Caching Foundation

WordPress caching eliminates redundant PHP execution and database queries by storing rendered output at multiple layers — from the server edge to the user's browser. Most sites see their biggest TTFB and LCP gains by getting page cache and CDN cache correct before chasing deeper PHP or database optimizations.

Core Idea

A proper cache strategy involves layering Edge (CDN), Page (Server), and Object (Database) caches to serve content as quickly as possible while ensuring dynamic actions like checkout and admin functions remain uncached.

Cache Stack at a Glance

Layer	Purpose	Hits	Misses
Browser cache	Reuse assets already on the user's device.	Repeat visits; static files.	First visit; HTML.
CDN edge cache (APO/Tiered)	Serve HTML/assets from a nearby POP.	Public pages; static assets.	First-ever region hits; bypassed cookies.
Server page cache (LSCache)	Serve HTML from the web server without PHP.	Public pages; guest users.	Personalized, cart/checkout, logged-in pages.
Object cache (Redis)	Speed DB-dependent queries when HTML isn't cached.	Logged-in/admin/checkout; miss after purge.	None (if disabled).

Action Explanation & Estimated Impact

Typical Ranges

Ranges are from real-world WordPress sites on decent VPS + Cloudflare. Your mileage varies by theme/plugins, geography, and traffic.

Action	Why it Matters	Typical Impact
Enable Server Page Cache (LSCache)	Serves prebuilt HTML directly from server store; skips PHP/DB.	TTFB 0.2–0.6s, LCP 0.4–1.0s, PSI +5–15
Enable Cloudflare APO + Tiered Cache	HTML and assets at the edge; fewer origin trips; Tiered improves hit-rate globally.	TTFB 0.3–0.9s global, LCP 0.3–0.8s, PSI +8–20
Set Browser Cache for static assets (1y immutable)	Repeat visits reuse cached JS/CSS/fonts/images instantly.	Repeat-visit LCP 0.3–0.7s, PSI +5–10
Turn on Redis Object Cache (persistent)	Speeds DB reads/writes for uncached pages (logged-in, cart, search).	Uncached TTFB 0.15–0.4s, INP steadier, PSI +3–8 on dynamic pages
Brotli @ CDN + HTTP/2/3	Smaller transfers and better multiplexing; quicker first paint.	FCP/LCP 0.1–0.3s
Guest Mode/Guest Optimization (LSCache)	Ensures first-time visitors get optimized, cacheable HTML even before session starts.	TTFB 0.2–0.4s, LCP 0.2–0.5s, PSI +4–8
ESI blocks for cart/minicart	Cache whole page; hole-punch tiny dynamic fragments.	LCP 0.2–0.5s on Woo product/listing
Smart Purge + Warmup	Keeps hit-rate high post-publish; avoids cold cache.	Sustained TTFB/LCP stability, PSI +3–6 consistency
Stale-While-Revalidate @ CDN	Serves slightly stale HTML instantly while refreshing in background.	Perceived TTFB near-zero on edge hits

Static vs Dynamic Strategy

Static (Blog/Docs/Marketing)

Aggressive ON, long TTL for Server Page Cache. ON (APO), long TTL for CDN HTML Cache.

• Highest scores; minimal exceptions.
• Object cache is optional.
• ESI is not needed.

Dynamic (Woo/LMS/Membership)

ON with careful bypass rules for Server Page Cache. ON with careful bypass for CDN HTML Cache. ON (Redis) for Object cache. ON for cart/minicart for ESI.

• Cache except /cart/, /checkout/, /my-account/; vary on auth cookies.

Safe Baseline (LSCache + Cloudflare) with Explanations

Area	Setting	Why
LSCache Cache	Cache = ON, Cache Mobile = ON (separate)	Delivers HTML from server store; separate mobile to avoid DOM/CSS drift.
LSCache TTL	HTML 17 days, Front page 1 day	Long TTL maximizes hit-rate; purge keeps content fresh.
LSCache Do Not Cache	/cart/, /checkout/, /my-account/, ?add-to-cart=	Prevents caching personalized/transactional routes.
LSCache ESI	ESI = ON, ESI for cart/minicart	Hole-punch tiny dynamic blocks; keep page cacheable.
LSCache Purge	Auto Purge on post/page update; selective purges	Keeps only affected pages fresh; avoids full cold cache.
Cloudflare APO	ON + Tiered Cache + Brotli + Early Hints	Edge HTML+assets, better global hit-rates, smaller transfers, faster start.
Cloudflare Cache Rules	Bypass for /wp-admin*, cookies like wordpress_logged_in_	Prevent serving cached HTML to logged-in/admin.
Browser Cache (headers)	Static assets: cache-control: max-age=31536000, immutable	Ensures instant loads on repeat visits.

