70% Of Top Retailers Are Invisible To Agentic Commerce – Here’s Why

Ecommerce is getting into a brand new section – one the place transactions can occur completely off-site, making web site visits elective, if not redundant. In case your ecommerce and search engine optimization groups are nonetheless taking part in the outdated recreation, there could possibly be bother forward.

In fact, I’m speaking about agentic commerce, the place the complete buyer journey – from discovery to checkout – can occur inside AI. And boy do issues transfer shortly!

In September 2025, OpenAI and Stripe announced their Agentic Commerce Protocol (ACP), permitting clients to not solely analysis and determine which product to purchase, but additionally full single-item transactions proper there within the chat with On the spot Checkout.

In January, Google announced the Common Commerce Protocol (UCP), “a brand new open normal for agentic commerce that works throughout the complete purchasing journey – from discovery and shopping for to post-purchase help.”

After which in March, after solely six months, OpenAI backed away from Instant Checkout, claiming that the preliminary model “didn’t provide the extent of flexibility that we aspire to supply.” As an alternative, ACP is inserting extra concentrate on product discovery – driving worthwhile visibility in your model and merchandise inside ChatGPT – whereas retailers can determine if and the way they combine their very own checkout experiences.

In the identical month, Google rolled out a UCP update that included a bunch of recent checkout and catalog capabilities, together with purchasing carts.

So, if there’s one factor we’ve realized already, you’d higher keep alert.

My sincere prediction is that, someday within the first half of subsequent 12 months, we’ll begin listening to from some retailers that they obtain a greater return on funding from agentic commerce than, say, paid social. The distinction, and the aggressive alternative, will turn out to be clear.

So, I made a decision to do a little analysis to see simply how prepared a few of the top-performing retailers are for agentic commerce. And it seems numerous them could also be extraordinarily unprepared.

However earlier than we get to the analysis information, it’s necessary to know why agentic commerce presents a really completely different problem to retailers.

The Finish Of Site visitors-First Ecommerce

We’ve identified for some time that AI is steadily coopting extra of the client journey, significantly in the discovery, analysis and consideration phases.

The first problem for retailers has been stepping into that all-important consideration set by guaranteeing their model and merchandise are really useful at key factors in related AI conversations. Get that proper, and hopefully, when the client makes a buying resolution, it’s your web site they click on via to.

For 3 a long time now, web site visitors has been on the coronary heart of ecommerce. On-line retailers have devoted a lot time, effort and cash to refining the artwork of buyer attraction – via search engine optimization and content material, via e-mail and social media, via particular presents and loyalty applications – something that may lead extra folks to their product pages.

Sure, there’s extra to ecommerce than simply visitors. There are conversion charges and purchasing cart optimizations, personalization and cross-selling alternatives and … effectively, it’s an extended record. However these all depend on visitors. A possible buyer has to land in your web site earlier than any of those different components come into play.

Nevertheless, when the client doesn’t want to go to your web site – once they can full the complete transaction inside an AI dialog – all earlier ecommerce methods start to unravel.

On the identical day as Google’s UCP launch announcement, Goal put out its personal press release describing what the new agentic purchasing expertise would appear like.

“A visitor begins a dialog in AI Mode or the Gemini app – for instance, ‘I’m stepping into figuring out and need to be each comfy and classy on the health club. Assist me discover cute and reasonably priced floral leggings, in a light-weight colour.’ They’d then see completely different choices to think about and buy with out having to go away the chat.”

So, how do you see that your merchandise are included amongst these choices?

The Mechanics Of Agentic Commerce

Agentic commerce doesn’t work like natural search. This isn’t about rankings the place a decrease place remains to be seen to clients who scroll far sufficient. You’re both in or out.

It additionally doesn’t work in the identical approach as AI citations or mentions. All that work your search engine optimization group would possibly already be doing to enhance your model’s visibility in AI gained’t enable you to right here.

Slightly than crawling the customer-facing content material in your product pages to extract the required data, each protocols as a substitute draw information from the service provider feed or product feed and the on-page schema.

Due to this fact, agentic commerce isn’t a content material optimization downside a lot as an information plumbing one. The AI brokers are merely conduits, piping uncooked product information a technique for AI to relay to the client and feeding transaction information again the opposite approach as soon as the acquisition is full.

That is necessary for 2 causes.

Firstly, it means each protocols retain the retailer because the service provider of file. OpenAI and Google aren’t reselling your wares in the identical approach that, say, the Apple Retailer does.

Secondly, for agentic transactions to achieve success, each protocols want detailed, correct, and up-to-the-minute data that – as you’ll see – numerous retailers don’t at the moment embrace of their feeds.

Proper now, most retailers I come throughout treat their product or merchant feeds as a side project; a handy approach to replace dynamic advert campaigns with primary product information. However the motive advert campaigns require so little data – a product identify, a picture, maybe a particular provide – is as a result of they’re nonetheless about driving visitors again to the product web page. The transaction nonetheless occurs on the web site.

