In This Article
The Industry Selling Everything on Earth Can’t Tell a Machine What Day It Updated Its Own Homepage
Amazon. Alibaba. Walmart. Shopify. eBay. Etsy. Temu. JD.com. MercadoLibre. Rakuten. Between them, these ten companies process a meaningful share of every online retail transaction on the planet, run some of the busiest websites in existence, and in two cases, quietly operate the cloud infrastructure that half the internet runs on. If any sector should have solved basic machine-readability by now, it’s this one.
That’s the question I set out to answer, not from a GMV, market-cap, or checkout-conversion angle, but from a structural one. When an AI system, ChatGPT, Perplexity, Gemini, or an AI shopping agent, tries to identify, parse, and cite one of these companies, what does it actually find. I ran the ten largest e-commerce companies by global relevance through the AI Visibility Inspector using the Ivica Srncevic Framework, and the results are the latest entry in this independent research series.
This installment follows analyses of the legal industry, global pharmaceutical, SaaS CRM, global banking, industrial tools manufacturing, life and health insurance, automobile industry, commercial vehicle sector, hospitality and tourism, global light vehicle manufacturers, chemicals and petrochemicals, and most recently, media and entertainment.
Key Takeaways
- Sector average: 52.0, sitting right on the boundary between Grade D and Grade C, and it’s a boundary this dataset straddles literally. Four companies land in Grade C, five in Grade D, and one collapses into Grade F.
- Temu, not Amazon, not Alibaba, posts the highest score in the dataset at 73, Grade C – Fair. The best-structured storefront in this sector is the one that spent the least time building an SEO reputation.
- Freshness has effectively collapsed sector-wide. Average score: 1.2 out of 100. Eight of ten companies scored a literal zero. There is no outlier here propping the average up, unlike prior installments in this series.
- The sector splits almost exactly in half between two distinct failure modes. Five companies (Shopify, Temu, Walmart, eBay, MercadoLibre) trigger a missing dateModified warning. The other five (Alibaba, Amazon, Etsy, JD.com, Rakuten) trigger a missing H1 warning, meaning AI parsers can’t anchor a primary topic at all.
- JD.com’s global-facing domain scores 28, Grade F – Critical, the single lowest result in this entire research series to date, and Amazon’s amazon.com homepage posts the exact same 47, Grade D – Poor it scored in the media and entertainment installment of this series, because it’s the same page.
What “AI Visibility” Means for an E-commerce Company
AI visibility is the measurable degree to which an AI system, an LLM-based search engine, a shopping assistant, or an autonomous research agent, can correctly parse, verify, and cite a company’s website when answering a query about it. It has nothing to do with conversion rate, page speed, or checkout funnel design. It’s a structural property: does the page declare one clear topical anchor, does it carry schema markup that lets a machine assert facts with confidence, and can the system verify when that content was last true.
For an e-commerce company specifically, that translates into a very concrete commercial risk, and it’s a sharper one than in most sectors this series has covered. When someone asks an AI shopping agent where to buy something, which marketplace has the better return policy, or whether a retailer still operates in a given country, the AI answers from whatever it can retrieve with structural confidence. A retailer with strong logistics and a weak structural footprint doesn’t get excluded gently. It gets replaced in the answer by a competitor whose homepage the machine trusts more, and unlike a media brand losing a narrative point, this is a retailer losing the actual transaction.
A Methodology Note Specific to This Sector
Two companies in this dataset, Amazon and Alibaba, are also two of the largest cloud infrastructure providers on earth. AWS and Alibaba Cloud were deliberately excluded from this analysis. This installment evaluates each company’s primary e-commerce storefront, amazon.com and alibaba.com, because that’s the canonical global identity for the retail side of the business and the domain most AI shopping and research queries would actually be routed toward. A separate assessment of cloud-computing domains specifically would need its own dataset and its own framework, since the queries an AI system fields about “the best cloud provider for X” are structurally a different problem than “where do I buy X.”
