U50
The discovery and structured-data work from the July audit was executed, and it worked. The site is now legible to machines and still invisible in its market.
Did the GEO score improve?
Yes, substantially. Nearly every machine-readability defect named in the July audit has been fixed and verified: homepage title, description, H1, Open Graph image, Organization and WebSite schema, Product schema across the entire catalog, FAQ schema, placeholder text, and insecure social image URLs.
The overall score did not move with it. The July total came from a rubric weighted toward on-page and creative factors. This run weights authority, reviews, and off-page presence at 40 of 100, because those decide whether a commerce site can compete. On that basis U50 is still a Poor property: the fixes were confined to markup, and nothing was done about authority, reviews, or the trust and funnel defects.
How the GEO subscore is calculated
The July report published a category score of "SEO & discovery 34." That came from the July rubric, with different components and weights. It is not the same measure and is not used in any comparison here.
To produce a like-for-like movement, both dates are scored on this run's GEO components. The July column is reconstructed from signals the July report explicitly documented, not re-measured, because the site has since changed and that state cannot be re-crawled.
| GEO component | Max | July | Now |
|---|---|---|---|
| Page-level discovery metadata (title, description, H1, Open Graph) | 6 | 0 | 5 |
| Entity schema present and valid | 8 | 0 | 5 |
| Product and commerce structured data | 5 | 0 | 5 |
| Rating and review data available to AI answers | 3 | 0 | 0 |
| Agent surfaces (llms.txt, agents.md, UCP, MCP) | 5 | 4 | 4 |
| Crawl accessibility for AI retrieval | 3 | 3 | 1 |
| Total | 30 | 7 | 20 |
Normalized: July 23/100, now 67/100.
What was fixedverified against live pages
Every row below was a named defect on July 14 and is now resolved.
| Signal | July 14 | August 3 |
|---|---|---|
| Homepage title | "u50.com" | Descriptive, 54 characters |
| Homepage meta description | Absent | 143 characters, specific |
| Homepage H1 | Absent | 1, semantic |
| Homepage Open Graph image | Absent | Present, HTTPS |
| Homepage JSON-LD | 0 blocks | Organization + WebSite with SearchAction |
| URLs carrying JSON-LD | 0 of 71 | 10 of 45 crawled, all products included |
| Product schema on PDPs | 0 | 14 of 14 sampled, 100 percent |
| Product offer data | None | 5 to 24 variants per PDP, with availability and brand |
| FAQPage schema | Absent | Present |
| Sitemap URLs missing description | 31 of 71 | 3 of 45 crawled |
| "IMAGE COMING SOON" strings | 11 on one PDP | 0 sitewide |
| Social image on insecure http:// | Yes, on PDPs | 0, all HTTPS |
| Shipping page H1 | Said "Returns & Exchanges" | Says "Shipping" |
The finding that matters most
| Category probe | U50 present | Who returned instead |
|---|---|---|
| "best premium heavyweight cotton modal hoodie 430 GSM" | No | TENORE, Independent Trading, Free Fly, TRANSITION, PANGAIA, Redfeather |
| "performance sportswear brand men women recovery travel hoodie" | No | Ridge Merino, Unbound Merino, Wool&Prince, MISSION, VIRUS |
| "U50 United50 performance sportswear" (branded) | Yes, first | Four more U50 pages follow |
The brand owns its name and nothing beyond it. Perfect Product markup does not create demand-side visibility on its own.
GEO gaps that remain
- Zero collection-level schema. All 17 collection pages carry no structured data. No CollectionPage, no ItemList. Collections are the category-intent entry points and they are machine-invisible, which connects directly to the category-probe failure above.
- No entity graph. Not one node carries an
@id. Organization, WebSite, Product, and BreadcrumbList exist as isolated islands rather than a linked graph, so engines cannot resolve them to a single brand entity. - Organization node is minimal. Name, url, and sameAs only. No logo, contactPoint, address, or founding detail.
- No rating data. Product schema is complete on every other axis. AggregateRating is the highest-value missing property for both rich results and AI answers about product quality.
- Boilerplate descriptions. 29 of 45 crawled pages share one identical 150-character brand sentence. The homepage and all product pages have unique descriptions, so the fix landed on audited templates and was skipped on the rest.
- Two pages still lack an H1: /pages/accessories and /pages/seasonal-looks. Both named in July.
