Most "speed up WordPress" advice is a pile of plugin recommendations with no sense of which lever actually moves the needle. The truth is that a handful of fixes account for almost all the gain, and the order you do them in matters. Below are twelve tactics, ranked roughly by impact-per-effort, that genuinely improve real-world load times and your Core Web Vitals — not just your synthetic test score.

For reference, the targets you are chasing in 2026 are LCP under 2.5s, INP under 200ms (INP replaced FID as a ranking signal back in March 2024), and CLS under 0.1. A good server TTFB sits under 200ms. Keep those numbers in mind as you work — chasing a "100 score" in a lab tool while your INP is 400ms on a real phone is the classic beginner trap.

Fix the foundation first (tactics 1–4)

1. Put it on a host that isn't the bottleneck

No plugin can rescue a slow server. If your TTFB is consistently above 600ms, your host is the problem, full stop. Shared bargain hosting (the $3/month tier) routinely delivers 800ms–1.5s TTFB under any real traffic. Move to a host built for WordPress — Kinsta, WP Engine, Cloudways, or SiteGround's GrowBig+ tiers — and confirm you are running PHP 8.2 or 8.3. The jump from PHP 7.4 to 8.x alone often cuts execution time 20–30% for free. This single change frequently does more than the other eleven combined.

2. Add full-page caching

Page caching stores the finished HTML so WordPress doesn't rebuild the page from PHP and database queries on every visit. This is the biggest software lever you have. If your host runs LiteSpeed, install LiteSpeed Cache — it's free and excellent. Otherwise WP Rocket (premium, near-zero config) or FlyingPress are the go-to choices. On managed hosts like Kinsta or WP Engine, server-level caching is already on; don't stack a second page cache on top or you'll fight cache-purge bugs.

3. Turn on object caching

Page caching skips work for anonymous visitors, but logged-in users, WooCommerce carts, and dynamic pages still hammer the database. Redis (or Memcached) object caching stores the results of expensive database queries in memory. This is the unsung hero for dynamic sites — stores, membership sites, forums. Most managed hosts offer Redis as a toggle; pair it with the free Redis Object Cache plugin. If your site is a brochure-style blog with no logged-in traffic, you can skip this one.

4. Put a CDN in front of everything

A CDN serves your static assets — and increasingly your cached HTML — from a location physically near each visitor, slashing latency for anyone outside your server's region. Cloudflare's free tier covers most sites; BunnyCDN is cheap, fast, and pay-as-you-go if you want more control. Enable Cloudflare's "Cache Everything" page rule or APO for WordPress to cache HTML at the edge, not just images and CSS.

Trim the page weight (tactics 5–7)

5. Compress and convert images to WebP or AVIF

Images are the single heaviest thing on most pages, and the unoptimized hero image is almost always your LCP element. Serve modern formats — WebP is universally supported now, AVIF is even smaller — using ShortPixel, Imagify, or EWWW Image Optimizer. A 2MB JPEG hero compressed to a 150KB WebP can pull seconds off LCP. Also stop uploading 4000px-wide photos into a 1200px column; resize before upload.

6. Lazy-load images, iframes, and video embeds

There's no reason to download a YouTube embed or a footer image before the visitor scrolls to it. WordPress lazy-loads images natively via loading="lazy", but the big win is facade-loading heavy third-party embeds — replacing a full YouTube iframe (which drags in 500KB+ of JavaScript) with a clickable thumbnail. WP Rocket, FlyingPress, and LiteSpeed Cache all do this. One caveat: never lazy-load your above-the-fold LCP image, or you delay the very thing Google measures.

7. Switch to a lightweight theme

Bloated multipurpose themes (the kind bundled with a page builder and 40 demo sites) can ship 300–500KB of CSS and JS before you write a word. Lean, performance-first themes — GeneratePress, Kadence, Blocksy, or Astra — load a fraction of that. If you're already deep into a heavy theme, a re-theme is a project, but it's often the difference between fighting for every point and having performance built in by default.

Tame the render-blocking front end (tactics 8–10)

8. Remove unused CSS and inline the critical path

The average WordPress page loads CSS for components it never renders — slider styles on a page with no slider, WooCommerce CSS on your About page. Remove unused CSS (a feature in WP Rocket, FlyingPress, and LiteSpeed/QUIC.cloud) strips it per-page, and critical CSS inlines just the styles needed to paint above-the-fold content, deferring the rest. This is one of the highest-impact fixes for LCP and the "render-blocking resources" warning in PageSpeed Insights. Test thoroughly — overly aggressive CSS removal can break layouts.

9. Defer and delay JavaScript

JavaScript is the main villain behind a bad INP score, because parsing and executing it blocks the main thread and makes the page feel unresponsive to taps and clicks. Two moves: defer non-critical scripts so they load after HTML parsing, and delay scripts (analytics, chat widgets, ad tags, Facebook Pixel) until the user first interacts. Delaying third-party scripts is often the single biggest INP improvement available, and every major caching plugin offers it.

10. Self-host fonts and preload them

Calling fonts from Google's servers means an extra DNS lookup and connection to a third-party domain before any text renders. Self-host your web fonts locally (OMGF automates this, or do it manually), add font-display: swap so text shows immediately in a fallback, and preload the one or two font files used above the fold. Subset to only the character sets and weights you actually use — most sites ship five weights and use two.

Keep it lean over time (tactics 11–12)

11. Clean and optimize the database

Over months, your database fills with post revisions, expired transients, spam comments, and orphaned metadata that bloat queries. Run WP-Optimize or Advanced Database Cleaner to clear them, and add define('WP_POST_REVISIONS', 5); to wp-config.php to cap revisions going forward. A leaner database means faster queries, which directly improves TTFB on dynamic pages. Always back up before a cleanup pass.

12. Audit plugin bloat and rein in the Heartbeat API

Every active plugin is potential weight — extra queries, extra scripts, extra autoloaded options. Use the free Query Monitor plugin to see exactly which plugins are running slow queries or loading assets site-wide. Deactivate anything you can live without, and look for "single-purpose" plugins doing what a caching plugin already handles. Separately, WordPress's Heartbeat API pings the server every 15–60 seconds in the admin and editor; throttle or disable it (Heartbeat Control plugin) to cut needless server load, especially on busy admin areas.

How to actually measure progress

Don't trust a single lab score. Test with PageSpeed Insights for lab data and field (real-user) data, run a few passes through WebPageTest for a waterfall view of what's actually slow, and watch the Core Web Vitals report in Google Search Console for the trend across real visitors over 28 days. Field data is the only thing Google ranks on — a perfect lab score with poor field INP helps nobody.

Work top to bottom. The host, caching, and image fixes are where the real seconds live; the front-end and database work is what gets you from "fast enough" to genuinely fast. You almost never need to do all twelve at once — fix the foundation, re-measure, and only keep going if the numbers still aren't where you want them.

The 12 WordPress Speed Tactics That Actually Work