# 7 Universal Commerce Protocol Challenges (and the Opportunities They Create)

*By Joshua Cogar · 2026-09-30 · Updated 2026-10-07 · AI Search*

> UCP turns your storefront into an API. The seven challenges it creates for businesses, the opportunity inside each one, and what to change on your website now.

*This article is based on my talk at the Las Vegas AI Meetup at Tech Alley Vegas in August 2026. The full recording and transcript are above.*

**The Universal Commerce Protocol (UCP) creates seven problems for businesses that sell online: visibility, brand, price and margin, the customer relationship, operations, channel control, and measurement.** Each one comes from the same shift. AI agents now shop for people, reading product feeds and checking out through an API, so the shopper never sees your storefront. Every one of those challenges also opens an opportunity for businesses that act early, because most of their competitors haven't started.

I've run a marketing agency for about eight years, working on go-to-market strategy and ecommerce. I'll make what may sound like an unreasonable claim: this is the biggest change to the internet since the internet itself. Most businesses have no idea how much their market will change over the next two to five years.

If you want the definitional background first (who built UCP, how it works, and the Shopify enrollment details), start with [How the New Universal Commerce Protocol Affects Marketing and Website Requirements](/blog/universal-commerce-protocol-marketing/). This article is about what UCP breaks and what you should do about it.

## The Shift in One Paragraph

Before agents ever reach a checkout, AI has already hollowed out discovery. This is zero-click search. In the first four months of 2026, [68% of U.S. Google searches ended without a click](https://sparktoro.com/blog/in-2026-less-than-one-third-of-google-searches-still-send-a-click/), up from about 60% in 2024, according to SparkToro's analysis of Similarweb data. Fewer than one in three searches now sends a click to the open web. The top of the funnel isn't necessarily disappearing, but it's moving out of your control. I'm a marketer, and most of the tactics I've used in that awareness layer are going away.

![Slide: Zero-click is eating awareness. Fewer than one in three Google searches sends a click, demand capture breaks first, with questions to ask about where the awareness budget goes](/images/ucp-talk/ucp-slide-02.jpg)

UCP takes that one step further. The [Universal Commerce Protocol](https://ucp.dev/) is an open, free standard, [co-developed by Google and Shopify](https://ucp.dev/) along with retailers and payment companies, for how AI agents discover products and complete purchases. A shopper asks ChatGPT, Gemini, or Claude for "a men's size 10 zero-plastic running shoe for wide feet under $150." The agent searches product directories it already has access to, not the open web. It narrows the options, then reads the finalists' product feeds and checks out over UCP.

![Slide: Two very different shopping journeys. The journey businesses built for, from Google search to storefront checkout and retargeting, versus the arriving journey where an agent reads the product feed and checks out over UCP rails](/images/ucp-talk/ucp-slide-03.jpg)

**Your storefront is becoming an API.** The shopper's loyalty moves to their agent, not to a business they never visited.

This isn't speculative. Generative AI traffic to U.S. retail sites [rose 1,300%](https://news.adobe.com/news/2025/1/adi-pr-full-season-recap) during the 2024 holiday season compared with the year before, according to Adobe. By March 2026, AI shoppers [converted 42% better](https://techcrunch.com/2026/04/16/ai-traffic-to-us-retailers-rose-393-in-q1-and-its-boosting-their-revenue-too/) than other traffic, after converting 38% worse a year earlier. On Shopify, orders from AI-powered searches grew [nearly 13x year over year](https://www.shopify.com/blog/how-agentic-commerce-works) in Q1 2026. And starting the week of March 24, 2026, Shopify made eligible merchants' products [discoverable in ChatGPT by default](https://www.shopify.com/news/agentic-commerce-momentum) through Agentic Storefronts. Many owners don't know it happened.