One domain-selection note worth flagging plainly: JD.com’s result reflects global.jd.com, the company’s international-facing site, rather than its primary Chinese consumer marketplace. That distinction matters for reading the score below. It isn’t a flaw in the analysis. It’s the first finding of the analysis, and this series has now seen the same pattern surface with Netflix’s localized homepage in the media and entertainment installment.
The Scores
| Company | AI Retrieval Score | Grade | Structure | Depth | Schema | Freshness |
|---|---|---|---|---|---|---|
| Temu | 73 | C – Fair | 76 | 85 | 80 | 0 |
| Shopify | 57 | C – Fair | 100 | 75 | 35 | 0 |
| Walmart | 57 | C – Fair | 75 | 80 | 50 | 0 |
| MercadoLibre | 56 | C – Fair | 100 | 85 | 26 | 0 |
| eBay | 55 | C – Fair | 95 | 75 | 35 | 4 |
| Rakuten | 48 | D – Poor | 60 | 80 | 35 | 8 |
| Amazon | 47 | D – Poor | 60 | 100 | 25 | 0 |
| Alibaba | 46 | D – Poor | 60 | 80 | 35 | 0 |
| Etsy | 53 | D – Poor | 60 | 80 | 50 | 0 |
| JD.com | 28 | F – Critical | 35 | 44 | 15 | 0 |
Sector average: 52.0 – balanced almost exactly on the line between Grade C and Grade D, AI Retrieval Index
Four companies landed in Grade C. Five landed in Grade D. One, JD.com, collapsed into Grade F. No company in this dataset reached Grade B or A, and none came close.
Six Findings the Sector Cannot Ignore
Finding 1: The Discount Disruptor Beat the Incumbents at Their Own Game
Temu’s 73 isn’t just the highest score in this dataset. It’s built on the strongest Schema result in the entire sector, 80, more than double Amazon’s 25 and triple Alibaba’s 35. A company whose entire go-to-market strategy runs on paid social and aggressive discounting posted cleaner structured data than marketplaces that have spent two decades investing in organic search. That’s the surprise this dataset keeps handing back across sectors: SEO tenure and AI-retrieval structure are turning out to be two separate skill sets, and this time the newer entrant is the one that happened to get the second one right.
Finding 2: Freshness Didn’t Just Weaken Here. It Nearly Vanished
Sector-wide Freshness averages 1.2, the lowest this series has recorded to date, and there’s no single outlier doing the heavy lifting the way Disney’s 96 did in the media and entertainment installment. Eight of ten companies scored a flat zero. The best result in the entire dataset is Rakuten’s 8. When an AI system can’t verify how current a page’s claims are, in a sector where prices, stock availability, and return policies change constantly, that’s not a cosmetic gap. It’s the exact signal an AI shopping agent needs most and the one this sector has, almost uniformly, failed to provide.
Finding 3: The Sector Splits Cleanly Into Two Different Diseases
This dataset breaks almost exactly in half between two distinct Structural Decay warnings. Shopify, Temu, Walmart, eBay, and MercadoLibre all trigger a missing dateModified warning, meaning the content itself is reasonably well anchored but its currency can’t be verified. Alibaba, Amazon, Etsy, JD.com, and Rakuten all trigger a missing H1 warning instead, meaning an AI parser can’t confidently identify what the page is even about in the first place. Unlike the media and entertainment installment, where seven of ten companies shared the same H1-based failure, this sector doesn’t have one dominant disease. It has two, evenly distributed, and each demands a different fix.
