X9Elysium SEO Audit — Full Report

Site: https://x9elysium.com Audit date: 2026-05-02 Business type: Agency (Shopify consulting) with strong local-service signals (Toronto, Calgary, Mississauga) Hosting platform observed: Hostinger (response header platform: hostinger, x-powered-by: Next.js) — note: CLAUDE.md and netlify.toml describe a static export to Netlify; current deploy does not match.

---

Executive Summary

Category	Weight	Score	Weighted
Technical SEO	22%	45	9.9
Content Quality	23%	55	12.7
On-Page SEO	20%	40	8.0
Schema / Structured Data	10%	85	8.5
Performance (CWV)	10%	55	5.5
AI Search Readiness	10%	60	6.0
Images	5%	65	3.3
SEO Health Score			53.8 / 100

Headline finding: Foundations are well thought-out (rich JSON-LD graph, AI-crawler allowlist, llms.txt, per-blog metadata), but two production-grade defects are dragging the site down hard:

1. Every blog post detail page returns HTTP 500. All 18 URLs in the sitemap under /blog/<slug> 5xx. Confirmed on post-1, post-6, best-shopify-plus-agencies-toronto-2026, launching-dtc-brand-toronto-90-day-playbook. RSS feed and /blog index work, but the actual articles do not. This is the #1 ranking-blocker on the site.
2. Five primary pages share the exact same <title> and meta description: /, /about, /services, /work, /contact. They all return X9Elysium — Shopify Commerce Consulting and the same description, because each subpage is a "use client" component and Next 14 App Router silently disallows metadata exports from client components — so they fall through to the root-layout defaults.

Fix those two and the score jumps roughly 20 points before any other work.

Top 5 critical issues (do this week)

1. /blog/<slug> returns 500 — HTTP 500 on every post. Internal links from /blog, /locations/*, footer, and llms.txt all lead to error pages. Search engines will deindex.
2. Duplicate title/meta on 5 main pages — homepage, about, services, work, contact all share the same title tag. Google treats this as a single canonical entity and will demote the duplicates.
3. No canonical on the 5 main pages — only /blog, /locations/*, /careers, blog-post pages have <link rel="canonical">. Homepage and money pages have none.
4. 404 page returns 500 — /this-page-does-not-exist-12345 returns HTTP 500 instead of 404. Google's crawler treats 5xx as a temporary issue and will retry indefinitely; Search Console will fill with soft errors.
5. output: "export" removed from next.config.js — CLAUDE.md describes a static export to Netlify; the current next.config.js has no output field and netlify.toml still says publish = "out". The build is in an inconsistent state and the live site is being served from Hostinger.

Top 5 quick wins

1. Convert app/about/page.tsx, app/services/page.tsx, app/work/page.tsx, app/contact/page.tsx to server components with metadata exports (split client-only sections into child components). One commit fixes finding #2.
2. Add alternates: { canonical: "https://x9elysium.com" } to root layout and per-page exports.
3. Delete content/posts/post-1.md … post-10.md. They are leftover Hugo template fluff (generic ROI / strategic-planning essays) that hurt E-E-A-T. Remove from sitemap, add 410 (or 301 to a relevant real post).
4. Create /public/images/og-image.png (1200×630). Today the blog metadata fallback points to a 404.
5. Generate a real /404 page (Next App Router's app/not-found.tsx) so soft-404s become hard 404s.

---

Technical SEO

Crawlability

• ✅ robots.txt is well-built: User-agent: * allows site, blocks /api/, lists sitemap. Explicit allow-list for 12 AI crawlers (GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-Web, PerplexityBot, Perplexity-User, Google-Extended, CCBot, anthropic-ai, Amazonbot, Applebot-Extended).
• ✅ sitemap.xml resolves at root, lists 33 URLs with priority, lastmod 2026-05-02 on most entries.
• 🛑 Sitemap lists URLs that 500. All 18 blog posts and the /terms-policy page are in the sitemap; blog posts are broken. Either fix the routes or remove from sitemap until they're back.
• ⚠️ Trailing-slash behaviour returns 308. Internal links use the non-trailing form, which is fine, but do not let any external system link with a trailing slash — every redirect costs link equity.

