What questions should boards ask management about agentic commerce readiness?

By James Thomson, Managing Partner of Equity Value Advisors · Published October 12, 2026 · Last updated October 12, 2026

Short answer

Boards should ask management seven questions, one for each dimension of the Agentic Commerce Control Model: whether AI systems describe the products accurately, which agents and sellers carry them, what customer data the company keeps, what agent-driven sales contribute after all costs, how agent authority is verified, who owns cross-functional decisions, and what triggers escalation. The board oversees dependency. Management runs the channel.

Key points

Agentic commerce becomes a board matter when it affects the investment thesis, enterprise economics or risk profile, not when it becomes fashionable.
The seven questions test control, not activity. A long list of AI pilots answers none of them.
Good answers name an owner, a measure and a point where the company can intervene.
Boards that learned the Amazon lesson late can ask these questions early, while the cost of changing course is still low.

Why do boards need their own agentic commerce questions?

Management will naturally report agentic commerce as a growth opportunity: new traffic, new integrations, new pilots. The board’s job is different. It must understand whether the company is building deliberate delegation or accidental dependency on a new layer of intermediaries.

The scale makes the question worth asking now. McKinsey estimates that AI agents could orchestrate $900 billion to $1 trillion of US B2C retail revenue by 2030. Adobe reported that traffic from generative AI platforms to US retail sites grew 693% year over year during the 2025 holiday season, and that those visitors converted 31% more often than visitors from other sources. Neither figure means a given brand faces material exposure today. Both mean a typical private equity holding period now overlaps the years in which the channel takes shape.

Many boards treated Amazon as “another sales account” until it shaped pricing, advertising, inventory and enterprise value. The same pattern can repeat with AI platforms, faster.

What are the seven questions?

1. Product truth: Do AI systems get our product facts from an authoritative source, or are they guessing?

Agents rebuild a brand’s presentation from feeds, retailer listings, reseller pages and reviews. If the company has no single authoritative version of product facts, every intermediary will assemble its own. A good answer names the system of record, the owner of product truth and how often external listings are reconciled against it.

2. Brand representation: Do we know how AI assistants describe and compare our products today?

Management should be able to show recent results from structured testing across the major assistants, not anecdotes from a single prompt. A good answer includes an accuracy rate, the most common errors and the sources causing them. See how to protect product accuracy in AI recommendations.

3. Distribution control: When an agent recommends our product, which seller gets the order?

An agent can identify the right product and still route the sale to an unauthorized seller or a low-margin channel. A good answer covers authorized seller policy, how agent-routed orders are monitored and what happens when a gray-market seller wins them. See whether AI shopping agents can increase unauthorized sellers.

4. Customer control: Which parts of the customer relationship do we keep in each agent channel?

Identity, consent, transaction history, loyalty, service and repeat purchase can each sit with a different party. A good answer states which elements the company must retain, which it trades away and why the trade is worth it. See who owns the customer relationship when an AI agent buys.

5. Economic control: What does an agent-driven sale contribute after fees, advertising, returns and reacquisition?

Top-line agent revenue tells the board little. Two channels with identical revenue can differ sharply in value if one yields a known repeat customer and the other an anonymous sale that must be bought again. A good answer reports contribution margin and incrementality, not just attributed sales.

6. Permission and risk control: Can we show which agent acted, under what authority and with what data?

Software acting for a customer must have demonstrable authority. A good answer explains how the company recognizes legitimate agents, limits what they can do, handles disputes and wrong purchases, and keeps records. See the legal, liability and fraud risks of agentic commerce.

7. Governance control: Who decides when ecommerce, marketing, IT, finance and legal disagree?

Each function can act sensibly within its remit and still produce poor enterprise economics. A good answer names the executive with authority over agent connections, data sharing, commercial terms and channel conflict. See who should own agentic commerce.

How should the board read management’s answers?

Signal	Strong answer	Weak answer
Ownership	A named executive with matching authority	“Everyone is involved” or a committee with no decision rights
Measurement	Accuracy rate, contribution margin, concentration trend	Pilot counts, traffic screenshots, press coverage
Intervention	Clear ability to change terms, data feeds or integrations	Integrations that cannot be modified or disconnected
Economics	Full cost view, including reacquisition	Attributed revenue only
Risk	Documented agent verification and dispute handling	An assumption that platforms handle it

Apply four tests to every answer: Do we know what is happening? Do we know who owns it? Can we intervene if conditions change? Can we measure the economic consequence?

When should these questions move from annual to quarterly review?

An annual review suits most consumer brands today. Move the topic onto the quarterly agenda when any of these triggers appears:

An AI platform becomes a meaningful source of demand or contribution.
The company loses material access to customer data in a channel.
One AI platform or marketplace agent accounts for a concentrated share of discovery.
Acquisition economics deteriorate because agents compress price comparison.
Fraud, consumer trust or regulatory exposure becomes significant.

When the topic becomes material, report it inside the existing ecommerce board pack, not in a separate AI dashboard. Agentic commerce competes for the same customers, inventory and capital as Amazon, DTC and wholesale, so the board needs one view of enterprise economics.

What should the board avoid asking?

Boards add little by asking which protocol to adopt, which agency to hire or which prompts to test. Those are management decisions. The board adds value by asking whether management can explain the dependency it is creating and reverse it if the terms change. Protocol choice matters to the board only when it locks the company into one platform’s economics; see whether boards should care which protocol management adopts.

What good looks like

Management can answer all seven questions in a short written briefing, with data. It has tested how AI assistants represent the products, knows which sellers capture agent-routed demand, can show contribution by channel and has an executive empowered to resolve conflicts. The board receives enough information to see material changes in enterprise value and risk, and no more.

Frequently asked follow-up questions

When does agentic commerce become a board-level issue?

Agentic commerce becomes a board issue when it affects the investment thesis, enterprise economics or risk profile: meaningful contribution, dependency on one AI platform, loss of customer data access, significant fraud or regulatory exposure, or a shift in how customers discover the company’s products.

Should the board create an agentic commerce committee?

Usually not. Most boards should fold agentic commerce into existing ecommerce and risk oversight. A management steering group can help coordinate functions, but the board’s role is to confirm that one accountable executive has authority over the decisions that matter.

What metrics should the board see on agentic commerce?

The core set is agent-driven revenue and contribution, customer data capture by channel, product information accuracy in AI answers, authorized versus unauthorized fulfillment, platform concentration, and return and dispute rates on agent-placed orders.

Sources

McKinsey & Company, The agentic commerce opportunity, October 17, 2025.
Digital Commerce 360, Adobe data on generative AI holiday shopping traffic, 2025, January 13, 2026.

What changed in agentic commerce, and what it means for brands.

A concise weekly brief for brand executives navigating the shift from human-led shopping to agent-mediated commerce. Follow the developments that matter, the implications for brand control and governance, and the questions senior teams should be asking now.

Read The Agentic Commerce Brief on Substack