- Does this really change anything for my company, or is it still stuff for big retailers?
- What exactly happened in a few weeks?
- What happens to my analytics data when an agent buys on behalf of the customer?
- Is blocking AI agents the right choice, or is it a boomerang?
- Is there a middle ground between blocking everything and opening everything?
- Three concrete scenarios: what changes for an e-commerce, a manufacturing company, and a professional firm?
- Which marketing levers really remain in the brand's hands?
- Why are we talking about this now: is the market really ready or is it still just hype?
- What are the most frequent mistakes companies are already making?
- How to start concretely: the steps to adapt the strategy?
- How to tell if my strategy is working with AI agents?
- So, is it worth adapting now or can you wait?
- How we see it at SHM
- Frequently Asked Questions
- Consumer AI agents (like Meta's Muse) complete purchases without the customer ever visiting your site, taking away visibility into preferences and behaviors.
- AI traffic to retail sites converts 42% better than regular traffic (Adobe Analytics, March 2026): blocking agents means losing your most decisive customers.
- The levers that remain in the brand's hands are: quality of the data catalog, access policy for verified agents, and oversight of the pre-agent moment.
- Deciding how to handle AI agents isn't a technical choice: it's a marketing decision that needs to be made now, not when the numbers drop.
An AI agent has already bought something on behalf of one of your customers — and you don't know it yet. Muse, Meta's new agent, has surpassed 2.5 million downloads in two weeks: faster than ChatGPT to the top of the Apple App Store.
Does this really change anything for my company, or is it still stuff for big retailers?
This changes everything for anyone selling online or managing a digital customer journey. AI agents are no longer just an enterprise tool: they are in consumers' pockets.
Meta's Muse, OpenAI's Dots, Anthropic's Cowork: these tools — called AI agents for consumers , that is, software that acts autonomously on behalf of a person — they don't wait for the customer to visit your site. They compare products, read return policies, and complete the purchase on infrastructures that the brand never touches.
The Italian e-commerce market is worth 66.6 billion euros in 2026 , up 6% compared to the previous year, according to the Politecnico di Milano Netcomm B2c eCommerce Observatory . A growing slice of these purchases is about to pass through agents that the brand does not control. The topic also concerns an industrial spare parts store, a consulting firm with online quotes, or a retail network with digital booking.
| Your situation | Do AI agents concern you? | why |
|---|---|---|
| Sell products online (e-commerce) | Yes | Agents compare prices and complete checkouts autonomously |
| You have a product/service catalog available online | Yes | Agents read the catalog and use it to make recommendations to customers |
| You use pixels and analytics to track behavior | Yes | Agent-based checkout doesn't trigger pixels: data disappears |
| Sell only offline or via sales rep | No, for now | The impact is indirect, but the comparison still happens online |
| You already have a brand-side AI agent | Yes | You need to decide how it interacts with your customers' agents |
What exactly happened in a few weeks?
In less than a month, three major players have made opposite decisions regarding the same AI agent. The market has split in two, and every brand must now choose which side to be on.
On September 20, 2026, Amazon showed a pop-up to Muse users browsing its store: unauthorized access, violation of terms of use. Two days later, PayPal announced that its customers can make purchases and checkouts via Muse at PayPal merchants worldwide. The day before that, Shopify had enabled Shop Pay for Muse on all its stores — and it did so default , with manual opt-out for merchants.
According to Shopify , merchants enrolled automatically must disable the feature from their admin settings if they do not wish to participate. On September 28, Shopify extended support for browser agents with three new tools that allow them to read, update, and complete transactions — always with the buyer's explicit authorization, as reported by TechCrunch .
| Company | Decision on Muse | Stated motivation |
|---|---|---|
| Amazon | Block with pop-up | Muse does not identify itself, captures credentials, no prior notification |
| PayPal | Full access | Enabling PayPal merchants to reach new customers via agent |
| Shopify | Access enabled by default | Shop Pay for all eligible merchants, manual opt-out available |
| Target | Partial access (no autonomous checkout) | Blocking automatic clicks at checkout, navigation allowed |
What happens to my analytics data when an agent buys on behalf of the customer?
Agent-based checkout doesn't trigger your tracking pixels. The order comes through, but the behavioral history that led to the purchase disappears.
When Muse completes a purchase on Shopify, Google Analytics and custom pixels don't trigger. Shopify only passes server-to-server events for "checkout started" and "checkout completed." The order shows up for the merchant, but the journey that led the customer to choose that product — the pages visited, the comparisons made, the time spent — never reaches your analytics tools.
