Agentic Commerce: Write for the AI Agent and the Buyer
Lab NotesStrategyNote #20

Agentic Commerce: When an AI Shops for Your Buyer, Every Page Needs Two Readers

Half of shoppers already use AI somewhere in the purchase, and agents are starting to do the comparing for them. The technicians are racing to make product data machine-readable. That's half the job. The human still approves the buy, and the human still needs a reason.

Fred Smith, founder of iFix Marketing Fred SmithDigital futurist · Founder, iFix Marketing 9 min read
// The direct answer9 min read

Agentic commerce means AI assistants and agents increasingly do the research, comparison and shortlisting for shoppers, and in some cases complete the purchase. That makes clean, accurate, structured product facts essential, because an agent can't recommend what it can't read. But surveys show shoppers still want to approve the final decision, so the purchase has two readers: the agent that shortlists and the human who says yes. I call a page built for both the Two-Reader Page. Facts get you on the shortlist. Words win the approval.

Your ecommerce lead came back from a conference with a new phrase and a new panic: agentic commerce. AI agents that research, compare and even buy on a shopper's behalf. The vendors are selling feeds, protocols and “agent-ready” product data. The board wants to know if you're ready. Nobody can agree on what ready means.

I'm a digital futurist, and I think this shift is real and arriving faster than most plans assume. But the conversation has the same blind spot as every technical wave before it. Everyone's asking how to make the machine understand the product. Almost no one's asking what the person who approves the purchase needs to hear.

// 01

What the agentic commerce data shows

The surveys define “using AI to shop” differently, so compare numbers only within the same study. The direction is consistent:

  • It's mainstream already. One large 2026 survey of about 21,000 consumers across 11 markets [1] found 53% now use AI for shopping, with the US slightly lower at 48%. Among US shoppers who use AI, completing a purchase with it rose about 70% since 2025.
  • General assistants lead. The same survey found 56% had used a general AI chat or search tool for shopping, against 27% who'd used a retailer's own AI assistant.
  • Most of it is still research. A separate study [2] found AI shopping use concentrated before the purchase: about 41% researching products, 33% looking for reviews and 31% hunting deals.
  • People want the final say. A survey of 2,000 US consumers [3] found shoppers open to AI doing more, but only when they feel protected, in control and able to approve the final decision.

So the agent is becoming the researcher and the comparison engine. The human is still the buyer.

// 02

What the technician sees, and what the strategist sees

The technician sees a data problem. Product feeds need to be complete, accurate and structured. Prices, inventory, specs, shipping and returns need to be readable by machines. Checkout needs to work for an agent. All of it is necessary, and most catalogs are messier than anyone admits.

The strategist sees the moment after the agent hands over its shortlist. “Here are three options that match what you asked for.” Now the person reads three summaries, maybe clicks through, and decides. What makes them pick yours? Not the spec table. All three products cleared the spec table, or they wouldn't be on the list. They pick the one whose reason makes sense to them: the sentence that says who it's for, what it solves and why it's the safe choice.

The technician makes you findable to the agent. The strategist makes you chosen by the person.

I spent almost four years with an automaker's dealer advertising association in the Western region, helping the ad teams understand what the dealers were saying and how the money would be spent each quarter. That's a two-audience job: every message has to make sense to the people writing the ads and to the people spending the money. Agentic commerce is the same structure: one reader wants facts it can match, the other wants a reason it can believe.

The split

The agent decides ifyou're on the list.The person decidesif you're theanswer.

// 03

The mechanism: the Two-Reader Page

A Two-Reader Page serves the agent and the human on the same page, without one getting in the other's way. It has two layers:

  • The fact layer, for the agent. Complete, consistent, structured truth: exact specs, real prices, availability, shipping times, return terms, compatibility, materials, sizes. No marketing language in the facts. Agents compare, and vague facts lose comparisons.
  • The reason layer, for the human. One plain sentence on who this is for and what it solves. The top objection, answered. Proof a person can trust: real reviews, specific guarantees, honest limits. This is where the buyer decides.

The two layers fail in opposite ways. Pages with great copy and messy data never make the shortlist. Pages with perfect data and no reason make the shortlist and lose the approval, because the human sees three spec-identical options and picks the one with the clearest story, or the cheapest.

That second failure is the dangerous one. If facts are all you give the person, you've turned yourself into a commodity, and commodities compete on price.