![Slide: The 2026 numbers. AI-referred traffic to US retail sites up 1,324%, 42% better conversion from AI-referred shoppers, 13x growth in AI-driven Shopify orders, and stores auto-enrolled in March 2026](/images/ucp-talk/ucp-slide-04.jpg)

*The slide shows 1,324%. Adobe's published figure for the 2024 holiday season is 1,300%, and that's the number to cite.*

One more point before the challenges. **UCP is plumbing, not a marketplace.** You stay the merchant of record, you keep the customer data, and your existing store fees don't change. It carries the same products and prices you have today, so nothing about the protocol makes you win. **Your data makes you win.**

![Slide: UCP, new rails, same store. Buyer agents such as Gemini, Copilot, ChatGPT and Claude connect through the open UCP standard to your store, where you stay merchant of record and keep customer data](/images/ucp-talk/ucp-slide-05.jpg)

## Challenge 1: Visibility. Machines Read Feeds, Not Homepages

![Slide: Challenge 1 of 7, visibility. Machines read feeds, not homepages; feed quality is the new SEO, with questions to ask such as whether an agent will even find your products](/images/ucp-talk/ucp-slide-06.jpg)

**The challenge.** Most business owners haven't absorbed this yet. As agentic commerce grows, what your website *looks like* counts for less and less. You still want good UX, since nobody wants a page that looks machine-built. But I work with business owners constantly, and they obsess over aesthetics. What counts now is infrastructure. **SEO is becoming a function of infrastructure.**

SEO isn't dead. [SEO, AEO, and GEO](/blog/what-is-generative-engine-optimization/) are layers working toward the same goal. If your traditional SEO is weak, your AEO will be too. Be skeptical of anyone promising to "rank you in the LLMs," because the data doesn't support it. The most useful metric available is **share of voice**: how often your business appears among the options an AI returns, tracked as a trend over time.

**The opportunity.** Feed quality is the new SEO, and most feeds are bad. Budgets are shifting toward answer engine optimization, meaning getting LLMs to [cite your business](/blog/how-to-get-cited-by-llms/) in their answers. A business with clean, consistent, machine-readable data can show up in AI answers even if its pages don't rank on keywords.

**What to change on your website:**

- **Remove contradictions.** Google compares the prices in your product feed with your landing page and structured data, and products whose data doesn't match [may be disapproved](https://support.google.com/merchants/answer/12159029?hl=en). The same logic applies to your name, address, and phone number (NAP) across [every listing on the web](/blog/top-local-citation-sources/). LLMs look at everything, and consistency is how they judge credibility. We're moving into a trust economy.
- **Add rich Q&A-style attributes.** These are being added specifically for agents to read.
- **Stop hiding answers behind JavaScript.** Many WordPress and Elementor sites show FAQ answers in accordions that expand with JavaScript. If the answer only appears after a script runs, crawlers and agents often can't see it. Put the text in the HTML.
- **Look at fan-out queries.** Google's AI Mode uses a [query fan-out technique](https://blog.google/products-and-platforms/products/search/google-search-ai-mode-update/), breaking one question into subtopics and running many searches at once. You can research what those look like and build content and data around them. More on this in [How to Optimize for LLM Search](/blog/how-to-optimize-for-llm-search/).

## Challenge 2: Brand. The Interchangeable Inputs Problem

![Slide: Challenge 2 of 7, brand. The interchangeable inputs problem, with brand moving to face-to-face presence and questions about your moat when a machine does the choosing](/images/ucp-talk/ucp-slide-07.jpg)

**The challenge.** When an agent squeezes the whole buyer journey into one recommendation, brand storytelling and merchandising disappear from the decision. How do you have a brand to a machine?

The questions this raises:

- What's my moat when a machine does the choosing?
- Should I build a brand agent, or an app inside the major AI services?
- How do I stay a brand and not a row in a comparison table?

Most business owners don't know to ask these yet. I had a LinkedIn exchange with a senior leader at HubSpot who thought the idea that the top of the funnel is shrinking was ridiculous. Even large brands don't see this coming.

**The opportunity.** Brand building moves offline, into places a machine can't flatten. Expect more face-to-face events and sponsorships, like a shoe company sponsoring a race. The goal is to have shoppers tell their own agents, "Always check this brand first." People can write that preference into their AI instructions, and brands are already working to be first in line.

**What to change on your website:** Put your differentiators into structured data, not just marketing copy. If your value lives in materials, warranty, bundles, or shipping speed, the agent can only weigh it if it's in the data.