Indexability

• ✅ <meta name="robots" content="index, follow"> on all main pages.
• ⚠️ Five main pages have no canonical: /, /about, /services, /work, /contact. Without an explicit canonical, Google may pick the wrong URL when query strings or UTM params are appended (e.g., from Tawk.to or Clarity).
• 🛑 Blog post 500s — confirmed on:
  • https://x9elysium.com/blog/post-1 → 500
  • https://x9elysium.com/blog/post-6 → 500
  • https://x9elysium.com/blog/best-shopify-plus-agencies-toronto-2026 → 500
  • https://x9elysium.com/blog/launching-dtc-brand-toronto-90-day-playbook → 500
  • Trailing-slash variants 308 → 500.
• 🛑 Hard 404 missing — random URL returns 500, not 404. app/not-found.tsx does not exist.

Build / hosting consistency

• 🛑 next.config.js no longer sets output: "export". Per CLAUDE.md, this site is intended to deploy as a static export to Netlify (netlify.toml still declares publish = "out"). The live response headers show the site running on Hostinger with x-powered-by: Next.js. Either:
  • re-enable static export and deploy out/ to Netlify; or
  • commit to a Node host, update CLAUDE.md and netlify.toml, and figure out why blog/[slug] specifically 500s on Hostinger.
• ⚠️ next.config.js line 6: images: { unoptimized: true }. This is required for static export but means every <Image> ships the original file at every viewport — no WebP, no responsive srcset.

Security / headers

• ✅ HTTPS enforced. content-security-policy: upgrade-insecure-requests is present.
• ⚠️ No Strict-Transport-Security, no X-Content-Type-Options, no Referrer-Policy. Add these via netlify.toml (or whichever host's headers config is live).
• ⚠️ Tawk.to and Microsoft Clarity load via Script strategy="afterInteractive". Both ship third-party JS that can degrade INP. Prefer lazyOnload for Tawk.

Core Web Vitals

• Could not pull live CrUX data without auth. Static analysis flags:
  • LCP risk: 4.4 MB hero-video.mp4 + an animated emerald-glow gradient hero. Heavy first paint, especially on mobile.
  • INP risk: Tawk.to widget mounts a chat iframe + listeners; Clarity mounts a session-recorder. Both run on afterInteractive. Combined, this often pushes mobile INP over 200 ms.
  • CLS: Likely fine — Inter via next/font/google reserves font space, all <Image> components have explicit width/height.
• Action: run a real Lighthouse on / and /locations/toronto; a 4.4 MB hero video should be preload="none", playsinline, and ideally below-the-fold or replaced with a poster image.

---

On-Page SEO

Title tags & meta descriptions

URL	Title	Description	Canonical
/	"X9Elysium — Shopify Commerce Consulting"	shared default	❌ none
/about	"X9Elysium — Shopify Commerce Consulting"	shared default	❌ none
/services	"X9Elysium — Shopify Commerce Consulting"	shared default	❌ none
/work	"X9Elysium — Shopify Commerce Consulting"	shared default	❌ none
/contact	"X9Elysium — Shopify Commerce Consulting"	shared default	❌ none
/blog	"Insights | X9Elysium — Shopify Strategy for Toronto, Calgary & Vancouver"	unique	✅
/locations/toronto	"Shopify Plus Agency in Toronto | X9Elysium"	unique	✅
/locations/calgary	"Shopify Agency in Calgary | X9Elysium"	unique	✅
/careers	"Careers | X9Elysium — Build the Future of Canadian Commerce With Us"	unique	✅

Root cause: app/about/page.tsx:1, app/services/page.tsx:1, app/work/page.tsx:1, app/contact/page.tsx:1 all start with "use client". Client components in the App Router cannot export metadata — Next silently ignores it and falls back to the root layout. The careers/page.tsx works because it is a server component.

Fix pattern: keep the page as a server component, export metadata, and move client-side state/animations into a child component:

// app/services/page.tsx (server component)
import type { Metadata } from "next";
import ServicesClient from "./ServicesClient";

export const metadata: Metadata = {
  title: "Shopify Plus Services | X9Elysium",
  description: "Shopify store audits, custom apps, platform migrations, and unified commerce strategy for Canadian retailers.",
  alternates: { canonical: "https://x9elysium.com/services" },
};

export default function ServicesPage() {
  return <ServicesClient />;
}

Heading structure

• ✅ Single <h1> on every page audited.
• ⚠️ Earlier WebFetch read of /work summarized three H1-style strings ("Our Work", "Client Stories", "Let's build something that lasts."). Raw HTML shows only one <h1> tag, but verify the visual hierarchy matches semantic levels — sometimes Tailwind classes paint H2/H3 to look like display text and confuse the reader.