Finding 4: Amazon Posted the Exact Same Score Twice, and That’s Actually the Point
Amazon’s amazon.com scored 47, Grade D – Poor, in this dataset. That’s not a coincidence worth glossing over: it’s the identical score amazon.com posted in this series’ media and entertainment installment, because it’s literally the same page evaluated twice by the same framework. Amazon’s perfect Depth score of 100 (the highest in this entire dataset) hasn’t moved the needle either time, because Schema sits at 25 and Freshness at 0 in both. That consistency is itself a finding. The world’s largest online retailer has built the same structural gap into its homepage regardless of which lens you evaluate it through, retail or media, and it’s been sitting there long enough to show up identically twice.
Finding 5: Structure Shows the Widest Spread of Any Dimension in This Dataset
Shopify and MercadoLibre both posted a perfect 100 on Structure. JD.com posted 35. That’s a 65-point gap on a single dimension inside one ten-company sector, wider than the spread this series recorded on Structure in either the media and entertainment or chemicals and petrochemicals installments. It’s a reminder that “e-commerce” isn’t one homogeneous technical category. A platform-as-a-service company like Shopify, whose entire product is a well-templated storefront engine, structurally outperforms companies that built their homepage as a bespoke, decade-old asset nobody has revisited from an AI-parsing standpoint.
Finding 6: JD.com’s Result Is the Lowest This Series Has Ever Recorded, and Likely for the Wrong Reason
At 28, Grade F – Critical, JD.com’s global.jd.com posted the single lowest score across every industry this series has analyzed so far, industrial manufacturing, banking, pharma, all of it. Structure at 35 and Schema at 15 both sit near the bottom of the entire dataset. Given the domain-selection caveat above, the honest read is that this score likely reflects an internationally-facing site that receives less structural investment than JD’s primary Chinese consumer marketplace, not necessarily the company’s real-world AI readiness in its core market. That’s still worth stating plainly rather than softening: whatever the reason, this is the address an English-language AI query would currently retrieve, and right now it fails almost every dimension this framework measures.
What AI Actually Sees
eBay’s result is the clearest illustration in this dataset of a company doing almost everything right structurally and still tripping on the cheapest fix available. A Structure score of 95 and Depth of 75 show an AI parser has no trouble identifying what the page is or extracting substance from it. But the missing dateModified signal means the system still can’t confirm when any of that was last verified, the same gap Shopify, Temu, Walmart, and MercadoLibre share. This is the fixable half of the sector’s problem: the content and topical anchoring are largely sound, and a JSON-LD dateModified tag is a same-week engineering ticket, not a content strategy overhaul.
Alibaba, Amazon, Etsy, JD.com, and Rakuten present the harder half. A missing H1 tag means the AI system has no declared primary topic to anchor to at all, regardless of how deep or fresh the content underneath might actually be. That’s a structural problem sitting one layer below content quality, in the page’s basic HTML hierarchy, and it’s the reason four of these five companies land in Grade D or worse despite respectable Depth scores across the board.
This isn’t a judgment of any of these companies’ logistics networks, product selection, seller ecosystems, or market share. Nothing in this dataset should be read as commentary on Amazon’s fulfillment network, Alibaba’s B2B reach, or Shopify’s merchant tooling. This is strictly a structural, machine-readability assessment of one primary domain per company, evaluated at a single point in time. A company can move more packages than anyone on earth and still be poorly represented to the AI systems now mediating a growing share of how people discover and compare where to buy something.
Cost of Inaction
Every quarter these gaps go unaddressed, more of the transactional, purchase-intent questions this sector lives on, where’s this cheapest, which retailer ships fastest to my country, is this seller legitimate, get answered by an AI shopping agent pulling from whichever competitor’s page it can verify with confidence. Unlike a media brand losing a narrative point to Wikipedia, this is a direct, first-party revenue leak. A shopper who never sees your storefront in an AI-generated answer never opens your app at all. A sector averaging 52.0, with Freshness collapsed to 1.2 and half the dataset unable to confirm a page’s own primary topic, isn’t losing visibility gradually. It’s already losing the AI-mediated share of a shopping decision it used to own outright, and every week that gap persists, it compounds against whichever retailer’s structured data an AI system happens to trust more.