Expected impact (combined baseline): TTFB 0.5–1.3s, LCP 0.6–1.6s, PageSpeed +15–35 (desktop often >95 for static sites).

Verifying with CLI (commands + expected output)

info

Run from your workstation or server shell.

Check CDN/Server cache headers

check-headers.sh

curl -I https://example.com/

Expected output (key lines):

curl-output-headers

HTTP/2 200
x-litespeed-cache: hit
cf-cache-status: HIT
cache-control: max-age=0, must-revalidate, no-cache, no-store

Interpretation: HTML served from LSCache (hit) and Cloudflare (HIT). HTML often uses short/zero max-age; static assets differ.

Check a static asset

check-static-asset.sh

curl -I https://example.com/wp-content/uploads/2025/hero.webp

Expected output:

curl-output-asset

HTTP/2 200
cache-control: max-age=31536000, immutable
cf-cache-status: HIT
content-encoding: br

Interpretation: 1-year browser cache, immutable, edge hit, Brotli on.

Confirm no-cache route

check-bypass-route.sh

curl -I https://example.com/checkout/

Expected output:

curl-output-bypass

HTTP/2 200
x-litespeed-cache: miss, no-cache
cf-cache-status: BYPASS

Interpretation: Checkout bypasses caches as intended.

Purge & Warmup Patterns (with benefits)

Pattern	How	Benefit	Impact
Selective Purge	Purge updated post + homepage, categories, tags	Keeps TTL long elsewhere	Maintains high hit-rate
Scheduled Warmup	Use crawler to prefetch top pages after purge	Fewer cold misses after deploy	TTFB/LCP steady after changes
Stale-While-Revalidate	CDN serves stale briefly while refreshing	Zero-downtime feel on updates	TTFB feels ~0s on edge hits

Common Pitfalls & Fixes

Symptom	Likely Cause	Fix
Users see old content	Over-long TTL + no purge	Enable auto-purge on update; selective manual purge.
Wrong version to logged-in users	Edge caching HTML without cookie bypass	Add cookie-aware bypass rules (WordPress auth cookies).
Cart/checkout broken	Cached dynamic routes	Exclude /cart/, /checkout/, ESI cart.
Cache fragmentation (low hit-rate)	Vary by user-agent/device/language too broadly	Minimize varies; only separate where layout truly differs.
Double optimization conflicts	LSCache + another plugin both minifying/caching	Let LSCache handle page caching; JS/CSS handled by your chosen layer (e.g., Perfmatters).

Quick Lab (60-90 minutes)

Lab Instructions

1. Baseline: Record TTFB/LCP/PSI before changes (mobile + desktop).
2. Enable LSCache with rules above; purge all once.
3. Enable Cloudflare APO + Tiered + Brotli; create bypass rule for admin/auth cookies.
4. Set Browser Cache for assets (1y immutable).
5. Turn on Redis (Object Cache Pro/Redis plugin) and confirm Object Cache: Enabled in WP.
6. Warm critical routes (home, blog, category, product, product list).
7. Re-test: first view and repeat view from two locations.
8. Compare deltas: target TTFB 0.8s, LCP 1.0s (static), and PSI +20.

Checklist (print-ready)

View Checklist

• LSCache page cache ON; mobile cache ON (separate).
• Exclusions: /cart/, /checkout/, /my-account/, ?add-to-cart=.
• ESI cart/minicart ON.
• TTLs: HTML 17d; front page 1d.
• Cloudflare APO + Tiered + Brotli + Early Hints ON.
• Cache Rules: bypass admin + auth cookies.
• Browser cache for static assets: max-age=31536000, immutable.
• Redis object cache enabled (persistent).
• Purge on update (selective), crawler warmup configured.
• Verified with curl -I and PageSpeed/Lighthouse from two regions.

Cheat Sheet (one page)

View Cheat Sheet

Task	Setting
Server page cache	LSCache ON, TTL 17d, exclude cart/checkout/account
Edge HTML	Cloudflare APO ON, Tiered ON, Brotli ON
Browser	Static assets max-age=31536000, immutable
Object cache	Redis ON (persistent), monitor hit-ratio
Dynamic fragments	ESI for cart/minicart
Purge	Auto on update; selective manual purges
Warmup	Crawler on sitemap/top pages after purge
Verify	curl -I headers, Lighthouse, WebPageTest
Conflicts	One tool for page cache; one for JS/CSS

What's Next

• Server Cache
• Browser Cache
• Object Cache (Redis)
• CDN Cache