How to read the agentic shift before it lands on your category is a regular topic inside the iFix Lab.

lab/bench/the-math-illustrative-shIllustrative
// 04

The math (illustrative): shortlisted, then what

These are modeled numbers for illustration, drawn from no real account.

The setup. A retailer does $30M a month. Say AI-assisted shopping touches 20% of its category's purchases this year and that share is growing.

Facts only. The team cleans its feed and gets onto agent shortlists for 40% of relevant queries. When it's shortlisted against two rivals, it wins a third of the time, the expected share for an undifferentiated option. Most of those wins go to whoever's cheapest that day, so the team discounts to compete. Margin falls three points on that revenue.

Facts plus reasons. Same feed work, same 40% shortlist rate. But each product page now opens with a plain sentence on who it's for and what it solves, answers the top objection, and shows specific proof. Say the win rate when shortlisted rises from 33% to 45%, without discounting. On the agent-touched slice of the business, that's roughly a third more revenue at full margin.

The compounding part. Agents learn from outcomes. Products that get chosen and don't get returned tend to get recommended more. Reasons that match reality reduce returns. Good words feed the machine better data over time.

Run your own numbers, starting with how often you're shortlisted and how often you win from the shortlist.

IllustrationWin rate once an agent has you on the shortlist
33%Facts only45%Facts plus reasons
Clean feed, discounting to winSame feed, a plain sentence of why

Roughly a third more revenue on the agent-touched slice, at full margin.

// 05

Writing the reason layer

The reason layer is short. That's what makes it hard. For each product or offer:

  • The who-and-what sentence. “For runners with flat feet who want stability without a stiff, heavy shoe.” Not “Engineered for performance.”
  • The top objection, answered. In the buyer's words. “Will it fit wide feet?” answered in one line, with the specific fit detail.
  • Proof a person trusts. Reviews that mention the specific benefit, a guarantee stated plainly, honest notes on who it's not for.
  • Consistency with the facts. The reason can't promise anything the fact layer contradicts. Agents check, and people check returns.

Shoppers have been telling salespeople “show me, don't sell me” for decades. The reason layer is how you show them in one screen.

// 06

The returns test

There's a quick way to see whether your reason layer is honest: read your returns and your one-star reviews next to your product copy.

If the copy says “fits true to size” and the returns say “runs small,” the reason layer is lying, and agents will learn it. If the copy says “easy setup” and the reviews say “took me an afternoon,” the human approved a promise the product didn't keep.

Fix the sentence, not just the product listing. An honest “runs half a size small, so most people size up” loses a few impulse buyers and keeps far more of the ones who stay. In a world where agents learn from what gets returned, honest words aren't only ethical. They're a ranking signal.

Breaks at scale
// 07

Where the common advice breaks between $1M and $10M a month

At $1M a month, the advice is “get your product feed agent-ready.” For a small catalog, cleaning the feed is quick and the copy is often written by the founder, so both layers come out fine.

Between $3M and $10M, catalogs grow to thousands of items, copy gets generated at scale, and the reason layer quietly disappears. Every product page reads like a spec sheet with adjectives. The feed gets you shortlisted, and the page gives the human nothing to choose by.

The limit

An agent-readycatalog with noreasons is acommodity catalogwith betterformatting.

$10M+
// 08

What changes once you're already past $10M a month

  • Own the fact layer like finance owns the books. One source of product truth, audited, consistent across your site, feeds and marketplaces. Errors cost shortlists.
  • Write the reason layer for your top sellers by hand. The products that drive most of your revenue deserve a person's words. Machines can handle the long tail with strict rules.
  • Track shortlist rate and win rate. Ask the major assistants your buyers' shopping questions and log how often you're shortlisted and how you're described.
  • Watch returns as a words metric. Returns often mean the reason promised something the product didn't deliver. Fix the reason.
  • Plan for agent checkout, carefully. Where agents can complete purchases, make sure the human still sees what matters before they approve: price, terms, the reason, the limits.
// 09

The decision framework I'd use

Five checks for every important product page:

  1. Is every fact complete and consistent? Same specs, price and terms everywhere.
  2. Is there a who-and-what sentence? Plain words, first screen.
  3. Is the top objection answered? In the buyer's language.
  4. Is the proof specific? Reviews and guarantees that name the benefit.
  5. Would a person pick this from a list of three? If the only difference is price, the reason layer isn't done.
// 10

Questions senior buyers ask

Is agentic commerce real or hype?