## Challenge 3: Price and Margin. Comparison Becomes Perfect and Instant

![Slide: Challenge 3 of 7, price and margin. Every agent query is a price-comparison event, with questions about competing on price and what visibility will cost](/images/ucp-talk/ucp-slide-08.jpg)

**The challenge.** LLMs can query many sources at once, so poor data hurts you fast. If your feed is weak, or your price and margin are out of line, the agent skips you.

- **Every agent query is a price comparison** across every participating merchant.
- A **universal cart** shows alternatives from different retailers side by side.
- Instant comparison risks **training buyers to wait for discounts.**

Then there's paid placement. AI platforms are starting to sell ads, and OpenAI has [begun testing ads in ChatGPT](https://openai.com/index/testing-ads-in-chatgpt/). It's the Google pattern repeating: the company that controls discovery also sells ads inside it, which is a real conflict of interest. Ad costs are already climbing: Google search cost per click rose about [39% from 2020 to mid-2024](https://www.karooya.com/blog/cpc-growth-trends-google-paid-search-shopping-ads-text-ads-2020-2024/). Agentic commerce could have been an escape from that cycle, but it may simply rebuild it.

**The opportunity.** Perfect comparison rewards merchants who compete on complete information, not just price. A full feed that states total value (loyalty points, shipping, returns, warranty) gives the agent something to compare besides the sticker price.

**What to change on your website:** Keep price and availability accurate in real time and consistent everywhere they appear. Include loyalty and member pricing in the feed so it shows up in the comparison.

## Challenge 4: The Customer Relationship. Who Owns the Shopper?

![Slide: Challenge 4 of 7, the customer relationship. The conversation lives with the platform, with questions about repeat business, loyalty programs and replacing the email list](/images/ucp-talk/ucp-slide-09.jpg)

**The challenge.** You're still the merchant of record, and you keep the order data. But **the conversation lives with the platform.** You may know an order came from ChatGPT, but not why the shopper was buying or what else they compared. That's the top-of-funnel problem again, now happening at the moment of purchase.

Repeat purchases get harder, because you have less insight into intent. One attendee at the talk, who works for a large manufacturer and retailer, pointed out another loss: **average order value.** On your site, a shopper sees "customers also bought," recommendations, and checkout upsells. An agent sent to buy one item buys one item.

**The opportunity.** Everything after the order is still yours, and it counts more than ever. **Reviews become a primary ranking input.** Businesses with great service earn great reviews, and the AI agent reads those reviews when it decides between you and a competitor. That decision is very black and white.

**What to change on your website:**

- **Build loyalty into your product feed.** If a shopper earns points with you and not with a comparable merchant, that can tip the agent's decision. UCP supports this.
- **Treat packaging, shipping, customer service, and returns as your brand.** Control what you can control, and do it excellently.
- **Build up your reviews** on purpose and consistently.

## Challenge 5: Operations. The Table Stakes Went Up

![Slide: Challenge 5 of 7, operations. Real-time inventory and price accuracy are mandatory, several categories are excluded from native checkout, and you lose the chance to compete](/images/ucp-talk/ucp-slide-10.jpg)

**The challenge.** AI agents enforce the standards shoppers always *hoped* for. In an agent world there's no grace.

- **Stale price or stock data gets you excluded.**
- **Some categories are excluded from native checkout**, including subscriptions, pre-orders, and personalized goods. If you sell subscriptions, UCP is a problem for you today.
- **You don't lose a sale; you lose the chance to compete.** Operational slips cost you future queries, not just one order.

Dynamic pricing came up in the Q&A. UCP accommodates it, so prices served through the API can change based on factors like delivery urgency. The bigger worry is manipulation. I expect **sponsored APIs**, where a platform that can see your session steers you toward a vendor who has paid for preference. Taken to a draconian extreme, a platform with access to credit or financial data could charge more because it knows you can afford it.

**The opportunity.** Operational excellence becomes a visibility advantage. Businesses that keep their data clean will be included in more queries while sloppier competitors drop out.

**What to change on your website:** Sync inventory and pricing in real time. Make your shipping, returns, privacy, and terms policies complete and easy for machines to read. Unanswered questions give an agent a reason to skip you.