Named in July, still open
These were not markup problems, and they were not addressed.
| July finding | Status |
|---|---|
| 03 / Impact and wellness claims outrun the proof | Unchanged. The page still cites "our partnership," "national campaigns reaching hundreds of millions," and "measurable impact" with no named partner, program, methodology, or dated result. |
| 04 / Wholesale page asks for a form that is not there | Unchanged. Copy still reads "please complete the inquiry form below." No inquiry form exists. The page offers only a PDF titled Wholesale Application Form, returned by email. |
| 06 / Footer navigation reads "WOMAN'S" | Unchanged. |
| 07 / Index hygiene: duplicate home URL, empty blog, wishlist pages | Unchanged. /pages/home still returns 200 with its own canonical. /blogs/news still contains zero articles. Wishlist pages remain indexable with no descriptions. |
| 01 / Discount modal is the first experience | [unavailable] this run. Requires a rendered browser session. Neither confirmed fixed nor confirmed present. |
| 05 / Homepage merchandising payload | Homepage HTML grew from 315 KB to 364 KB, so the payload was not reduced. Full session measurement [unavailable]. |
New since July
- A duplicate product URL is live and indexable: /products/flow-high-waist-body-shorts-copy, carrying full Product schema. A working-copy artifact that should not be public.
- The origin rate-limits aggressively. 29 of 74 URLs returned HTTP 429 on the first crawl pass and required serial retries with backoff. Legitimate crawlers and AI retrieval agents hit the same throttling, which directly undercuts the structured-data investment just made.
Authority and off-page3 / 25
- A domain-level link probe surfaced only u50.com's own pages plus unrelated entities: an AMD Alveo U50 accelerator card and a Nevada business register entry for U50, LLC. The brand name collides with hardware and shell-company records, adding an entity-disambiguation burden.
- No editorial press coverage of the brand surfaced across a press-intent probe.
- No review platform presence surfaced.
- Quantified backlink and referring-domain counts: [unavailable]. Backlink index access required. Ahrefs units burned: 0.
Absence of discoverable citations across six probes is evidence, not a measured link count.
Scorecard
| Area | Earned | Max | |
|---|---|---|---|
| Authority and off-page signals | 3 | 25 | |
| Reviews, reputation, and proof | 2 | 15 | |
| Content depth and competitive substance | 6 | 15 | |
| Schema and entity integrity | 9 | 15 | |
| Technical crawl and indexation | 9 | 12 | |
| Agentic and AI-surface readiness | 6 | 10 | |
| Delivery, performance, and security | 4 | 8 | |
| Total | 39 | 100 |
Bands: 85 to 100 strong, 70 to 84 good, 55 to 69 fair, 40 to 54 weak, 0 to 39 poor. Authority, reviews, and off-page presence carry 40 of 100 because they determine whether a commerce property can compete; markup and technical hygiene carry 37. This weighting differs from the July baseline, so the two overall totals are not comparable. The GEO subscore comparison is shown separately above and is computed on this run's components for both dates. Provisional, judgment-weighted score against observable evidence, not an industry-standard tool score.
Recommended work
The machine-readability phase is largely done. This is the demand phase.
| # | Workstream | Impact | Effort | Window |
|---|---|---|---|---|
| 1 | Build category-level authority: editorial placement, category and comparison content, retail and marketplace presence for the material claims the brand already owns | High | High | 90 to 180 days |
| 2 | Launch review capture and add AggregateRating to the existing Product schema | High | Medium | 30 to 90 days |
| 3 | Substantiate or remove the impact and wellness claims: name the partner, link evidence, quantify contributions | High | Low | 2 weeks |
| 4 | Replace the wholesale contradiction with one native qualification form | High | Low | 1 week |
| 5 | Add CollectionPage and ItemList schema to all 17 collections; write unique descriptions for the 29 boilerplate pages | Medium | Medium | 2 to 4 weeks |
| 6 | Add @id to every schema node to form one linked entity graph; enrich Organization with logo, contactPoint, address | Medium | Low | 1 week |
| 7 | Index hygiene: redirect /pages/home, remove the duplicate "-copy" product URL, noindex wishlist pages, add missing H1s, publish or remove the empty blog | Medium | Low | 1 week |
| 8 | Investigate origin rate-limiting so crawlers and AI agents are not throttled; reduce the homepage payload | Medium | Medium | 2 to 4 weeks |
Method and limitations
- The July 14 report was retrieved from its live published page and compared item by item against current live evidence.
- Sitemap index plus five child sitemaps parsed: 74 unique URLs. 45 fully audited, plus 14 product pages verified by targeted sampling across every product family.
- Verified directly: robots.txt, llms.txt, agents.md, UCP discovery document, agentic discovery sitemap, canonicals, meta descriptions, H1 structure, Open Graph and Twitter metadata, JSON-LD parse validity and node types, product offer and availability data, hostname redirect behavior.
- Six discovery probes across brand, category, link, and press intent.
- [unavailable] Quantified backlinks and referring domains, Shopify admin and analytics, Search Console, GA4, Core Web Vitals, rendered browser session, conversion and modal behavior, and live AI-assistant answer probes. No browser surface exists in this runtime.
- 29 of 74 URLs required serial retry after HTTP 429 responses. Sampling covered every distinct product family rather than all 37 products individually.