Internal linking

• ✅ Solid: nav links, footer, breadcrumb JSON-LD on inner pages, llms.txt cross-references.
• 🛑 Internal links to /blog/<slug> from /blog index, location pages, and llms.txt all currently lead to 500-ing pages. Until #1 is fixed, every internal blog link is hostile to crawlers.

---

Schema / Structured Data

This is the strongest area of the site.

• ✅ Root layout ships an @graph with three nodes: Organization, WebSite, ProfessionalService (app/layout.tsx:58-115). Includes ContactPoint, PostalAddress, sameAs (Instagram, Facebook, LinkedIn).
• ✅ Blog index emits Blog + BreadcrumbList (app/blog/page.tsx:34-70).
• ✅ Blog post detail emits BlogPosting (with inLanguage: "en-CA", wordCount, author with sameAs), BreadcrumbList, and FAQPage only when the post defines FAQs (app/blog/[slug]/page.tsx:77-138). Conditional FAQ emission is correct — meets the Aug 2023 Google guideline.
• ✅ Locations page ships ProfessionalService (LocalBusiness shape) + BreadcrumbList with areaServed listing Toronto, Mississauga, Brampton, Markham, Vaughan, Hamilton, GTA admin area (app/locations/toronto/page.tsx:32-77).
• ✅ Career detail emits JobPosting with hiringOrganization, jobLocationType: TELECOMMUTE, applicantLocationRequirements: Country/Canada (app/careers/[slug]/page.tsx:58-83).

Improvements

• ⚠️ JobPosting is missing validThrough, baseSalary, and directApply: false. Google now requires validThrough to keep job listings eligible for the rich result.
• ⚠️ Add WebPage nodes to inner pages (services, work, contact) once they have unique metadata.
• ⚠️ Consider Service schema for the six service offerings on /services (one per service).
• ⚠️ Organization.address is set to Mississauga only; llms.txt and copy claim a presence in Vancouver and Calgary. If Vancouver is a real office (not just a service area), add OfficeLocation or a second LocalBusiness node.
• ⚠️ No aggregateRating / Review schema on the agency. If client testimonials are real and verifiable, this is a high-leverage add for AI citations.

---

Content Quality (E-E-A-T)

What's working

• Real authored content on the 8 location-aware MDX posts (Toronto, Calgary, Canada-wide). 1.4k–2k words each. First-person, opinionated, dated.
• Strong author signal: Darshan Patel with avatar, LinkedIn sameAs, bio attached to every post via DEFAULT_AUTHOR (app/lib/blog.ts:23-29).
• Locations pages have unique metadata, schema, and pull only category-tagged posts.

What's hurting E-E-A-T

• 🛑 content/posts/post-1.md … post-10.md are leftover Hugo template fluff. Topics: "Maximizing ROI with E-commerce Consulting", "Maximizing ROI with Strategic Planning", etc. — generic, AI-flavoured prose that doesn't cite client work, doesn't mention Shopify specifically, doesn't reflect the agency's real expertise. They dilute the topical authority the 8 real posts are building.
  • post-6.md is only 350 words — thin content.
  • post-2.md even contains a placeholder broken image: ![ROI](https://your-image-url/roi.png).
  • These 10 URLs are in the sitemap, indexed (when they don't 500), and pull the site's authorship signal toward generic blog spam.
• ⚠️ Real posts have ISO dates of 2023-04-04 (post-1) up through 2026 (real ones). When the post-N files get fixed and indexed, mixed-era content with no dateModified updates looks stale.
• ⚠️ No author page (/authors/darshan-patel). The author is referenced as https://x9elysium.com/about instead — fine, but a dedicated author entity strengthens E-E-A-T more.

