Why Is My WordPress Site So Slow? 15 Real Causes & Fixes (2026)

1. Measure Before You Fix

Before you touch a single plugin, measure. Optimizing blind is how people waste weekends chasing the wrong problem. Open any of these tools, enter your URL, and write down three numbers:

• TTFB (Time to First Byte) — how long your server takes to respond after the browser asks for a page. Under 200ms is excellent. 200–500ms is fine. Over 800ms means your host is the bottleneck. Measured by PageSpeed Insights and web.dev.
• LCP (Largest Contentful Paint) — how long before the hero element paints. Under 2.5s passes Google's Core Web Vitals. Over 4s is a ranking liability.
• Total page weight — how many megabytes load. The HTTP Archive median is ~2.3 MB. If your page is 6 MB, images are likely the problem.

Two diagnostic tools you'll reach for daily: GTmetrix shows a waterfall so you can see which request is blocking, and our free PageSpeed tool gives you the same Lighthouse scores Google uses.

The rule: if TTFB is bad, your site is fast — your host is slow. Close the plugin optimization checklists. Move to better hosting first, then optimize on top of it. You'll save yourself 20+ hours of tinkering.

2. The #1 Cause — Your Host

Cheap shared hosting ($2–3/month from the biggest brands) is designed around one thing: packing thousands of sites onto a single overloaded server. Here's what "oversold" actually looks like under the hood:

• Spinning HDD or consumer SSD storage. Every database query waits on disk I/O. NVMe is 6–10× faster than SATA SSD and 50–100× faster than HDD for the random-read workload WordPress generates. See NVMe on Wikipedia for the technical spec.
• Apache with no object caching. Every page view re-queries MySQL from scratch. A single homepage can trigger 80–200 database queries on a bloated WordPress install.
• CPU and RAM throttled to the minimum. The moment you get a traffic spike, you're throttled — not upgraded. Many hosts return HTTP 503 "Service Temporarily Unavailable" the moment you exceed your plan's CPU allotment.
• Shared MySQL servers. A noisy neighbour's slow query tanks your site. You have no visibility into this and no way to fix it.
• PHP 7.4 or older. Some cheap hosts still run end-of-life PHP. The PHP release schedule makes clear that 7.x has been unsupported since November 2022.

This is why the same WordPress install, on the same theme, with the same plugins, can load in 500ms on one host and take 6 seconds on another. The WordPress code is identical. The infrastructure isn't. See our deep-dive on shared hosting CPU limits for what those throttles actually look like under load.

3. The 15 Fixes, Ranked by Impact

1. Move to NVMe + LiteSpeed hosting

This is the single biggest fix and it dwarfs everything below it. NVMe SSDs are 6–10× faster than SATA SSDs for the random-read workload WordPress generates. LiteSpeed with LSCache outperforms Apache by 3–5× on PHP workloads and has a native WordPress caching plugin that needs zero config. Our Growth plan includes both by default.

2. Install a caching plugin (LSCache or WP Rocket)

If you're on LiteSpeed hosting, install LiteSpeed Cache — it's free and drops TTFB by 80–95% for logged-out visitors. On Apache/Nginx hosts, WP Rocket is the best paid option. W3 Total Cache is a capable free alternative if you can tolerate its UI.

3. Put Cloudflare (or another CDN) in front

Cloudflare's free tier is enough for most sites. It serves cached HTML and all your images from 320+ edge locations, which is why a site loading in 400ms from Germany might load in 2s from Australia without a CDN. Turn on APO (Automatic Platform Optimization for WordPress) for another 30–50% gain.

4. Compress and properly size images

A 4MB unoptimized hero image is the single most common page-speed killer. Use an image optimization plugin (ShortPixel, Smush, or the LSCache image optimizer) to convert to WebP or AVIF and strip metadata. Serve correctly sized images — a 2000px-wide photo displayed at 800px still downloads the full 2000px file unless you use srcset. See MDN on responsive images.