If you need agentic commerce to full that transaction in your behalf, it’s going to want much more data.

The three Schema Fields You Can’t Miss

When Google Gemini evaluates whether or not to advocate a product or not, it isn’t solely involved with primary transactional metadata. Google’s UCP documentation lists three different necessary alerts:

1. priceValidUntil – confirming the pricing data remains to be present.
2. shippingDetails.deliveryTime – offering transport data the client will count on to see.
3. hasMerchantReturnPolicy.merchantReturnDays – supplying the agent with data on return home windows, and many others. to provide the client some purchaser safety.

If any of these fields are absent or incomplete, Gemini gained’t merely rank the product decrease or much less prominently. Your product gained’t be included. Interval.

Pace issues, too. AI doesn’t need to waste worthwhile time and CPU making an attempt to determine issues out. If an agentic agent calls your API and the response is sluggish, it’ll be much less probably to make use of your feed subsequent time.

The identical goes for stale or inaccurate data that outcomes in failed transactions, reminiscent of an merchandise turning out to be unavailable although the feed claimed it was in inventory.

All these points might trigger the AI to view your feed as unreliable, main it to advocate merchandise from manufacturers with extra reliable feeds in future.

The Analysis Findings

So, now we all know what’s required. How are a few of the prime retailers doing?

We recognized 207 top-traffic product element pages (PDPs) from 29 completely different retailers in line with their natural visitors quantity in Ahrefs. In fetching information from these pages, 52 weren’t readable (returning 403, 404, or standing 0 errors), whereas 14 others turned out to not be PDPs (misidentified class or content material pages).

This left us with 141 PDPs to audit – all at the moment successful the old-school search engine optimization recreation.

We then scored every web page towards a 10-point UCP-readiness rubric primarily based on Google’s UCP documentation. (The factors in daring mirror the important thing alerts talked about above.)

1. Has Product / ProductGroup / IndividualProduct JSON-LD on the web page.
2. Has GTIN identifier (gtin / gtin8 / gtin12 / gtin13 / gtin14).
3. Has MPN or distinct SKU.
4. Has model object (identify or @id).
5. Has value + priceCurrency on at the least one provide.
6. Has availability (InStock / OutOfStock / and many others.).
7. Has priceValidUntil.
8. Has shippingDetails.deliveryTime.
9. Has hasMerchantReturnPolicy.merchantReturnDays.
10. Has aggregateRating or overview array.

The findings are stark.

1. The Fundamentals Are Positive

Let’s begin with the excellent news. Just about all of the audited PDPs cowl the fundamentals.

• 99% carry value and availability.
• 99% carry MPN or SKU.
• 96% carry model.

Plus, all 141 of the PDPs seem to have absolutely adopted the product schema really useful by Google since 2014.

2. 70% Miss All Three Of The Most Essential Attributes

Nevertheless, the three key fields I discussed earlier had been solely added to Schema.org between 2020 and 2021. And that is the place issues start to crumble.

• Solely 18% carry priceValidUntil.
• Solely 13% carry shippingDetails.deliveryTime.
• Solely 11% carry merchantReturnDays.

It appears not everybody has stored their schema templates updated.

3. 65% Don’t Embrace A GTIN Subject

Not like the three fields above, the dearth of a World Commerce Merchandise Quantity (GTIN) gained’t essentially stop AI from utilizing your feed, however it’ll handicap the agent’s skill to check your providing with different retailers.

The GTIN is how Google’s UCP is aware of that Product X in your web site is similar bodily merchandise as Product X on a competitor’s web site. And not using a GTIN, the AI agent gained’t acknowledge your product as the identical, treating it as a substitute as a singular merchandise. This implies it may possibly’t do issues like evaluate pricing. If a buyer asks, “Which retailer has one of the best value on Product X?” the reply gained’t be you.

4. The Problem Is Configuration, Not Which Platform You Use

Solely three of the audited manufacturers scored a median of 8 or greater throughout their prime PDPs:

It’s value noting that not one of the top-scoring manufacturers would have wanted to change platforms or rebuild to make themselves agentic commerce prepared. All they did was reconfigure their present CMS to reveal the further fields.

And that’s most likely the greatest takeaway right here: Left unaddressed, the issue of incomplete feeds and schema might see your on-line retailer left behind by agentic commerce. However fixing the points shouldn’t take numerous effort and expense.

5. 15% Had been Utterly Invisible To Agentic Commerce

One attention-grabbing discovering happened due to the information we couldn’t see.

Of the 207 URLs we audited, 32 returned an HTTP 403 Forbidden error, together with PDPs from main manufacturers like Adidas UK, UGG, Converse, and Christian Louboutin. Mainly, our instruments had been unable to fetch the required information as a result of these web sites have defenses in place to forestall bots from scraping their websites.

