Ecommerce is entering a caller shape – 1 where transactions tin hap wholly off-site, making website visits optional, if not redundant. If your ecommerce and SEO teams are still playing the aged game, location could beryllium problem ahead.
Of course, I’m talking astir agentic commerce, where the full customer journey – from find to checkout – tin hap within AI. And boy do things move quickly!
In September 2025, OpenAI and Stripe announced their Agentic Commerce Protocol (ACP), allowing customers to not only investigation and decide which merchandise to buy, but also complete single-item transactions right there in the chat with Instant Checkout.
In January, Google announced the Universal Commerce Protocol (UCP), “a caller unfastened modular for agentic commerce that useful crossed the full shopping journey – from find and buying to post-purchase support.”
And past successful March, after only six months, OpenAI backed distant from Instant Checkout, claiming that the first type “did not connection the level of elasticity that we aspire to provide.” Instead, ACP is placing much attraction connected merchandise find – driving valuable visibility for your marque and products wrong ChatGPT – while retailers tin determine if and really they merge their ain checkout experiences.
In the aforesaid month, Google rolled out a UCP update that included a bunch of caller checkout and catalog capabilities, including shopping carts.
So, if there’s 1 point we’ve learned already, you’d amended enactment alert.
My honorable prediction is that, sometime successful the first half of adjacent year, we’ll start proceeding from immoderate retailers that they execute a amended return connected finance from agentic commerce than, say, paid social. The difference, and the competitory opportunity, will go clear.
So, I decided to do immoderate investigation to spot conscionable really fresh immoderate of the top-performing retailers are for agentic commerce. And it appears a batch of them whitethorn beryllium highly unprepared.
But earlier we get to the investigation data, it’s important to understand why agentic commerce presents a very different challenge to retailers.
The End Of Traffic-First Ecommerce
We’ve known for a while that AI is gradually coopting much of the customer journey, particularly in the discovery, research and information stages.
The superior situation for retailers has been getting into that all-important information group by ensuring their marque and products are recommended astatine cardinal points successful applicable AI conversations. Get that right, and hopefully, erstwhile the customer makes a purchasing decision, it’s your website they click done to.
For 3 decades now, website postulation has been astatine the bosom of ecommerce. Online retailers have dedicated so overmuch time, effort and money to refining the creation of customer attraction – done SEO and content, done email and societal media, done typical offers and loyalty programs – thing that mightiness lead much group to their merchandise pages.
Yes, there’s more to ecommerce than conscionable traffic. There are conversion rates and shopping cart optimizations, personalization and cross-selling opportunities and … well, it’s a agelong list. But these each trust on traffic. A imaginable customer has to land connected your tract earlier immoderate of these different factors travel into play.
However, erstwhile the customer doesn’t request to sojourn your website – erstwhile they tin complete the full transaction wrong an AI speech – each erstwhile ecommerce strategies statesman to unravel.
On the aforesaid time arsenic Google’s UCP motorboat announcement, Target put retired its own press release describing what the new agentic shopping experience would look like.
“A impermanent starts a speech successful AI Mode aliases the Gemini app – for example, ‘I’m getting into moving retired and want to beryllium some comfortable and stylish astatine the gym. Help maine find tiny and affordable floral leggings, successful a ray color.’ They’d then spot different options to see and purchase without having to time off the chat.”
So, really do you spot that your products are included among those options?
The Mechanics Of Agentic Commerce
Agentic commerce doesn’t activity for illustration integrated search. This isn’t astir rankings wherever a little position is still visible to customers who scroll acold enough. You’re either successful aliases out.
It also doesn’t work in the aforesaid measurement as AI citations aliases mentions. All that work your SEO team might already be doing to increase your brand’s visibility successful AI won’t help you here.
Rather than crawling the customer-facing content connected your merchandise pages to extract the basal information, both protocols instead draw data from the merchant feed or product feed and the on-page schema.
Therefore, agentic commerce isn’t a contented optimization problem truthful overmuch arsenic a information plumbing one. The AI agents are simply conduits, piping earthy merchandise information 1 measurement for AI to relay to the customer and feeding transaction information backmost the different measurement erstwhile the acquisition is complete.
This is important for 2 reasons.
Firstly, it means both protocols retain the retailer arsenic the merchant of record. OpenAI and Google aren’t reselling your wares successful the aforesaid measurement that, say, the Apple Store does.
Secondly, for agentic transactions to beryllium successful, some protocols need detailed, accurate, and up-to-the-minute information that – arsenic you’ll spot – a batch of retailers don’t presently see successful their feeds.
Right now, astir retailers I travel crossed treat their merchandise aliases merchant feeds arsenic a broadside project; a convenient measurement to update move advertisement campaigns pinch basal merchandise data. But the logic advertisement campaigns require truthful small accusation – a merchandise name, an image, possibly a typical connection – is because they’re still astir driving postulation backmost to the merchandise page. The transaction still happens connected the website.
If you want agentic commerce to complete that transaction connected your behalf, it’s going to need a batch much information.
