SEO Audit — aaronkr-courses.github.io

SEO Audit — aaronkr-courses.github.io

Site: https://courses.aaron.kr
Date: 2026-05-01
Auditor: Claude Code (claude-sonnet-4-6)
Jekyll plugins: jekyll-seo-tag, jekyll-sitemap, jekyll-feed

Summary Table

# Check Before After Status
1 <title> — unique, descriptive Duplicate (head.html + seo-tag) seo-tag only; index title improved FIXED
2 <meta name="description"> — all pages Missing on all _pages; course pages used structured code string Added to all 9 _pages; course pages get auto-generated human description FIXED
3 <link rel="canonical"> Present (jekyll-seo-tag) Present (jekyll-seo-tag) PASS
4 og:title Present (jekyll-seo-tag) Present PASS
5 og:description Structured code string on courses; missing on _pages Human-readable description injected; _pages get site fallback FIXED
6 og:image Absent on all pages Course img field wired; site.logo fallback for all other pages FIXED
7 og:url Present (jekyll-seo-tag) Present PASS
8 og:type article on courses (jekyll-seo-tag default) Unchanged — acceptable; not harmful NOTED
9 twitter:card summary (jekyll-seo-tag default) summary_large_image when page.img present FIXED
10 twitter:title Present (jekyll-seo-tag) Present PASS
11 twitter:description Structured code string on courses Human-readable description injected (overrides seo-tag) FIXED
12 twitter:image Absent Course img or site.logo fallback FIXED
13 meta name="robots" Absent index, follow on all Jekyll pages; /attend pages noindex,nofollow FIXED
14 meta name="viewport" Present Present PASS
15 html lang attribute site.lang = en (correct for server-side) Present and correct PASS
16 robots.txt — valid, references sitemap Present but missing Disallow for /attend Added Disallow: /attend/ FIXED
17 sitemap.xml — present and correct Present; includes PDF assets; URLs correct in live build Added sitemap:false for /attend and /assets/pdf paths FIXED
18 Sitemap URL correctness _site/ shows localhost:4000 (local build artifact) _config.yml url is correct; live build uses https://courses.aaron.kr PASS
19 hreflang N/A — JS-only bilingual toggle (not URL-based) Skipped by design N/A
20 Structured data — Course schema BlogPosting on courses (jekyll-seo-tag default) Added proper Course + CourseInstance JSON-LD FIXED
21 Structured data — EducationalOrganization Absent Added Person + WebSite JSON-LD on homepage FIXED
22 One <h1> per page Present on all layouts Present PASS
23 Crawlable links All <a href> links present; nav uses relative_url Pass PASS
24 Descriptive link text Nav uses lang-en/lang-ko spans; course cards have aria-label Pass PASS
25 Duplicate <title> tag Both head.html manual title AND jekyll-seo-tag outputting <title> Removed manual <title> from head.html; seo-tag is authoritative FIXED

Per-Check Findings and Fixes

1. <title> Tag — FIXED

Before: head.html had a hardcoded <title>SEO Audit — aaronkr-courses.github.io — Aaron Snowberger — Courses</title> AND `

SEO Audit — aaronkr-courses.github.io | Aaron Snowberger — Courses

` also outputs <title>. This produced duplicate title tags in every rendered page.

After: Removed the manual <title> tag from _includes/head.html. `

SEO Audit — aaronkr-courses.github.io | Aaron Snowberger — Courses

` (jekyll-seo-tag) is now the sole title tag generator. It produces page.title | site.title format with an | separator (e.g., C++ Programming | Aaron Snowberger — Courses).

Also updated index.md title from Courses (generic) to Teaching & Courses (more descriptive for og:title).

2. <meta name="description"> — FIXED

Before: No description: front matter on any _pages file. All 9 pages (archive, office-hours, policies, policy-ai, policy-attendance, policy-assignments, policy-collaboration, policy-late, policy-tests) fell back to site.description (“Courses taught by Prof. Aaron Snowberger at five Korean universities.”) — same text on every page.

For course pages, page.description held the structured display string CODE • SEMESTER • UNIVERSITY (e.g. INFO2118 • 2026년 1학기 • 한밭대학교) — not suitable as a meta description.