If you have built over the years a CDP (Customer Data Platform, i.e., a centralized system that collects and unifies customer data), that infrastructure doesn't capture anything from an agent-assisted checkout. It's as if the customer entered through the back of the store, grabbed the product, and left without passing through the visible checkout.
This topic is intertwined with the growing tracking difficulty already discussed in the article on Google Site Kit and automatic tracking on WordPress : data gaps multiply, and every new channel adds a blind spot.
Is blocking AI agents the right choice, or is it a boomerang?
Blocking agents keeps customers who give up on the agent on your site. But those who continue to use the agent take themselves—and all their data—to a competitor that accepts them.
Amazon can afford to block Muse: it generates over 68 billion dollars in annual advertising revenue from people browsing its pages, and it has its own purchasing agent (Buy for Me) that identifies itself and allows brands to opt-out. Most companies have neither a comparable advertising business nor a proprietary agent to protect.
The reasoning is simple: a block only works if the customer abandons the agent and goes back to browsing manually. If, however, the customer prefers the agent, they take it to a competitor that accepts it. And with them, they take all the context the agent has stored: preferences, dietary restrictions, sizes, purchase history. Data that the brand will never see.
On the legal front, in August 2026 the Ninth Circuit (US federal court) ruled that the "user" accessing a site is the natural person, and the AI agent acts on their behalf. This limits the ability to use anti-hacking laws to block agents operating on the customer's mandate.
A year ago, AI traffic converted worse. Today, it converts 42% better. Blocking this traffic in 2026 means blocking the customers most intent on purchasing. The full data is available in the Adobe Digital Insights report .
Is there a middle ground between blocking everything and opening everything?
Yes: allow verified agents and block malicious bots. Payment networks have already built the infrastructure to do this.
Visa has developed the Trusted Agent Protocol in collaboration with Cloudflare: it allows merchants to verify an agent's digital signature and distinguish it from a scraper (a program that copies data without permission). Mastercard requires the registration and verification of agents before they can transact on its network, and every Mastercard issuer in Europe is already enabled for Agent Pay.
In practice: a brand can configure its system to accept verified agents at checkout and block unidentified ones. It is not a binary choice between "open to everything" and "closed to everything."
- 1 · Marketing team Define the access policy Deciding which agents to accept, with what level of autonomy, and at which stage of the journey (browsing, cart, checkout).
- 2 · IT / security team Implementing verified agent recognition Integrate Trusted Agent Protocol (Visa/Cloudflare) or Mastercard registration requirements to separate legitimate agents from bots.
- 3 · E-commerce team Configure checkout for agents If you're using Shopify, check your default settings (Sales channels > Agentic) and decide whether to keep or disable automatic access.
- 4 · Analytics team Adapt tracking Switching to server-to-server events to capture at least checkout data, even when the pixel doesn't fire.
- 5 · Marketing team Optimize the catalog for agent readability Structure product data, pricing, policies, and availability in a machine-readable format (schema markup, structured feeds).
Three concrete scenarios: what changes for an e-commerce, a manufacturing company, and a professional firm?
The impact varies by sector, but the logic is the same: those who arrive via an agent have already made up their minds, and the brand must make itself readable and reliable before the agent chooses.
Illustrative examples: numbers are meant to show the reasoning, they are not real customer data.
| Company type | Scenario with AI agent | Main risk | Lever to activate immediately |
|---|---|---|---|
| Fashion e-commerce (50K orders/year) | The agent compares sizes, prices, and delivery times across 5 stores in 10 seconds and completes the checkout on the most convenient one. | Losing orders without knowing why (pixel doesn't trigger) | Structured product feed, clear and machine-readable return policy, checkout open to verified agents. |
| Industrial manufacturer (B2B spare parts) | The purchasing manager's agent compares availability and prices between suppliers and sends the order to the fastest one | Not being in the agent's comparison set if the catalog isn't structured | Technical catalog with structured data, integration with B2B ordering systems, accessible APIs |
| Professional firm (online quotes) | The customer agent collects quotes from 3 firms, compares them, and books the first appointment | The quote is not readable by the agent if it is in PDF or an unstructured format | Quote form with structured output, calendar availability exposed via API or standard feed |
Which marketing levers really remain in the brand's hands?
Three levers remain controllable: the moment before the agent acts, the quality of the data the agent reads, and the direct relationship with the customer after the purchase.
The AI agent acts on the customer's behalf, but the customer has already formed their preferences before delegating to the agent. If your brand is in the customer's mind when they configure the agent, you have already won half the battle. This is the moment of Brand Awareness and of pre-purchase content : articles, videos, reviews that the customer read before opening the agent's app.