## Challenge 6: Channel Control. A New Management Burden

**The challenge.** There isn't one agent channel. There are many: Gemini, ChatGPT, Copilot, and more.

- They run on **two protocols: UCP and the [Agentic Commerce Protocol](https://www.agenticcommerce.dev/) (ACP)**, each with different fees.
- **Fees are inconsistent, and they keep changing.** In January 2026, Shopify merchants selling through ChatGPT checkout were [reportedly set to pay OpenAI a 4% fee](https://www.pymnts.com/news/ecommerce/2026/shopify-merchants-to-pay-4percent-fee-on-sales-made-through-chatgpt-checkout/). Then OpenAI [ended Instant Checkout](https://www.modernretail.co/technology/what-went-wrong-with-chatgpts-instant-checkout/) in March 2026 and moved purchases into merchants' own apps inside ChatGPT. Some platforms charge a commission, some don't, and some haven't published terms.
- **Shopify enrolled stores by default.** Most owners never read the new terms, so they don't know what changed.
- **Per-channel controls exist**, including one for smaller and emerging AI agents.
- **Unverified bots hit your store too.** Agent identity standards are still maturing.

**The opportunity.** Few merchants are managing these channels on purpose, so anyone who does gets an edge by default.

**What to change on your website:** Answer three questions. Which agent services will you enable? Do you trust unknown or unverified agents? Who on your team owns this channel? On Shopify, the settings are under Settings, then Sales Channels.

## Challenge 7: Measurement. Flying Blind

![Slide: Challenge 7 of 7, measurement. Agentic orders generate zero page views and last-click attribution collapses, with questions about replacing the conversion funnel](/images/ucp-talk/ucp-slide-12.jpg)

**The challenge.** The analytics stack was built on **sessions, clicks, and funnels. Agentic orders produce none of them.** Last-click attribution breaks down when the "click" is an API call from an agent. Attribution is the backbone of conversion rate optimization, and it's hard to A/B test an API call. As my co-founder Tim May put it during the Q&A, what's missing is attribution. Without session data, you lose the journey you'd use to market to that customer again.

**The opportunity.** Businesses that set up per-channel measurement now will know their real margins when fees and ad placements settle. Everyone else will be guessing.

**What to change on your website:** Track orders and average order value by agent channel. Start monitoring share of voice across AI answers as your visibility metric, instead of relying on sessions alone. We help with this in our [analytics and optimization](/services/analytics-optimization/) work.

## The Biggest Challenge: Most Service Businesses Are Still Left Out

Service work is most of the U.S. economy. Service-providing industries employ about [137 million people](https://www.bls.gov/iag/tgs/iag07.htm) (August 2026), compared with about [21.6 million in goods-producing industries](https://www.bls.gov/iag/tgs/iag06.htm), according to the Bureau of Labor Statistics. They also account for about 10.6 million of the roughly 12 million private business establishments BLS counted in early 2026. **UCP still doesn't cover most quote-based local service businesses.** Construction, clinics, salons, agencies, financial services, real estate, trades: anything priced by estimate, without a GTIN or a firm list price, has no path yet.

UCP has started to move past retail. At Google Marketing Live on May 20, 2026, Google said it would [bring UCP to hotel booking and local food delivery](https://blog.google/products-and-platforms/products/shopping/shopping-updates-google-marketing-live), and [ucp.dev](https://ucp.dev/) now lists a draft Lodging specification, with a Food specification marked as coming soon. Both are categories with fixed, published prices. A plumber's quote or a remodel estimate still has nowhere to go.

Tim described the scope precisely in the Q&A. UCP targets products **where there's no negotiation**, meaning retail and fast-moving goods. Anything with individual pricing or elaborate bundling falls outside it. UCP stops at: *here are the products and prices, pick one, and I'll hand you to the merchant.*

That's an enormous gap, and a large opportunity. Travel shows why direct connections pay off. Hotels typically pay online travel agencies [15% to 30% or more in commission](https://www.cloudbeds.com/articles/online-travel-agencies/commissions/) per booking. I used to work at Carnival Cruise Line, and companies like that judge acquisition cost against a customer's lifetime value, so they'll spend heavily up front to win a customer. If acquiring a customer dropped from, say, $800 to $80 because an agent could talk directly to every hotel in a destination, going direct would be a no-brainer. The aggregator's reason to exist shrinks.