After:

  • Added unique description: front matter to all 9 _pages files (120–160 chars each).
  • Added description: to index.md.
  • _includes/head.html now generates a human-readable meta description for course pages by parsing page.description components and building: {title} — {semester}, {university}. Taught by Prof. Aaron Snowberger. Find the syllabus, schedule, textbooks, and grading breakdown. (~148 chars).
  • Course authors may optionally add seo_description: front matter to any course file to override the auto-generated description.

3. Canonical Link — PASS

jekyll-seo-tag generates <link rel="canonical"> on every page using site.url + page.url. No action required.

4–5. Open Graph title / description — FIXED

og:title was already present via jekyll-seo-tag. og:description on course pages used the structured code string. Fixed by injecting <meta property="og:description"> after `

SEO Audit — aaronkr-courses.github.io | Aaron Snowberger — Courses

` on course pages (crawlers use the last duplicate occurrence).

6. og:image / twitter:image — FIXED

Before: No og:image or twitter:image tags anywhere on the site.

After (three-tier logic in head.html):

  1. If page.img is set (course pages) → https://courses.aaron.kr/assets/img/{page.img}
  2. Else if site.logo is set → https://courses.aaron.kr/assets/img/aaron.jpg
  3. Else → no tag (shouldn’t occur given site.logo is now configured)

Also added logo: /assets/img/aaron.jpg to _config.yml so jekyll-seo-tag can reference it for the site-wide SEO JSON-LD Person schema.

7–8. og:url / og:type — PASS / NOTED

og:url is correct via jekyll-seo-tag. og:type outputs article for course collection pages because jekyll-seo-tag detects a modification date (from file mtime). This is not harmful — article is an acceptable type for educational content pages. Changing it would require forking the seo-tag plugin or suppressing its entire output and rewriting manually; not worth the maintenance cost.

9–12. Twitter Card — FIXED

Before: twitter:card: summary (jekyll-seo-tag default), no twitter:image.

After:

  • Pages with page.img: twitter:card: summary_large_image + twitter:image pointing to course book cover
  • All other pages: twitter:image pointing to site.logo (aaron.jpg), keeping summary card
  • twitter:description overridden on course pages with human-readable text

13. meta name="robots" — FIXED

Before: No <meta name="robots"> anywhere.

After:

  • _includes/head.html now outputs <meta name="robots" content="index, follow"/> on all Jekyll-rendered pages.
  • attend/index.html and attend/admin.html both have <meta name="robots" content="noindex, nofollow"/> added directly.

14. meta name="viewport" — PASS

Was already present: <meta name="viewport" content="width=device-width,initial-scale=1.0"/>. No change.

15. html lang attribute — PASS

default.html: <html lang="en" data-theme="dark">. site.lang: en in _config.yml. Since the EN/KO toggle is JavaScript-only (no separate URL routes), lang="en" is correct as the server-side base. The JS toggle uses document.documentElement.classList.toggle('ko', ...) for CSS targeting. hreflang is intentionally omitted (see #19).

16. robots.txt — FIXED

Before:

User-agent: *
Allow: /
Sitemap: https://courses.aaron.kr/sitemap.xml

Note: The file had --- front matter delimiters (needed for Liquid processing of https://courses.aaron.kr), so the Sitemap line would render correctly in production.

After: Added Disallow: /attend/ to prevent crawlers from indexing the Firebase attendance app.

17–18. sitemap.xml — FIXED

The _site/sitemap.xml examined during the audit showed localhost:4000 URLs — this is a local build artifact. The live build uses url: https://courses.aaron.kr from _config.yml.

Fixes applied:

  • Added sitemap: false default for path: "attend" in _config.yml so jekyll-sitemap won’t include /attend/* pages.
  • Added sitemap: false default for path: "assets/pdf" to exclude PDFs from sitemap (they appeared in the local sitemap and add noise/waste crawl budget).

19. hreflang — N/A (Intentionally Omitted)

The bilingual EN/KO toggle is JavaScript-only with a single URL per page. There are no separate /ko/ URL variants. hreflang is URL-based and is not applicable here. Implementing it would require adding x-default and ko alternates pointing to the same URL — which is technically valid but misleading and offers no SEO benefit. Correctly omitted.

20. Structured Data — Course Schema — FIXED

Before: jekyll-seo-tag outputs BlogPosting JSON-LD for all course pages — incorrect type.