The 3 Schema Fields You Cannot Miss
When Google Gemini evaluates whether to urge a merchandise aliases not, it isn’t only concerned with basic transactional metadata. Google’s UCP documentation lists three other important signals:
- priceValidUntil – confirming the pricing accusation is still current.
- shippingDetails.deliveryTime – providing shipping information the customer will expect to see.
- hasMerchantReturnPolicy.merchantReturnDays – supplying the supplier pinch accusation connected return windows, etc. to springiness the customer immoderate purchaser protection.
If any of those fields are absent or incomplete, Gemini won’t simply rank the merchandise little or less prominently. Your product won’t be included. Period.
Speed matters, too. AI doesn’t want to discarded valuable clip and CPU trying to fig things out. If an agentic supplier calls your API and the consequence is slow, it’ll beryllium little apt to usage your provender adjacent time.
The aforesaid goes for old aliases inaccurate information that results in grounded transactions, specified as an item turning retired to beryllium unavailable moreover though the provender claimed it was successful stock.
All these issues whitethorn origin the AI to position your feed as unreliable, leading it to urge products from brands pinch more trustworthy feeds successful future.
The Research Findings
So, now we cognize what’s required. How are immoderate of the apical retailers doing?
We identified 207 top-traffic merchandise item pages (PDPs) from 29 different retailers according to their integrated postulation measurement successful Ahrefs. In fetching information from these pages, 52 were not readable (returning 403, 404, aliases position 0 errors), while 14 others turned retired not to beryllium PDPs (misidentified class aliases contented pages).
This near america pinch 141 PDPs to audit – all currently winning the old-school SEO game.
We past scored each page against a 10-point UCP-readiness rubric based connected Google’s UCP documentation. (The points successful bold bespeak the cardinal signals mentioned above.)
- Has Product / ProductGroup / IndividualProduct JSON-LD connected the page.
- Has GTIN identifier (gtin / gtin8 / gtin12 / gtin13 / gtin14).
- Has MPN aliases chopped SKU.
- Has brand object (name aliases @id).
- Has value + priceCurrency on astatine slightest 1 offer.
- Has readiness (InStock / OutOfStock / etc.).
- Has priceValidUntil.
- Has shippingDetails.deliveryTime.
- Has hasMerchantReturnPolicy.merchantReturnDays.
- Has aggregateRating or reappraisal array.
The findings are stark.
1. The Basics Are Fine
Let’s start pinch the bully news. Virtually all the audited PDPs screen the basics.
- 99% transportation value and availability.
- 99% transportation MPN aliases SKU.
- 96% transportation brand.
Plus, all 141 of the PDPs look to person afloat adopted the merchandise schema recommended by Google since 2014.
2. 70% Miss All Three Of The Most Important Attributes
However, the 3 cardinal fields I mentioned earlier were only added to Schema.org between 2020 and 2021. And this is wherever things statesman to autumn apart.
- Only 18% carry priceValidUntil.
- Only 13% carry shippingDetails.deliveryTime.
- Only 11% carry merchantReturnDays.
It seems not everyone has kept their schema templates up to date.
3. 65% Don’t Include A GTIN Field
Unlike the three fields above, the deficiency of a Global Trade Item Number (GTIN) won’t necessarily forestall AI from utilizing your feed, but it will handicap the agent’s expertise to comparison your offering pinch different retailers.
The GTIN is really Google’s UCP knows that Product X connected your website is the aforesaid beingness point arsenic Product X connected a competitor’s site. Without a GTIN, the AI supplier won’t admit your merchandise arsenic the same, treating it alternatively arsenic a unsocial item. This intends it can’t do things for illustration comparison pricing. If a customer asks, “Which retailer has the champion value connected Product X?” the reply won’t beryllium you.
4. The Issue Is Configuration, Not Which Platform You Use
Only three of the audited brands scored an mean of 8 aliases higher crossed their apical PDPs:
| Brand | Platform | Avg. | Top Failing Fields |
| Uplift Desk | Custom | 9.0 | GTIN |
| Carbon38 | Shopify | 8.6 | merchantReturnDays |
| Sigma Beauty | Shopify | 8.0 | priceValidUntil, shippingDeliveryTime |
It’s worthy noting that nary of the top-scoring brands would person needed to move platforms aliases rebuild to make themselves agentic commerce ready. All they did was reconfigure their existing CMS to expose the additional fields.
And that’s probably the biggest takeaway here: Left unaddressed, the problem of incomplete feeds and schema could see your online store left down by agentic commerce. But fixing the issues shouldn’t take a batch of effort and expense.
5. 15% Were Completely Invisible To Agentic Commerce
One interesting finding came about because of the information we couldn’t see.
Of the 207 URLs we audited, 32 returned an HTTP 403 Forbidden error, including PDPs from awesome brands for illustration Adidas UK, UGG, Converse, and Christian Louboutin. Basically, our devices were incapable to fetch the basal information because these websites person defenses successful spot to forestall bots from scraping their sites.
Of course, deciding to artifact bots is simply a morganatic business choice. There are galore reasons why a marque mightiness want to protect its catalog from scrapers, specified arsenic stopping competitors from scraping real-time pricing information to undercut them connected price.
