Strategy & Planning

Tech Stack: Current vs. Recommended

The current learn.apmex.com is built on WordPress + Elementor, a page-builder stack that limits filtering, structured data, and content reuse. Below is a side-by-side comparison with our recommended headless rebuild and the rationale behind every layer change.

Current stack blockers

Native faceted filters (current)

MIT

All proposed tool licenses

Proprietary deps eliminated

Stack comparison

Current learn.apmex.com today

CMS

WordPress 6.4.4

Monolithic PHP CMS. Content is stored as post meta, not structured data.

Couples content to presentation

Page builder

Elementor Pro

Paid plugin. Drag-and-drop layout stored as shortcodes in post content - not queryable, not filterable, not portable.

Paid license No structured data

Dynamic fields

JetEngine

Paid plugin for custom post types and meta. Adds filtering capability, but only within WordPress and with performance trade-offs.

Paid license Plugin dependency chain

WordPress default search

Full-text search only. No facets, no type-ahead, no relevance tuning. Users must scroll or know exact keywords.

No faceted search

CDN / Edge

Fastly + Cloudflare

Fastly for origin caching, Cloudflare for bot management and DDoS. Bot rules block all crawlers (returns 403).

Good coverage Blocks legitimate crawlers

SEO / Schema

Yoast SEO (inferred)

Plugin-based schema. Limited Article and FAQ schema. No structured product or price schema for precious metals content.

Thin schema coverage

Deployment

Managed WordPress hosting

Traditional server-rendered PHP. PHP build time on each request. Cache invalidation is manual.

Slower page delivery

Recommended Proposed rebuild

CMS

Sanity (headless)

Content stored as structured documents in a dedicated content lake. Schema is code - every field is typed, queryable, and portable.

Structured content Free tier available

Front end

Next.js or Astro

Static site generation with incremental regeneration. Pages are pre-built HTML served from edge CDN. Zero PHP runtime on requests.

MIT licensed Edge-native delivery

Dynamic fields

Sanity schema (built-in)

No plugin needed. Content types (article, video, Q&A, tool) are defined in the Sanity schema and available as typed API responses.

No paid plugins

Algolia

Dedicated search-as-a-service. Real faceted filtering by metal type, content type, difficulty, and tags. Sub-100ms results with typo tolerance.

Real facets Type-ahead Paid above free tier

CDN / Edge

Vercel Edge or Cloudflare Pages

Static assets served from 300+ edge locations. Bot rules reconfigured to allow legitimate crawlers (Googlebot, Bingbot) while blocking bad actors.

Global edge delivery Crawler-friendly config

SEO / Schema

JSON-LD (native)

Article, FAQPage, HowTo, BreadcrumbList, and SpecialAnnouncement schema generated from Sanity content at build time. No plugin required.

Full schema coverage AI-ready markup

Deployment

CI/CD via GitHub Actions

Push to main triggers build. Static HTML deployed to edge in under 2 minutes. Rollback is one click. Preview deployments for every PR.

Automated pipeline Preview URLs

What changes and why

Layer	Current pain	Proposed fix	Impact
Content model	Post meta + Elementor shortcodes - no queryable schema	Sanity typed documents - every field is an API endpoint	Enables filtering, related content, AI schema
Filtering	No native facets - JetEngine workaround with perf cost	Algolia faceted search with real-time results	Highest-impact UX addition on the rebuild
Page delivery	PHP render on every request	Pre-built HTML from edge CDN	Sub-100ms TTFB vs 800ms+ current
Plugin cost	Elementor Pro + JetEngine annual licenses	No paid build-layer plugins	License cost eliminated
SEO schema	Thin schema via Yoast - no precious-metals types	JSON-LD generated from content schema at build time	Rich results eligibility in Google Search
Crawler access	Cloudflare Bot Mgmt blocks all crawlers (returns 403)	Allowlist Googlebot and Bingbot in firewall rules	Critical - current setup blocks Google indexing
Deployment	Manual FTP / admin dashboard deploys	GitHub Actions CI/CD with preview URLs per PR	Faster iteration, safer rollbacks
Content portability	Content locked in WordPress database	Sanity content lake with REST and GraphQL APIs	Content reusable across web, app, and future AI agents

License and dependency audit

Every tool in the recommended stack is either MIT-licensed open source, has a free tier that covers the Knowledge Center's traffic level, or is a standard paid SaaS with no hidden dependency chain. No proprietary page-builder lock-in.

Tool	Layer	License	Free tier	Notes
Next.js	Front end	MIT	Open source	No runtime cost. Deploy to Vercel or any host.
Astro	Front end (alt)	MIT	Open source	Zero-JS by default. Excellent for content-heavy sites.
Sanity CMS	Content lake	Free tier	Up to 3 users, 10 datasets, 20 GB bandwidth	Studio UI is MIT. API is SaaS - paid above free tier limits.
Algolia	Search	Free tier	10,000 search requests/month, 1M records	Free tier covers low-traffic launch. Paid plans scale cleanly.
JSON-LD (native)	SEO schema	MIT	n/a - code standard	W3C standard. No library needed - plain script tags in HTML.
GitHub Actions	CI/CD	Free tier	2,000 minutes/month on public repos	Sufficient for content-site build cadence.
Cloudflare Pages	Hosting	Free tier	Unlimited requests, 500 builds/month	Alternative to Vercel. No egress cost.

Why this matters for APMEX

The Cloudflare crawler block is the most urgent issue

The current bot management configuration returns HTTP 403 to all automated crawlers - including Googlebot. This means pages on learn.apmex.com that are not already indexed cannot be crawled. Any new articles published since the bot rule was added may not appear in Google Search at all. Correcting the allowlist rule is a quick configuration change that should happen before any other rebuild work.

Risk: Elementor content migration

~200 existing articles are stored as Elementor shortcode blocks. Migration to Sanity requires a one-time conversion pass. The content itself is fine - the wrapper format is what needs to change. We would budget 2-3 weeks for this migration with tooling to automate the bulk of it.

Risk: Algolia index build time

Algolia needs an initial index build from the existing content. This is a one-time operation during launch that takes 30-60 minutes for ~200 articles. Ongoing indexing is event-driven via Sanity webhooks.

Upside: URL stability

Existing URLs (e.g. learn.apmex.com/articles/...) can be preserved exactly during the rebuild using Next.js dynamic routes or redirects. Zero SEO equity is lost from the URL migration.

Upside: AI-ready schema

Structured JSON-LD from Sanity makes APMEX content eligible for Google's AI Overviews and rich snippet features. Precious metals Q&A content is a strong candidate for FAQ rich results, which appear above organic links.