Recommendation

Treat post-1 … post-10 as a deletion candidate set. For each:

• If the topic is genuinely on-brand: rewrite from scratch with X9Elysium specifics, add a real dateModified, link to a real case study, and rename the slug to something descriptive.
• Otherwise: 410 (gone) via Netlify _redirects or 301 to the closest topical real post.

---

Performance

• ⚠️ Hero video 4.4 MB at public/hero-video.mp4. Even with playsInline and lazy-loading semantics, a 4 MB asset on a hero section is an LCP and data-cap risk on mobile. Either:
  • serve a poster JPG (~80 KB) as the LCP element and load the video on requestIdleCallback; or
  • re-encode at 1080p H.265 + WebM at ~1.5 MB; or
  • drop the video on viewports < 768 px.
• ⚠️ Tawk.to + Clarity both load on afterInteractive (app/layout.tsx:129-152). Combined that's ~150–200 KB of third-party JS plus runtime cost. Move Tawk to lazyOnload (or to delayed based on user interaction). Consider gating Clarity to non-iframe sessions only.
• ⚠️ images: { unoptimized: true } (next.config.js:6) — required for static export, but means no automatic WebP/AVIF, no responsive srcset. If you stay on Hostinger Node, drop this flag and let Next optimise.
• ✅ Inter via next/font/google with display: swap and explicit weights — eliminates CLS from font loading.

---

Images

• ✅ All 5 homepage <img> elements have non-empty alt text.
• ✅ Explicit width/height on every <Image> keeps CLS contained.
• ⚠️ No og-image.png in public/images/. Blog metadata fallback (app/blog/[slug]/page.tsx:32) points to https://x9elysium.com/images/og-image.png which 404s. Twitter / LinkedIn previews for posts without a heroImage will be image-less.
• ⚠️ Logo files duplicated: logo-1.png (86k), logo-x9e.png (86k), logo.png (12k), logo-dark.svg (11k), logo.svg (11k), logoo.jpg (63k), x9-logo.png (200k). Pick one canonical logo per format and delete the rest — keeps the bundle small and avoids cache thrash.
• ⚠️ flower.jpg (313k) and Instagram.html in repo root look like leftovers — confirm they aren't shipping.

---

AI Search Readiness (GEO)

• ✅ Excellent llms.txt (public/llms.txt) — TL;DR, ARR range, services list, locations served, recent insights with one-sentence summaries, RSS + sitemap pointers.
• ✅ All major AI crawlers explicitly allowed in robots.txt.
• ✅ inLanguage: "en-CA" set on BlogPosting schema — reinforces Canadian relevance for AI grounding.
• 🛑 Citability collapses on blog 500s. Perplexity, ChatGPT, and Gemini crawl /blog/<slug> to verify the claims in llms.txt. When those URLs 500, the model's confidence in citing X9Elysium drops — and the AI allowlist becomes meaningless.
• ⚠️ llms.txt mentions Vancouver as a served location, but there is no /locations/vancouver page. Either build one (high-leverage local SEO for an under-served BC market) or remove the Vancouver line from llms.txt and copy.
• ⚠️ No "Frequently Asked Questions" block on /services, /contact, or /about. AI grounding picks up self-contained Q/A passages well; the homepage FAQ is a good model — repeat it on /services.

---

Sitemap audit

• ✅ Sitemap is valid XML, generated dynamically from app/sitemap.ts, includes 33 URLs.
• ✅ Sensible priorities: / 1.0, /contact 0.9, /locations/* 0.85, /blog 0.8, /careers 0.75.
• 🛑 Sitemap lists 18 blog post URLs that 500.
• ⚠️ All lastModified values default to new Date() for static routes (app/sitemap.ts:11-57) — every build "updates" every page. Better to read git mtime or hard-code per-page lastmod.
• ⚠️ Sitemap doesn't include /blog/post-1 … /blog/post-10 URLs explicitly because getAllPosts reads them — once those template files are deleted, they automatically drop from the sitemap. Good.

---

Hreflang / international

• ✅ <html lang="en"> set (app/layout.tsx:118).
• ✅ inLanguage: "en-CA" on BlogPosting.
• ⚠️ Mixed signals: HTML lang is en (generic), schema is en-CA. Pick one. Recommend lang="en-CA" since target audience is Canadian.
• N/A: no other languages — no hreflang work needed.