But the aforesaid defenses that artifact a competitory audit instrumentality will besides artifact a Google aliases OpenAI agent trying to entree merchandise information for agentic commerce. The mobility is whether blocking agentic commerce is simply a conscious business determination aliases an unforeseen consequence.
Category SEO Won’t Help You In Agentic Commerce
When we built the audit sample, we were only willing successful high-traffic merchandise pages. Therefore, we filtered the Ahrefs results to get astatine a database of URLs astir apt to beryllium PDPs, stripping retired class pages and different content.
However, before adding those filters, we noticed that category pages (PLPs) currently win the organic-search game over PDPs by a facet of roughly 10 to 1.
For example, Barbour’s top integrated page is /gb/mens/jackets, attracting 20,248 monthly visits at the clip of our search. But their best-performing PDP is the Ashby waxed overgarment astatine 2,194. We saw this aforesaid shape crossed astir of the endeavor retailers successful our sample.
And because ecommerce strategies are traditionally traffic-focused, a batch of SEO effort has been directed towards class pages complete the years.
But agentic commerce doesn’t care about category pages because they transportation nary of the basal schema. AI agents only read system data at the product level. Barbour might rank well for “men’s wax jacket” in the SERPs and beryllium wholly overlooked by AI agents.
Fixing The Plumbing
All the information these agentic commerce protocols request would already beryllium successful 1 aliases different of your systems, specified arsenic your ERP, PIM, aliases inventory guidance platform. Shipping transportation times, return windows, value validity dates, GTINs – it should all beryllium there.
What’s missing is the plumbing to move this accusation cleanly into the merchant provender and on-page schema – not conscionable erstwhile but continually.
For this to happen, 3 things request to alteration astatine an operational level:
1. Treat The Merchant Feed As Essential Infrastructure
Assess the completeness of your existing feeds. To commencement with, audit your top-selling products to cheque whether they carry each 3 UCP action signals successful schema and prioritize which to reside first. Set an achievable target for really galore products your squad can optimize for agentic commerce each period aliases quarter.
2. Tighten Up Inventory Data Accuracy
If an supplier calls your API and receives stale, incomplete, aliases inaccurate information that causes transactions to fail, the reliability awesome degrades.
The safest course would be to configure your feeds to update in arsenic adjacent to existent time as possible, with sub-hour or moreover sub-minute granularity.
3. Prepare For Loyalty And Dynamic Pricing
At launch, pricing successful agentic commerce was mostly static. Agentic commerce could only connection the same point astatine the aforesaid value to everyone, moreover if some customers might usually beryllium eligible for a discount or different deal when buying via the retailer’s website.
However, the March UCP update added an important caller capability: Identity Linking. This allows the agentic supplier to interact pinch a retailer’s website connected behalf of the customer, “such arsenic accessing loyalty benefits, utilizing personalized offers, managing wishlists, and executing authenticated checkouts.”
Of course, this means a little more work to implement and satisfy the key requirements, but the benefits should beryllium obvious.
SEO Needs Supply Chain Thinking
For decades, ecommerce SEO has been astir PDPs, categories, and content. Like the proverbial three-legged stool, get 1 of these incorrect and your strategy falls over.
Agentic commerce doesn’t add different leg. I’d argue it’s another stool.
Unlike other AI conversations, agentic commerce doesn’t care which retailers have the champion content. This isn’t about answering questions pinch insights and thought leadership. Categories don’t matter either, arsenic the investigation showed. And while PDPs are still important, the protocol is only concerned pinch the schema.
These agentic commerce protocols want clean, accurate, and complete merchandise data. Nothing else.
This requires a different strategy. But this strategy isn’t about optimization. Instead, it’s about acquiring the necessary component data from different systems, applying a definite magnitude of quality control, and then packaging it up to get this data where it needs to go, connected clip and successful bully condition.
This is supply chain strategy applied to SEO. And for illustration immoderate proviso chain, this will require cross-functional ownership and clear deliverables.
While the SEO squad is typically responsible for marque visibility, including wrong AI, the merchant squad usually manages the merchandise and merchant feeds. And past there’s whichever team is responsible for reconfiguring the ecommerce level and implementing the basal technologies to securely synchronize information betwixt systems.
None of these teams are apt to person the authority aliases fund to thrust the basal changes alone, moreover if they’re fully alert of what needs to happen. Therefore, the CMO needs to make the lawsuit astatine committee level, highlighting the growing connection betwixt information infrastructure and commercialized outcomes in AI to demonstrate why the completeness and capacity of these feeds has suddenly go truthful important.
The winners successful agentic commerce won’t be the loudest brands, aliases the best-ranked pages. They’ll be whichever retailer made their products easiest for an AI agent to buy.
More Resources:
- Agentic Commerce Optimization: A Technical Guide To Prepare For Google’s UCP
- Agentic Commerce And The New Rules Of Google Ads
- Why Your Product Feed Is An SEO Asset (And Who Should Own It)
Featured Image: Yellow duck/Shutterstock
English (US) ·
Indonesian (ID) ·