Kevin D. Graham

SEO Implementation Spec — kevindgraham.com

How to use this: Drop this file in the root of the kevindgraham.github.io repo, then tell Claude Code:

Read seo-implementation-spec.md and work through it phase by phase. Start with Phase 0 and report what you find before making changes. Show me git status and git diff before any commit, and don’t push until I say go.

Phase 0 — Audit before you change anything

Report back on:

Is this a plain static HTML site or Jekyll? (Check for _config.yml, _layouts/, _includes/.)
List every page/route in the site and its current <title> and <meta name="description">.
Does sitemap.xml exist? robots.txt? A CNAME file?
Are there duplicate or missing <h1> tags on any page?
How many images lack alt attributes?
Is any structured data (JSON-LD) already present?

Do not make changes until this audit is reported.

Phase 1 — Technical foundation

robots.txt

Create at site root:

User-agent: *
Allow: /

Sitemap: https://kevindgraham.com/sitemap.xml

sitemap.xml

If Jekyll: add the jekyll-sitemap plugin to _config.yml and let it generate.
If plain HTML: generate sitemap.xml manually listing every real page with <loc> and <lastmod>. Exclude 404 pages, thank-you pages, and any test files.

Canonical URLs

Add a self-referencing canonical to every page:

<link rel="canonical" href="https://kevindgraham.com/PAGE-PATH/">

Pick ONE canonical form (with or without www, with or without trailing slash) and make every internal link match it.

404 page

Create a branded 404.html in the navy/gold palette that links back to the homepage, blog, and toolkit — not a dead end.

Phase 2 — Per-page metadata

For every page, write unique values. No duplicates across the site.

Element	Rule
`<title>`	50–60 characters. Front-load the search term, end with `\\| Kevin D. Graham`
`<meta name="description">`	140–160 characters. Written as a hook, not a summary
`<h1>`	Exactly one per page, and it should not be identical to the title tag
`<html lang="en">`	Present on every page
Heading order	H1 → H2 → H3, no skipped levels

Open Graph + Twitter cards

Add to every page so links look right when shared:

<meta property="og:title" content="...">
<meta property="og:description" content="...">
<meta property="og:image" content="https://kevindgraham.com/images/og-default.jpg">
<meta property="og:url" content="https://kevindgraham.com/PAGE-PATH/">
<meta property="og:type" content="website">
<meta name="twitter:card" content="summary_large_image">

Flag if no OG image exists at 1200×630 — that one needs to be designed, not coded.

Images

Add descriptive alt text to every image (describe the content, not “image of”).
Add width and height attributes to prevent layout shift.
Add loading="lazy" to any image below the fold.
Flag any image over 200KB and convert to WebP where possible.

Phase 3 — Structured data (JSON-LD)

Homepage — Person schema

{
  "@context": "https://schema.org",
  "@type": "Person",
  "name": "Kevin D. Graham",
  "url": "https://kevindgraham.com",
  "jobTitle": "Digital Media Educator and Content Creator",
  "description": "...",
  "sameAs": ["https://instagram.com/strategicdownloads"]
}

Book page — Book schema

Include name, author, description, bookFormat, and the purchase URL.

Blog posts — Article schema

Include headline, datePublished, dateModified, author, and image.

Any FAQ-style content — FAQPage schema

Only where real Q&A content exists on the page. Don’t fabricate questions to game it.

Validate every block against schema.org’s syntax before committing.

Phase 4 — Internal linking + crawlability

Every blog post should link to at least one pillar page and one other post.
Every pillar page should link to at least two related posts.
Use descriptive anchor text — “how to write better captions,” not “click here.”
Confirm no page is orphaned (reachable only by typing the URL).
Confirm nav links appear in the HTML source, not injected by JavaScript.

Phase 5 — Performance + accessibility

Inline critical CSS or minimize render-blocking resources.
Preload the primary web font; add font-display: swap.
Confirm color contrast meets WCAG AA against the navy/gold/cream palette.
Confirm the site works with JavaScript disabled (static site — it should).
Run a Lighthouse audit and report scores for Performance, Accessibility, Best Practices, and SEO. Target 90+ on all four.

Phase 6 — Blog scaffolding

If a blog post template exists, make sure every new post automatically gets:

Unique title, description, and canonical
Article JSON-LD
Published + modified dates in the HTML
OG tags pulled from the post’s own front matter

If it doesn’t exist, create the template so future posts inherit all of this without manual work.

Verification checklist

Before pushing, confirm:

Every page has a unique title and description
sitemap.xml returns valid XML and lists every live page
robots.txt points to the sitemap and blocks nothing important
Every page has exactly one H1
Every image has alt text
JSON-LD validates on the homepage, book page, and one blog post
No broken internal links
Lighthouse SEO score is 95+

Then show git status and git diff, and wait for approval before pushing to main.

What Claude Code CANNOT do — Kevin’s list

These require a human and an account:

Google Search Console — verify the domain, submit the sitemap, request indexing on key pages.
Bing Webmaster Tools — same, five-minute setup.
Google Business Profile — for anything local (photography, local visibility work).
Keyword decisions — what you actually want to rank for. Code can’t pick this.
Writing the content — the blog posts are the engine. Nothing above matters without them.
Backlinks — podcast appearances, guest posts, directory listings, book retailer pages.