The second lever is the quality of the data catalog. The agent reads your site like a machine: if information on price, availability, return policy, and technical specifications is not structured and updated, the agent ignores it or misinterprets it. Those who have invested in technical SEO and schema markup — as we explain in the article on AI agents in media buying — starts with an advantage.
The third lever is the post-purchase relationship. The order arrives anyway: the customer is yours. Post-purchase communication — emails, support, loyalty programs — is still the brand's territory, not the agent's.
| Journey stage | Brand control | Concrete action |
|---|---|---|
| Pre-agent: preference training | High | Educational content, brand awareness, reviews, presence on channels where the customer gets informed |
| Agent configuration | Medium | Being in the set of brands that the customer knows and considers reliable before delegating |
| Browsing and comparison (active agent) | Low | Structured catalog, updated data, schema markup, machine-readable feeds |
| Checkout via agent | Low-Medium | Access policy for verified agents, checkout enabled, server-to-server tracking |
| Post-purchase | High | Email, support, loyalty program, upsell: the customer is still yours |
Why are we talking about this now: is the market really ready or is it still just hype?
The numbers say it's not just hype. AI traffic to retail sites is growing in triple digits, and in Italy, the AI market reached 1.8 billion euros in 2025.
According to Artificial Intelligence Observatory of the Polytechnic University of Milan , the Italian AI market has reached 1.8 billion euros in 2025 , with 50% growth compared to the previous year. 71% of large Italian companies have started at least one AI project. SMEs are still lagging behind: less than 10% have an active project. But consumer AI agents aren't waiting for SMEs to be ready: they are coming from the customers, not the companies.
On the European front, the report IAB Europe on the impact of AI on digital advertising (September 2026) indicates that 86% of European organizations are already using AI for marketing, and more than half of the sector expects agentic ad buying to reach operational scale within a year.
This shift is intertwined with the broader transformation of paid media: those managing campaigns must already deal with agents that intermediate purchasing decisions, as analyzed in the article on how to scale campaigns with AI agents .
What are the most frequent mistakes companies are already making?
The most common mistake is letting the IT security team make the decision about AI agents, without marketing knowing what is happening.
The original MarTech source describes a classic case: Target blocked automated clicks at checkout, but that choice likely started as an anti-bot rule from the security team. Marketing found out when conversion numbers changed. This pattern repeats in many companies.
A second common mistake is thinking that the problem only concerns the checkout. AI agents collect data throughout the entire browsing experience: if your catalog is not structured, the agent will exclude you during the comparison phase, even before reaching the cart. The topic of machine-readable catalogs is the same one we addressed when talking about AI agent governance in media buying .
| Error | Consequence | How to avoid it |
|---|---|---|
| Leave the decision to the IT/security team | Marketing discovers the block when data drops | Create a cross-functional team (marketing + IT + legal) with a shared policy |
| Catalog not structured for machines | The agent excludes the brand during the comparison phase | Implement schema markup, structured feeds, real-time updated data |
| Total agent blocking without an alternative | High-intent customers go to competitors | Allow verified agents (Visa Trusted Agent Protocol, Mastercard Agent Pay) |
| Do not adapt analytics tracking | Orders without history: impossible to optimize campaigns | Switching to server-to-server events for agent-based checkout |
| Ignoring the pre-agent moment | The brand is not in the customer's consideration set | Investing in educational content and brand awareness before the customer delegates to the agent |
How to start concretely: the steps to adapt the strategy?
It starts with a quick audit of the current situation: what do agents see when they visit your site, and what happens to your data when they complete a purchase?
No need to wait for the problem to show up in the numbers. Just three initial checks are enough to see where you stand.
- 1 · Marketing + IT Catalog readability audit Verify that prices, availability, return policies, and technical specifications are in a structured format (schema markup, updated feeds). An AI agent reads this information like an advanced search engine.
- 2 · Marketing + Analytics Map tracking blind spots Identify which events aren't captured when checkout happens via an agent. Set up server-to-server events as a backup to pixels.
- 3 · Legal + Marketing + IT Define the agent access policy Decide which agents to admit, and with what level of autonomy. Evaluate integration with Visa Trusted Agent Protocol or Mastercard Agent Pay requirements.
- 4 · Marketing Take charge of the pre-agent moment Identify the channels where your customers form their preferences before delegating to the agent: educational content, reviews, comparison sites, social media. This is the new brand battlefield.
- 5 · Marketing Monitor specific KPIs for AI traffic Separate traffic coming from AI agents in your analytics system. Measure conversion rate, revenue per visit, and return rate separately.
How to tell if my strategy is working with AI agents?