An Uncomfortable Truth
This sector has spent twenty years optimizing checkout flows down to the millisecond and building recommendation engines that predict what you want before you type it. None of that matters to an AI system that can’t confirm what day your homepage was last updated. Freshness scored 1.2 across ten of the largest retailers on the planet, not because these companies lack current information, prices change hourly, but because almost none of them have bothered to expose that currency in a format a machine can verify. The next competitive battleground in this sector isn’t personalization. It’s proving, in markup, that your page is telling the truth right now, and right now, this entire industry is failing to make that case.
If your team is already thinking about what this means for your own domain, I work with enterprise organizations on exactly this gap through Enterprise Search Advisory, diagnosing where AI systems lose confidence in your content and fixing it at the structural level before it becomes a competitive disadvantage.
Frequently Asked Questions
As covered in Finding 1, Temu posted the strongest Schema result in the dataset, 80, well above Amazon’s 25 and Alibaba’s 35. Structured data quality, not company size or SEO tenure, is what drove the gap in this specific assessment.
As detailed in Finding 2, eight of ten companies scored a literal zero on Freshness, meaning their pages carry no verifiable dateModified or equivalent signal an AI system can check. This is a markup gap, not a content-currency gap; these companies almost certainly update pricing and stock constantly, they simply don’t expose that fact in a machine-readable format.
Not necessarily. As noted in the methodology section and Finding 6, this result reflects global.jd.com, the company’s internationally-facing domain, rather than its primary Chinese consumer marketplace. The score reflects what an English-language AI query would currently retrieve, not a verdict on JD’s real-world scale.
As covered in Finding 4, Amazon’s perfect Depth score is offset by weak Schema (25) and zero Freshness. A strong content-substance signal doesn’t compensate for missing structured data or unverifiable currency, both of which AI systems weigh heavily for confident citation.
For the five companies triggering a missing dateModified warning, Shopify, Temu, Walmart, eBay, and MercadoLibre, adding a dateModified JSON-LD property is a narrow, same-week technical fix. The five companies with a missing H1 warning face a slightly deeper structural issue, since it affects how an AI parser identifies the page’s primary topic in the first place.
Key Takeaways (Recap)
- Sector average of 52.0 sits on the exact boundary between Grade C and Grade D, the tightest split this series has recorded.
- Temu’s 73 is the highest score in the dataset, driven primarily by the strongest Schema result recorded, ahead of every legacy marketplace evaluated.
- Freshness has effectively collapsed sector-wide at 1.2, with no outlier propping up the average the way prior installments in this series have shown.
- The sector splits nearly in half between two distinct failure modes: missing dateModified signals and missing H1 anchors, each requiring a different remediation path.
Research Date: July 2026 | Methodology: Ivica Srncevic Framework + AI Visibility Inspector. This research is independent, not sponsored by any organization or legal entity. All company names and logos are used for identification and analysis purposes only.
Where This Goes From Here
If you’re evaluating your own organization against these findings, there are two ways I can help. For a hands-on structural diagnosis and remediation roadmap specific to your domain, my Enterprise Search Advisory engagement walks through the same framework applied here, at the level of your actual site architecture. For organizations that need ongoing, always-on monitoring of how AI systems are representing their brand, NovaX AI Visibility Intelligence tracks these exact signals continuously rather than as a point-in-time snapshot. Both start with the same question this article asked about Amazon, Temu, and the rest of this dataset: what does AI actually see when it looks at you.
This research is part of an ongoing independent series analyzing AI visibility across global industries. Previous installments cover the legal industry, global pharmaceutical, SaaS CRM, global banking, industrial tools manufacturing, life and health insurance, automobile industry, commercial vehicle sector, hospitality and tourism, global light vehicle manufacturers, chemicals and petrochemicals, and media and entertainment. All assessments use the AI Visibility Inspector and the Ivica Srncevic Framework.