That gap is why Tim and I are building **Scutara**, an agentic commerce marketplace that matches buyer agents with seller agents. It's designed for discoverability, anonymity, security, and neutrality where UCP stops. If you run a service business, [Agentic Commerce for Service Businesses](/blog/agentic-commerce-for-service-businesses/) walks through what you can do today.

![Diagram: Scutara, an agentic commerce marketplace connecting anonymous buyer agents with seller agents, built around discoverability, anonymity, security and neutrality](/images/ucp-talk/ucp-slide-14.jpg)

## Who Controls the Ranking?

The most technical question of the night was whether implementing UCP produces measurable lift if the agents don't use those signals for retrieval. Tim's answer goes to the core of it: **the indexes are owned by the big players.**

- Google has billions of products indexed. When Gemini or any other agent surface sends a request, **Google does the matching and the ranking.**
- In Tim's view, there are three major product indexes today: **Google, ChatGPT, and Shopify.** He doesn't see Claude as one of them yet, and he notes that Google queries both its own index and Shopify's.
- With a few hundred sellers of a product, ranking barely changes anything. With 10,000, pushing every candidate through an LLM becomes a token nightmare. Index owners will rank results to control cost, and then realize they can sell placement.

**Whoever owns the index owns the ranking, and whoever owns the ranking has a way to make money from it.** How to move up will depend on each index owner's strategy, and those strategies are still being written. At least now you know who the players are. And yes, Google is in charge of your destiny again.

## What About Local Businesses and Marketplaces?

**Local search.** LLMs draw on many sources, not just one directory like Yelp, so the outcome for local search is still uncertain. Google Maps also now lets people interact with a business without leaving Maps, so even "ranking in the map pack" may not deliver what it used to. A strong [Google Business Profile](/blog/optimize-your-google-business-profile/) and [local AEO](/blog/local-geo-and-aeo/) are still worth the work.

**Marketplaces like Etsy.** Ask why you're there. If it's only for visibility, feeding your product data into agent infrastructure may eventually handle discovery. But marketplaces still add AEO signal. Being on Etsy, on social media, in forum discussions, and on your own site, all saying the same thing, tells LLMs you are who you say you are and your product does what it claims.

**New products and small shops.** Keep doing SEO. But with a clean, accessible data feed, you can show up in LLM answers even if your site doesn't rank. This is where a small business can compete with bigger budgets: a strong website back end can beat spending.

## The Good News: Why UCP Favors the Prepared

![Slide: Clean data levels the playing field. Agents don't care how big you are, the traffic is better, Shopify does the heavy lifting, and most competitors are asleep](/images/ucp-talk/ucp-slide-15.jpg)

- **Agents don't care how big you are.** A perfect feed from a two-person shop can outrank a sloppy big-box listing, and plenty of product data out there is sloppy.
- **Shopify does the heavy lifting.** The protocol work is already done for Shopify merchants.
- **The traffic is better.** AI-referred visitors convert at higher rates.
- **Most competitors are asleep.** In May and June 2026 our prospecting scanner checked about 460 small business websites. The median overall marketing score was 38 out of 100, and AI search readiness was the weakest of five categories for 83% of them.

*Correction: in the talk I said we'd scanned "close to 800" sites and that most scored about 38 for AI readiness. The tool held 764 business records, including unscanned and archived ones, but about 460 had a completed score, and 38 is the median overall marketing score across five categories, not an AI readiness score. The corrected figures are above and in the table below. These are businesses our prospecting tool pulled, not a random sample of all websites.*

<table>
  <thead>
    <tr>
      <th>Prospecting Scanner Result (May 6 to June 9, 2026)</th>
      <th>Figure</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td>Small business websites with a completed score</td>
      <td>About 460</td>
    </tr>
    <tr>
      <td>Median overall marketing score (five categories)</td>
      <td>38 out of 100</td>
    </tr>
    <tr>
      <td>Median AI search readiness category score</td>
      <td>4 out of 25</td>
    </tr>
    <tr>
      <td>Share where AI search readiness was the weakest category</td>
      <td>83%</td>
    </tr>
  </tbody>
</table>