In fact, deciding to block bots is a reliable enterprise alternative. There are numerous the explanation why a model would possibly need to shield its catalog from scrapers, reminiscent of stopping opponents from scraping real-time pricing information to undercut them on value.

However the identical defenses that block a aggressive audit instrument may even block a Google or OpenAI agent making an attempt to entry product information for agentic commerce. The query is whether or not blocking agentic commerce is a acutely aware enterprise resolution or an unexpected consequence.

Class search engine optimization Gained’t Assist You In Agentic Commerce

Once we constructed the audit pattern, we had been solely enthusiastic about high-traffic product pages. Due to this fact, we filtered the Ahrefs outcomes to reach at a listing of URLs most certainly to be PDPs, stripping out class pages and different content material.

Nevertheless, earlier than including these filters, we seen that class pages (PLPs) at the moment win the organic-search recreation over PDPs by an element of roughly 10 to 1.

For instance, Barbour’s prime natural web page is /gb/mens/jackets, attracting 20,248 month-to-month visits on the time of our search. However their best-performing PDP is the Ashby waxed jacket at 2,194. We noticed this similar sample throughout a lot of the enterprise retailers in our pattern.

And since ecommerce methods are historically traffic-focused, numerous search engine optimization effort has been directed in the direction of class pages over time.

However agentic commerce doesn’t care about class pages as a result of they carry not one of the mandatory schema. AI brokers solely learn structured information on the product degree. Barbour would possibly rank effectively for “males’s wax jacket” within the SERPs and be fully ignored by AI brokers.

Fixing The Plumbing

All the information these agentic commerce protocols want would exist already in a single or different of your methods, reminiscent of your ERP, PIM, or stock administration platform. Transport supply occasions, return home windows, value validity dates, GTINs – it ought to all be there.

What’s lacking is the plumbing to maneuver this data cleanly into the service provider feed and on-page schema – not simply as soon as however regularly.

For this to occur, three issues want to vary at an operational degree:

1. Deal with The Service provider Feed As Important Infrastructure

Assess the completeness of your present feeds. To start out with, audit your top-selling merchandise to examine whether or not they carry all three UCP choice alerts in schema and prioritize which to handle first. Set an achievable goal for what number of merchandise your group can optimize for agentic commerce every month or quarter.

2. Tighten Up Stock Knowledge Accuracy

If an agent calls your API and receives stale, incomplete, or inaccurate information that causes transactions to fail, the reliability sign degrades.

The most secure course could be to configure your feeds to replace in as near actual time as potential, with sub-hour and even sub-minute granularity.

3. Put together For Loyalty And Dynamic Pricing

At launch, pricing in agentic commerce was largely static. Agentic commerce might solely provide the similar merchandise on the similar value to everybody, even when some clients would possibly usually be eligible for a reduction or different deal when shopping for through the retailer’s web site.

Nevertheless, the March UCP update added an important new capability: Identity Linking. This permits the agentic agent to work together with a retailer’s web site on behalf of the client, “reminiscent of accessing loyalty advantages, using personalised presents, managing wishlists, and executing authenticated checkouts.”

In fact, this means a bit of extra work to implement and fulfill the key necessities, however the advantages must be apparent.

search engine optimization Wants Provide Chain Pondering

For many years, ecommerce search engine optimization has been about PDPs, classes, and content material. Just like the proverbial three-legged stool, get one in every of these improper and your technique falls over.

Agentic commerce doesn’t add one other leg. I’d argue it’s one other stool.

Not like different AI conversations, agentic commerce doesn’t care which retailers have one of the best content material. This isn’t about answering questions with insights and thought management. Classes don’t matter both, because the analysis confirmed. And whereas PDPs are nonetheless necessary, the protocol is simply involved with the schema.

These agentic commerce protocols need clear, correct, and full product information. Nothing else.

This requires a distinct technique. However this technique isn’t about optimization. As an alternative, it’s about buying the mandatory part information from completely different methods, making use of a specific amount of high quality management, after which packaging it up to get this information the place it must go, on time and in good situation.

That is provide chain technique utilized to search engine optimization. And like every provide chain, this will require cross-functional possession and clear deliverables.

Whereas the search engine optimization group is often answerable for model visibility, together with inside AI, the service provider group normally manages the product and service provider feeds. After which there’s whichever group is answerable for reconfiguring the ecommerce platform and implementing the required applied sciences to securely synchronize information between methods.

None of those groups are prone to have the authority or price range to drive the required adjustments alone, even when they’re absolutely conscious of what must occur. Due to this fact, the CMO must make the case at board degree, highlighting the rising connection between information infrastructure and industrial outcomes in AI to show why the completeness and efficiency of those feeds has immediately turn out to be so necessary.

The winners in agentic commerce gained’t be the loudest manufacturers, or the best-ranked pages. They’ll be whichever retailer made their merchandise best for an AI agent to purchase.

Extra Assets:

Featured Picture: Yellow duck/Shutterstock