After: Created _includes/seo_course_schema.html with proper schema.org/Course JSON-LD:

  • @type: Course
  • name, description, url
  • provider → EducationalOrganization (university)
  • instructor → Person (Aaron Snowberger) with sameAs links
  • hasCourseInstance with location and schedule from page.information
  • offers with free/0 KRW pricing
  • Included from head.html for all layout: course pages

21. Structured Data — EducationalOrganization / WebSite — FIXED

Before: No site-level structured data beyond what jekyll-seo-tag provides.

After: Created _includes/seo_org_schema.html with @graph containing:

  • Person (@id: https://aaron.kr/#person) with all social profile sameAs links
  • WebSite (@id: https://courses.aaron.kr/#website) linked to person publisher
  • Included from head.html only on the homepage (page.url == '/')

22. One <h1> per page — PASS

  • default.html / page.html layout: <h1>SEO Audit — aaronkr-courses.github.io</h1> — one per page.
  • course.html: <h1 class="animate d2">SEO Audit — aaronkr-courses.github.io</h1> — one per course page.
  • Homepage (index.md): <h1 class="animate d2">Teaching &amp;<br><em>Courses</em></h1> — one h1.
  • Archive (archive.md): <h1>Course <em>Archive</em></h1> — one h1.
  • All _pages that use page.html layout produce one h1 via SEO Audit — aaronkr-courses.github.io.

No fixes needed.

23–24. Crawlable Links / Descriptive Link Text — PASS

All navigation links use <a href="">. No JavaScript-only navigation. No href="#" dead links in the nav (the archive occasionally uses ann.url | default: '#' for announcements without URLs — minor, not an SEO concern).

Course card links use aria-label=" — " — fully descriptive for screen readers and crawlers.

Nav links expose both lang-en and lang-ko spans — both are in the DOM and crawlable (CSS shows/hides based on .ko class). Google will see both text variants.

Files Changed

File Change
_includes/head.html Removed duplicate <title> tag; added meta robots index/follow; added course description override logic; added og:image / twitter:image injection; added twitter:card summary_large_image for courses; wired Course and Org structured data includes
_includes/seo_course_schema.html NEW — Course JSON-LD schema (schema.org/Course)
_includes/seo_org_schema.html NEW — Person + WebSite JSON-LD schema for homepage
_config.yml Added logo, social, twitter keys for jekyll-seo-tag; added sitemap: false defaults for /attend and /assets/pdf paths
robots.txt Added Disallow: /attend/
index.md Updated title to Teaching & Courses; added description: front matter
_pages/archive.md Added description: front matter
_pages/office-hours.md Added description: front matter
_pages/policies.md Added description: front matter
_pages/policy-ai.md Added description: front matter
_pages/policy-attendance.md Added description: front matter
_pages/policy-assignments.md Added description: front matter
_pages/policy-collaboration.md Added description: front matter
_pages/policy-late.md Added description: front matter; title improved to Late Submission Policy
_pages/policy-tests.md Added description: front matter
attend/index.html Added <meta name="robots" content="noindex, nofollow"/>
attend/admin.html Added <meta name="robots" content="noindex, nofollow"/>

Remaining Recommendations (Not Fixed — Require Content Decisions)

  1. og:type on course pages — Currently outputs article (jekyll-seo-tag default). Would ideally be website or a custom Course type, but overriding requires suppressing the entire seo-tag output. Low priority; article is not harmful.

  2. Course description: field dual-use — The structured CODE • SEMESTER • UNIVERSITY format is used both as display data in layouts and (previously) as the SEO meta description. The head.html now auto-generates a better SEO description from the parts. For any course where the auto-generated description is insufficient, add seo_description: "..." to that course’s front matter.

  3. Social preview image quality — Course book cover images (assets/img/books/) are small (local assets). Consider creating dedicated 1200×630px Open Graph images for the most important current-semester courses.

  4. Sitemap URL in pre-built _site/ — The committed _site/sitemap.xml shows localhost:4000 URLs. This is a local build artifact. The live GitHub Pages build will produce correct https://courses.aaron.kr/ URLs. Delete _site/ from the repository if it should not be committed.

  5. Google Search Console — Submit https://courses.aaron.kr/sitemap.xml to Google Search Console to ensure all 47+ course pages are indexed. Verify the canonical domain (CNAME is set; confirm redirect from aaronkr-courses.github.io → courses.aaron.kr is active).