The research and comparison part is already mainstream. Fully autonomous buying is earlier and growing. Plan for both, starting with the part that's here.

Do we need new technology to be agent-ready?

Mostly you need accurate, complete, consistent product data. Specific protocols will settle over time. Clean facts help under any of them.

Won't agents ignore marketing copy?

Agents ignore vague copy. They use clear, factual statements. The human still reads the reason before approving, and that's where your words decide the sale.

Should we build our own shopping assistant?

Maybe, but most shoppers use general assistants first. Make sure those can read and represent you accurately before you build your own.

How do we know if agents recommend us?

Ask them. Run your buyers' common shopping questions through the main assistants monthly and track whether and how you appear.

What happens to brand when agents do the comparing?

Brand becomes the reason a person picks you from the shortlist. That makes clear, specific words more valuable, not less.

// 11

Before your next catalog update

Ask an AI assistant to shortlist three products like yours for a specific buyer. Read the shortlist as the buyer would. If you're on it, ask whether you'd pick yourself. If you're not, look at your facts first.

Then write the one sentence that would make a person choose you from that list, and put it at the top of the page. Reading where the market is heading before it lands is what the iFix Lab does. The monthly Futures Readout is a recorded session that scores last month's calls, shows where signals are merging next and names three moves to make before the market notices, so your catalog is ready for both readers before your category's buyers hand their shopping to an agent. It's $97 a month, cancel any time, with a 14-day refund on month one. Join the iFix Lab.

// 12

Sources

  1. McKinsey and Company, Who wins when AI does the shopping? New evidence on agentic commerce, 2026.
  2. IBM Institute for Business Value with NRF, Own the agentic commerce experience, 2026.
  3. Synchrony, 2026 AI in Commerce Study: The Future of AI Shopping Depends on Trust, 2026.
// The rest of this note

The rest of this note is for people on the list.

Name, email, and where you sit in the industry. Thirty seconds. One sign-up opens every Lab Note on this browser.

  • The full note, the math and the framework
  • All 19 Lab Notes, open on this browser
  • The next note in your inbox before it goes up here
Your email gets a one-time code first, so it has to be real. Unsubscribe any time.
We send a one-time code to that inbox. Type it in and every note opens on this browser.

Already on the list?

Fred Smith, founder of iFix Marketing
// Who writes the Lab NotesFred Smith

Digital futurist and founder of iFix Marketing: the coach on the sideline with the clipboard. The Marine Corps trained me as a Combat Correspondent, and I started with words before I learned the dashboards. I write the Lab Notes and run the iFix Lab, flops included, so the next dollar goes out on evidence instead of guesses. The math in these notes is a model to think with, not a promise about your account.

The One Offer

Read it here.See what's nextbefore you spend.

This note is one idea, given away. Inside the iFix Lab you get the whole field. I watch answer engines, search, AI, signals and conversion from the sideline with the clipboard, and every week you get what's moving, what's hype and what's working, before your next dollar goes out.

  • 01Your Signal Map. Day one, one page: where you get found today, what AI says about you, what attention costs, and the seat you've been stuck in.
  • 02The Signal Briefing. Weekly: what moved in AEO, search, LLM discovery, signals and CRO, each with a verdict (hype, watch, working) and a move.
  • 03The Futures Readout. Monthly, recorded: last month's calls scored, where signals merge next, and three moves before the market notices.
  • 04The LLM Ops Vault. Copy-ready prompts labelled by job: LLM ops on top of offer and conversion.
  • 05The Hype vs Working Ledger. Dated trend verdicts, plus my own experiments log, flops included.
The iFix Lab · Membership
$97/month

Billed monthly until you cancel

Join the Lab for $97/month →
  • Cancel any time
  • 14-day month-one refund

First 14 days: if it isn't worth it, reply to any Lab email and month one is refunded. Cancel any time from your account, no calls needed.

The iFix Lab is a $97/month membership that renews monthly until you cancel. Cancel any time from your account. Lab Notes share my own tests and opinions. Results vary, and nothing here promises income or results. iFix Marketing, P.O. Box 1066, Jurupa Valley, CA 91752.

© 2026 iFix Marketing. All rights reserved.

The iFix Lab · $97/monthCancel any time
Join the Lab →
Read the full note · FreeThirty seconds to unlock
Unlock →