Evidence-first case study · sports brand SEO + GEO series, part 5 of 20
New Balance’s Korean site isn’t run by New Balance at all — nbkorea.com is operated by licensee E-Land World, and its robots.txt and llms.txt are, by a wide margin, the most AI-crawler-friendly configuration in this entire series: an explicit Allow: / for GPTBot, Google-Extended, ClaudeBot, PerplexityBot and CCBot, plus a meticulously written “Generative Engine Optimization Configuration” llms.txt spelling out allowed AI usage, attribution preferences and restricted paths. Yet that same Korean product page carries zero schema.org markup of any kind — no JSON-LD, no microdata — while the global newbalance.com site it has nothing to do with operationally still ships a full Review/ProductGroup schema block on its 574 Core page. The most sophisticated AI-policy document in the series sits on the site with the least structured data to back it up.
Actual public homepage captured during this audit, October 27, 2026.
What is Newbalance.com trying to do — and why audit it fifth?
This is part 5 of GeoParrot’s sports-brand series, running the same evidence-first method we used on Under Armour, Puma and Adidas. New Balance is a useful fifth name because its most iconic product, the 574, is a decades-old heritage sneaker much like Adidas’s Samba or Puma’s Suede Classic — but its Korea market setup turned out to be structurally unlike anything earlier in this series: not a subpath (Nike/Adidas/Puma) and not a brand-operated ccTLD (Under Armour’s underarmour.co.kr), but an entirely separate company’s domain.
Brand category Global performance footwear and apparel brand built on running heritage, with the 574 as its best-known heritage lifestyle silhouette and USA/UK manufacturing as a recurring brand pillar.
Primary user job Find the 574 (or a comparable heritage runner) in the right colorway and size, backed by visible reviews, and get order/returns help from a first-party support hub.
SEO + GEO challenge A Korea storefront run by an entirely separate licensee company, with a far more thoughtfully written AI-crawler policy than the parent brand’s own global site — but none of the structured data to let an AI engine act on it.
The product we tracked: 574 Core
US product page: $99.99, a 4.2-star rating linking to 2,416 reviews, and 5 colorways.Scrolling past the fold surfaces a “You May Also Like” cross-sell carousel (Made in USA 993/990v4/1300) rather than rendered review text — review content lives in structured data, not visible copy, on this pass.
The category layer: “Men’s Casual Shoes” at 173 items, with 574 as the second featured style tile after the 9060 — a mid-sized catalogue, narrower than Puma’s 747-item grid but wider than Under Armour’s 93-item Running Shoes listing.
AI citation snapshot: 10 real queries, Google AI Mode
We sent 10 verbatim queries to Google AI Mode (udm=50) and checked both direct a[href] links and citation chip aria-labels for the New Balance brand family. Method note, reported honestly: unlike the first four audits, this run’s direct link capture worked consistently (most queries returned 6-11 raw URLs), so counts here lean more on actual hrefs than chip labels alone. We counted a citation whenever newbalance.com, nbkorea.com, or any regional New Balance storefront (newbalance.ca, newbalance.co.uk, newbalance.com.au) appeared as a URL or chip label — the same regional-family counting convention used for Under Armour.
Reddit, kream.co.kr, Naver blogs, eland.co.kr — both newbalance.com and nbkorea.com cited, plus Korean label “뉴발란스”
best new balance running shoes 2026
❌ No
Believe in the Run, RunRepeat, Doctors of Running, Alastair Running, theruntesters.com — all third-party shoe review sites
new balance rewards membership benefits
✅ Yes
Zendesk help centres (NZ/AU), shopback.sg — newbalance.com, .co.uk and nbkorea.com (mobile) all cited
new balance made in usa explained
✅ Yes
footshop.eu, urbanindustry.co.uk, hypedpair.com, wwd.com, Reddit — two separate newbalance.com URLs captured
new balance vs nike running shoes
❌ No
runtothefinish.com, zelpio.com, rundreamachieve.com, TikTok, Instagram — nike.com appeared, newbalance.com did not
Result: New Balance’s domain family cited in 6/10 queries (Nike 5/10, Adidas 4/10, Puma 5/10, Under Armour 8/10 with a looser regional-counting rule). The named-product review query failed again — the fifth brand in a row where a dedicated third-party review site beat the brand’s own PDP on “[brand] [product] review.” Both head-to-head comparison queries also failed to cite NB, continuing the series-wide pattern that brand-vs-competitor queries are the hardest for any brand’s own domain to win — except the one comparison query against NB’s own 990, which succeeded, likely because it stayed within NB’s own product family rather than crossing to a competitor.
