High	Low	Mixed (Transaction only)
Control	Full	Low	Medium (MoR maintained)
Brand Visibility	High (Owned Experience)	Medium (Search-dependent)	Low (Agent-mediated)

The shelf is digital, the shopper is a bot, but the ultimate consumer remains, stubbornly and importantly, human. Build for both.