5. Audit your plugins

Every plugin adds database queries and HTTP requests. Deactivate anything you don't use. The biggest offenders: social share plugins that load jQuery on every page, related-posts plugins that do a full-text query on every request, and analytics plugins that are duplicating what Google Tag Manager already does. Use Query Monitor to see which plugins are firing the slowest queries.

6. Use a fast theme

Multipurpose themes (Avada, Divi, older Elementor builds) can add 3–5 MB of CSS/JS per page load. Modern lightweight themes like GeneratePress, Kadence, or Blocksy load under 100 KB and still look polished.

7. Optimize the database

WordPress databases bloat over time with post revisions, transients, and orphaned metadata. Run WP-Optimize or Advanced Database Cleaner monthly. Delete post revisions older than 30 days. Set WP_POST_REVISIONS to 5 in wp-config.php.

8. Upgrade to PHP 8.3

PHP 8.3 is 30–45% faster than PHP 7.4 on WordPress workloads. If your host is still on 7.x, that's another sign you need a new host. Check compatibility first with the PHP Compatibility Checker.

9. Enable HTTP/2 or HTTP/3 (QUIC)

HTTP/3 multiplexes requests and eliminates head-of-line blocking. Any modern host with LiteSpeed has it on by default. Check yours with http3check.net.

10. Disable WordPress heartbeat on the frontend

The admin-ajax heartbeat fires every 15 seconds and can spike CPU on shared hosts. Use Heartbeat Control to disable it except on the post editor.

11. Lazy-load images and iframes

WordPress 5.5+ lazy-loads images natively (loading="lazy"), but iframes still load eagerly in many themes. Add loading="lazy" to YouTube embeds, Google Maps, and anything else you embed.

12. Minify and defer JavaScript

Render-blocking JS is the most common Core Web Vitals flag. LSCache and WP Rocket both do this automatically. Test after each setting — some themes break with aggressive JS deferral.

13. Kill third-party scripts you don't need

Facebook pixel, Hotjar, Intercom, live chat, five different analytics tools — each one adds 100–400 ms. Audit with DebugBear or the PageSpeed "Reduce the impact of third-party code" flag. Remove anything that isn't directly driving revenue.

14. Enable OPcache and object caching

OPcache caches compiled PHP bytecode — on by default on any decent host. Object caching (Redis or Memcached) caches the results of database queries. LSCache includes native object cache support. On Growth and Scale, we ship Redis preconfigured.

15. Use a staging site, not production, to test optimizations

Every optimization above can break something. A staging site lets you swap the cache config, upgrade PHP, change themes — without risking your live traffic. All LaunchPad plans include one-click staging.

4. Cheap Host vs Quality Host — The Numbers

Below is a real benchmark of the same WordPress install (Astra theme, 12 plugins, 80 posts) on two different hosts. Tested with GTmetrix from a London test location, averaged over 5 runs:

Same site, same plugins, same images. The only variable is the infrastructure. This is why the page-speed checklists floating around SEO Twitter so often fail: they assume the host is fine.

5. When to Stop Optimizing and Switch Hosts

If you've done fixes 2–15 and TTFB is still above 800ms, your host cannot be optimized further. Shared hosts with oversold servers physically cannot deliver sub-200ms TTFB no matter what you do inside WordPress.

The signals it's time to switch:

• TTFB stuck above 800ms even on a brand-new default theme with no plugins
• Frequent "508 Resource Limit Reached" or "503 Service Unavailable" errors
• Your traffic is growing and the host keeps suggesting upgrades that don't fix the speed
• Support tickets take 24+ hours and get replies like "please optimize your site"
• You're on a $3/mo plan but your renewal quote is $15–$30/mo

Moving hosts is easier than it used to be. We handle the migration from Bluehost, GoDaddy, Hostinger, SiteGround, and others for free — you stay on your current host until the new site is tested, then we flip DNS. Typical downtime: zero. See also our comparisons of LaunchPad Host vs Bluehost and vs Hostinger.