G-score audit: the 8-pillar hygiene checklist, six pages
We scored six representative New Balance URLs with the same 8-pillar static-HTML checklist (JSON-LD present, FAQ content, visible date, author byline, concrete data points, table/list structure, 3+ subheadings, a 30–500 character early answer) used in our Week 17 quality-score experiment and on the first four brands. newbalance.com sits behind an Akamai-style WAF that returns a 403 to a plain HTTP fetch (the same pattern we hit on Adidas), so the four US pages were scored from browser-rendered HTML instead of a raw fetch; the two Korea pages fetched cleanly via plain HTTP.
Page
G-score
Passes
Fails
574 Core product page (US)
0.75
JSON-LD, date, author, data points, lists, subheadings
FAQ, early answer
Help centre (US)
0.625
JSON-LD, FAQ, data points, lists, subheadings
date, author, early answer
Korea ML574EVG product page
0.625
date, author, data points, lists, subheadings
JSON-LD, FAQ, early answer
US homepage
0.5
JSON-LD, data points, lists, subheadings
FAQ, date, author, early answer
Men’s Casual Shoes category
0.5
JSON-LD, data points, lists, subheadings
FAQ, date, author, early answer
Korea homepage (nbkorea.com)
0.5
date, data points, lists, subheadings
JSON-LD, FAQ, author, early answer
Average G-score across six pages: 0.5833 (Nike: 0.4375, Puma: 0.354, Under Armour: 0.5208, Adidas: 0.6458) — second-highest average in the series. But the two Korea pages average 0.5625 with a zero on JSON-LD both times, while the two highest-scoring US pages (product page and help centre) both carry JSON-LD — the Akamai-blocked US site has the stronger machine-readable backbone despite being harder to fetch in the first place.
robots.txt (US) — A long, standard Salesforce Commerce Cloud disallow list (cart, checkout, login, wishlist, loyalty pages, AB-test parameters) with crawl-delay rules for Pinterestbot, FacebookBot and bingbot, and a flat Disallow: / for PetalBot (Huawei’s crawler). No AI crawler (GPTBot, ClaudeBot, PerplexityBot, Google-Extended) is named at all, positive or negative — matching Nike’s and Puma’s silence.
robots.txt (Korea, nbkorea.com) — A completely different document, and the most AI-crawler-forward one in the series: after a short generic block, a dedicated “AI / LLM / Generative Search Bots” section explicitly Allow: /s GPTBot, Google-Extended, ClaudeBot, PerplexityBot and CCBot by name. Adidas’s robots.txt named ClaudeBot too, but only to set a crawl-delay — this is the first robots.txt in the series to proactively welcome AI crawlers rather than just acknowledge or throttle them.
llms.txt (US) — newbalance.com/llms.txt returns a clean HTTP 404 via the site’s own custom error page (not a raw server error) — no file exists, consistent with Nike and Under Armour before any file was added. See our explainer on what llms.txt actually does before treating its presence or absence as a ranking factor — Google Search doesn’t read it.
llms.txt (Korea, nbkorea.com) — the headline finding — Returns a clean HTTP 200: a genuinely well-structured “New Balance Korea – Generative Engine Optimization Configuration” document with named sections for Brand Definition, Purpose, Allowed AI Usage (crawl/index, summarize, generate answers, quote with attribution), Attribution Preference, Preferred Content for AI Reference, Content to Avoid (reviews, session pages), Restricted Paths, and a Licensing statement. It’s the most deliberately authored llms.txt in this series by a wide margin — Puma’s was a 2,672-link dump of the entire site, Under Armour’s was a readable-but-mechanical nav export. This one reads like it was written specifically to be read by an AI system, not generated from a sitemap.
Sitemap — US robots.txt lists sitemap_index.xml and sitemap_0.xml on newbalance.com, plus a separate hreflang sitemap hosted on an entirely different domain, nbxml.com. Korea’s nbkorea.com/sitemap.xml is a two-entry file with a code comment reading Generated by Screaming Frog SEO Spider 23.2 — a desktop SEO crawler tool, not a live CMS feed — and its two entries (bare domain and www) carry different, stale lastmod dates (2026-01-15 and 2024-09-03).
Product schema — the gap behind the headline finding — The US 574 Core page carries 7 separate JSON-LD blocks: a ProductGroup with full description, a standalone AggregateRating (4.2/2,416), a review array with per-review author, dates, headline and body text, an Organization block with address and social profiles, a BreadcrumbList, and two ItemList blocks for cross-sell products. The Korea ML574EVG page — despite a visibly displayed 4.9-star rating across 540 reviews — carries zero JSON-LD, zero schema.org references and zero microdata itemprop attributes anywhere in its rendered HTML. The site whose llms.txt explicitly invites AI systems to “generate search-based answers using this site as a primary source” gives those systems nothing structured to extract that answer from.