**A warning about one-click fixes.** Cloudflare recently released an agent readiness scan that generates a prompt to apply its recommended settings automatically. I ran it on my own site, which I'd just rebuilt for AI, and scored zero out of seven in one section, because it grades many things that don't apply to most websites. Clicking its suggested fix would have turned on bot blocking, blocked every AI agent, and tanked my SEO. Read my full [Cloudflare Agent Readiness Scan review](/blog/cloudflare-agent-readiness-scan-review/) before you click anything.

## The Three-Step Playbook for Your Website

![Slide: A three-step playbook. Get your feed agent-ready, take control of the channels, and own the relationship anyway](/images/ucp-talk/ucp-slide-16.jpg)

**1. Get your feed agent-ready.** Complete GTINs, full descriptions, accurate real-time stock, machine-readable Q&A, and fast page loads. Start by [scanning your site's AI readiness](/ai-readiness-tool/) for free.

**2. Take control of your channels.** Decide which agent services you'll work with, and put complete returns, shipping, privacy, and terms policies in your data.

**3. Own the relationship anyway.** Connect loyalty to customer identity, make fulfillment and returns part of your brand, build your reviews, and consider a brand agent that lives where your buyers are.

The full technical version is in our free checklist:

[[resource:agentic-commerce-checklist]]

**The storefront isn't dying. It's getting a new front door, and that will change how business gets done.**

## Universal Commerce Protocol Challenges FAQ

### What are the biggest challenges of the Universal Commerce Protocol for businesses?

There are seven: visibility (agents read feeds, not homepages), brand (products become interchangeable), price (comparison is instant), the customer relationship (the conversation stays with the platform), operations (bad data gets you excluded), channel control (multiple protocols and fees), and measurement (no sessions or clicks to track). For many businesses, the largest gap is that UCP still doesn't cover most quote-based services. Hotels and food delivery are being added, but estimates and custom quotes aren't.

### Is UCP a marketplace?

No. UCP is plumbing. It's an open protocol that connects AI agents to your store. You stay the merchant of record and keep your customer data, and your products and prices are the same as on your site. It doesn't make you win by itself. Your data does.

### Does UCP replace SEO?

No. Strong traditional SEO is still the foundation for showing up in AI answers. SEO is becoming a function of infrastructure: clean product feeds, consistent data, and content machines can read without running JavaScript.

### Can a small business compete with big brands in agentic commerce?

Yes. Agents rank on data quality, not company size. A complete, accurate feed from a small shop can outrank a poorly maintained listing from a major retailer, and most businesses haven't prepared yet.

### Will AI agents buy things without the shopper's approval?

Consumer tools like ChatGPT and Gemini generally stop short of completing a purchase without approval. Fully autonomous buying requires a custom buyer agent that the person has given permission and a budget.

### Who decides how products rank in AI shopping results?

Whoever owns the product index. In my co-founder Tim May's view, that means Google, OpenAI, and Shopify today. As the number of competing sellers grows, those index owners will rank results to control costs, and paid placement will likely follow.

### How do I measure sales that come from AI agents?

Traditional session-based analytics can't see them, because agent orders produce no clicks or funnels. Track orders and average order value by agent channel, and monitor your share of voice across AI answers over time.

### What should I fix on my website first?

Start with your data. Make your product feed complete and accurate, remove contradictions between your feed, site markup, and business listings, and put FAQ and policy content in plain HTML. Run a free [AI readiness scan](/ai-readiness-tool/) to see where you stand.

Advisors Marketing is a digital marketing agency in Henderson, NV. We help Las Vegas businesses and national brands get found through [SEO](/services/seo/), AEO, content marketing, digital ads, [website development](/services/website-design/), and AI integration.

---

Canonical page: https://seomarketingadvisors.com/blog/universal-commerce-protocol-challenges/
Site index: https://seomarketingadvisors.com/llms.txt · Agent directory: https://seomarketingadvisors.com/agents.json