The signals to monitor are different from the classic ones: it's not enough to look at total traffic, you need to separate AI traffic and measure its quality.
| Indicator | What it measures | Positive signal |
|---|---|---|
| Traffic from AI sources (segmented) | How many sessions come from AI agents or tools | Steady growth, in line with market trends (+127% YoY in August 2026 for US retail) |
| Conversion rate AI vs non-AI traffic | Whether AI visitors buy more or less than others | AI traffic converts better (benchmark: +42% in March 2026, Adobe) |
| Revenue per visit from AI traffic | Average value generated by each AI session | Higher than non-AI traffic (benchmark: +37%, Adobe) |
| Orders without a tracked session | Checkouts completed without navigation data | Number stable or decreasing after implementing server-to-server tracking |
| Presence in agent comparison sets | The brand appears when an agent compares products in your sector | Manual verification with tests on Muse, Dots, or other available agents |
| Post-purchase return rate via agent | Customers who bought via agent return directly | Return rate similar to or higher than customers who bought manually |
So, is it worth adapting now or can you wait?
It is worth it if you have an active e-commerce, an online product catalog, or a digital purchasing process. With each passing week, a growing share of your most motivated customers uses an agent. Adapting now costs less than catching up later.
A middle ground You sell mainly offline but have a digital presence. You can wait a few more months, but use this time to structure your catalog and define your policy: the cost of doing it in an emergency is much higher.
Not yet if You sell exclusively through direct offline channels, without a digital catalog and without an online purchasing process. The impact is still indirect, but monitor how your competitors are moving.
How we see it at SHM
For us, the arrival of consumer AI agents is not a threat to the brand-customer relationship: it is a test of the quality of that relationship. An AI agent chooses based on the data it finds. If your catalog is well-structured, your policies are clear, and your brand is already in the customer's mind before they delegate to the agent, you have already done the right work.
We believe AI should be integrated to free companies from routine and give back time — not to replace the ingenuity of those who build customer relationships. Consumer AI agents shift the battlefield, they don't eliminate the game. Those who own the pre-agent moment with quality content and those who structure data so machines can read it correctly will have a real advantage.
At SHM Studio we are working on this front with our clients: from the Strategic AI integration in marketing processes, at the SEO optimized to be found by AI agents , up to the revision of the paid campaigns in an ecosystem where the nature of traffic is changing. Think of a manufacturing company with a complex technical catalog: structuring that data to make it readable by AI agents is exactly the kind of work we have already done in international SEO, and which today is becoming urgent even for those who have never thought of agents as interlocutors.
You can explore the entire topic area in our section Marketing and ADV performance or contact us directly to understand where to start.
You can find more insights on how AI agents are changing campaigns in the article on agentic buying and automated media buying and in the one about creator marketing , where the fragmentation of attention follows a similar logic.
Frequently Asked Questions
A consumer AI agent is software that acts autonomously on behalf of a person: it browses sites, compares products, fills out forms, and completes purchases without the user having to do anything. A chatbot answers questions but does not take action. Meta's Muse, OpenAI's Dots, and Anthropic's Cowork are examples of consumer AI agents already available to the public.
Yes, the order arrives and is registered to the merchant. The problem is not the order itself, but the loss of behavioral data: the path that led the customer to choose that product is not tracked by standard pixels. To recover at least the checkout data, you need to implement server-to-server tracking.
Shopify enabled agent-based checkout by default for all eligible merchants in September 2026. If you want to disable it, you must do so manually from the admin settings: Sales channels > Agentic > Meta channel > Direct checkout OFF. The choice of whether or not to participate should be made consciously, not left to the default.
You can run a manual test using one of the available agents (Muse, Dots, or others) and check if it can find your products, read prices, and return policies. On the technical side, check for schema markup on products and the availability of updated structured feeds. A technical SEO audit often reveals the same gaps that hinder AI agents.
The legal landscape has become complicated. In August 2026, the US Ninth Circuit ruled that the user is the natural person and the agent acts on their behalf, limiting the use of anti-hacking laws to block agents. Terms of service remain a valid tool, but their enforceability depends on the context. The most viable path is to allow verified agents and block only unidentified ones.
Three low-cost actions: verify that your catalog has updated schema markup, do a manual test with an AI agent to see how your site looks, and check your checkout settings if you use Shopify. These three checks take a few hours and already give you a clear picture of where you stand.
The real competitive advantage isn't blocking AI agents: it's being the brand that the customer wants their agent to find.
Related articles
Discover more articles exploring similar topics, selected to offer you a more complete and stimulating perspective. Each piece of content is carefully chosen to enrich your experience.