Locale architecture — a third pattern in the series — newbalance.com/kr/ and /en-kr/ both render the plain US homepage verbatim, with no redirect and no real Korea content — orphan paths New Balance’s own site doesn’t route anywhere. The actual Korea storefront, nbkorea.com, is operated by a separate company entirely: its footer discloses “㈜이랜드월드패션사업부” (E-Land World Fashion Business Division), a Korean licensee, with its own business registration number and customer service lines, unconnected to New Balance’s own Delaware/Boston corporate entity named on newbalance.com. That’s a structurally different setup from Nike/Adidas/Puma’s subpaths and Under Armour’s brand-operated ccTLD: it’s the first brand-licensee locale split in this series.
What works / What to verify
WHAT WORKS
A genuinely well-authored llms.txt and an explicit AI-crawler-welcoming robots.txt on the Korea licensee site, a rich 7-block JSON-LD product schema on the US 574 Core page, and a visible 4.2-to-4.9-star review signal across both markets.
WHAT TO VERIFY
Whether nbkorea.com’s AI-friendly policy document was ever paired with a plan to add structured data, whether that gap is catalogue-wide or specific to this product, and whether New Balance’s global brand team is aware a licensee’s site carries more explicit AI-crawler permissions than its own .com domain.
Target query check — reported honestly, not inflated
We have not re-measured “new balance seo” or “new balance geo audit” query volume this session. Per the pattern already confirmed across nine earlier subjects in this series on October 20, 2026 (Keyword Planner, US) — the large majority showed no measurable search volume at all for “[brand] SEO/GEO audit” phrasing — we’re not claiming this title drives ranking-based traffic. The distribution case for this series rests on internal links, our own index, and readers already following the GEO Lab arc, not on search demand for “[brand] SEO audit” as a query.
Frequently asked questions
Does nbkorea.com’s AI-friendly llms.txt actually help New Balance get cited more?
Not clearly in this snapshot. New Balance’s domain family (including nbkorea.com) was cited in 6/10 queries — middle of the series range (Nike 5/10, Adidas 4/10, Puma 5/10, Under Armour 8/10). The one query where nbkorea.com itself showed up directly (“990 vs 574”) was a within-brand comparison, not a case where the detailed llms.txt policy appeared to change Google AI Mode’s citation behaviour on a broader query.
Why does New Balance Korea use a different company’s domain instead of newbalance.co.kr or a subpath?
We can only report what’s disclosed, not the business reasoning behind it: nbkorea.com’s footer names ㈜이랜드월드패션사업부 (E-Land World Fashion Business Division) as the operating entity, with its own business registration number. Licensee-operated regional e-commerce is common in footwear and apparel, but New Balance is the first brand in this series where we found it — Nike, Adidas, Puma and Under Armour all operate their Korea storefronts directly.
Does a well-written llms.txt matter if the page has no structured data?
Google Search doesn’t read llms.txt at all, and we have no confirmed evidence any major AI engine treats it as authoritative either — see our explainer. What an AI system can reliably extract from a product page — price, rating, review count, availability — typically comes from JSON-LD or microdata, which nbkorea.com’s ML574EVG page has none of despite the policy document’s claims about “generat[ing] search-based answers using this site as a primary source.”
Does a G-score of 0.75 on the US product page mean it’s more likely to be cited by AI engines?
Not reliably — this audit is itself a data point against that idea, consistent with all four prior brands. Our Week 17 experiment found hygiene/structure score was not a dependable predictor of AI citation. Here, the 574 Core PDP scored highest of any page in this audit (0.75) but our 10-query snapshot measures the domain family, not individual URL citation, so we can’t confirm this specific page drove any of the 6 successful citations.
Audit evidence, scope and limits
Inspected: October 27, 2026, public logged-out access from a Korea-based IP (real browser session used throughout). Screens: US homepage, Men’s Casual Shoes category, 574 Core product page (top and cross-sell), US Help centre, Korea homepage, a Korea “574” search (6 results), and the resulting ML574EVG PDP — plus robots.txt, sitemap and llms.txt checks on both domains, a 10-query Google AI Mode citation snapshot, and an 8-pillar G-score scored on six fetched pages.
Not measured: rankings, Search Console performance, analytics, conversion, signed-in/member flows, ChatGPT citations (not attempted, given the composer failures flagged on earlier audits in this series), or the complete product/locale catalogue beyond the one shoe and six URLs sampled here. This is a public-signal snapshot, not a performance claim or a full-catalogue crawl.
This is part 5 of a daily series. Every day, GeoParrot runs this same evidence-first method — real screenshots, a 10-query AI citation snapshot, an 8-pillar G-score audit, and an honest target-query check — against one more major sports and athletic-wear brand. Tomorrow: ASICS.
Leave a Reply