You ran your site through PageSpeed Insights, got a 34 in red, and a list of twenty warnings written in Google's language. Your web guy said "add a caching plugin". The score went to 41. The site still takes six seconds to show anything on a phone.
The problem with most advice on how to improve website speed is that it's a list of everything. Nobody tells you the order, and the order is the whole game. Four of the fixes below solve most slow business sites. The other eight are what you do after those four, if you still need to.
This is the order we use on the sites we build and the sites we're asked to rescue, with what each fix usually gains, what it costs, and which ones you can do yourself this afternoon.
Quick answer: To improve website speed, start with the four fixes that cut load time most: resize and compress images, remove plugins and scripts you don't use, move to faster hosting with a CDN, and stop loading fonts and chat widgets from third-party servers. Those four fix most slow business sites. Code-level work and rebuilds come after, not before.
Not sure which of the twelve applies to your site? Send us the link. We'll run it, read the field data and tell you the three fixes that matter for your pages, free. Book a free website and SEO audit.
Measure first: what "slow" means for your site
Before fixing anything, get two numbers from PageSpeed Insights on your busiest service page, not your home page. Test on mobile.
- LCP in the field data at the top of the report. That's how long real visitors waited to see the main content over the last 28 days. Under 2.5 seconds is Google's "good" line, per its Core Web Vitals guidance.
- The "Largest Contentful Paint element" in the diagnostics. That tells you what visitors are waiting for. It's almost always an image.
Ignore the big colored score for now. It's a lab estimate, and we've seen sites score 55 with perfectly good field data and sites score 90 while real visitors wait four seconds. If you want the full explanation of what each metric means and what it costs you, we wrote it up in our Core Web Vitals guide. This article is about the fixes.
The 12 fixes to improve website speed, ranked
Here is the whole list at a glance. Typical gains are what we see on business sites built on WordPress, Wix or an older custom build. Yours will vary, which is why you measured first.
| # | Fix | Typical gain on a slow site | Effort | Who does it |
|---|---|---|---|---|
| 1 | Resize and compress images | 1 to 4 seconds off LCP | Low | You, with an afternoon |
| 2 | Remove unused plugins and scripts | 0.5 to 2 seconds | Low | You or a developer |
| 3 | Faster hosting and a CDN | 0.5 to 2 seconds off first byte | Medium | Hosting move, a developer |
| 4 | Self-host fonts, drop third-party widgets | 0.3 to 1.5 seconds | Low to medium | Developer |
| 5 | Page and browser caching | 0.5 to 1.5 seconds for repeat and cached visits | Low | You, with one plugin |
| 6 | Preload the hero image, fix the LCP element | 0.3 to 1 second | Medium | Developer |
| 7 | Defer render-blocking CSS and JavaScript | 0.3 to 1 second | Medium | Developer |
| 8 | Replace hero videos and sliders | 1 to 3 seconds on mobile | Low | You and a developer |
| 9 | Remove redirect chains | 0.2 to 0.8 seconds | Low | Developer |
| 10 | Clean the database and update PHP | 0.2 to 1 second off first byte | Medium | Developer or host |
| 11 | Replace the theme or page builder | 1 to 3 seconds | High | Developer |
| 12 | Rebuild on a static or Next.js stack | Everything above, permanently | High | Agency |
Do them in order. Stop when your field LCP is under 2.5 seconds and your page responds to taps without lag. Most sites we work on stop somewhere between fix 5 and fix 7.
1. Images: the fix we do first on every site
Nine times out of ten, the LCP element is a photo, and that photo is a 4 MB file straight from a phone, displayed at 800 pixels wide.
The rule we use on our own blog: content images at 1200 pixels wide, hero and featured images at 1600, saved as JPEG at quality 75 or as WebP, under 150 KB each. That's not a compromise. On a phone screen nobody can tell the difference between that and the 4 MB original.
Do this yourself: export your ten heaviest images at those sizes and re-upload them. On WordPress, a plugin like ShortPixel or Imagify will convert the rest. Then make sure images below the fold are lazy-loaded and the hero image is not. Lazy-loading the hero is one of the most common ways a "speed plugin" makes a site slower.
2. Plugins and scripts you don't use
Every WordPress plugin can load its own CSS and JavaScript on every page, whether that page uses it or not. A contact form plugin, a slider, a social share bar, two SEO plugins, a table plugin from a page you deleted in 2022. Each adds requests, and the browser waits for all of them.
Go through the plugin list and deactivate anything you can't name a use for. Then look at the scripts: chat widgets, heat maps, three analytics tags, a Facebook pixel from a campaign that ended. Keep what you read weekly. Delete the rest.
3. Hosting and a CDN
If your site takes more than a second to send its first byte, nothing on the page can start. That's a hosting problem, and no plugin fixes it. Shared hosting at $5 a month is where most slow sites live.
Here's an example from our own setup. The WordPress backend behind our blog runs on a host that, on a bad day, takes about six seconds to answer a request. No visitor has ever waited for it, because the public site isn't served from WordPress. It's built into static pages and served from a CDN, so the slow backend only slows us down, never a reader. That's the same idea behind the Next.js vs WordPress decision, and it's why we push the public site onto a CDN even when the backend stays where it is.
If a rebuild isn't on the table, the cheaper version is managed WordPress hosting plus a CDN like Cloudflare in front of it. Expect $25 to $60 a month instead of $5, and expect the first byte to drop to a few hundred milliseconds.
4. Fonts and third-party widgets
Fonts loaded from Google Fonts, Adobe Fonts or a theme's CDN add a round trip to another server before any text can appear. We learned how fragile that is on our own office network, where Google Fonts requests time out often enough that our site would sometimes render with no fonts at all. Every site we ship now self-hosts its fonts, limited to two families, with `font-display: swap` so text shows immediately in a fallback font.
Widgets are worse. A live chat bubble, a reviews carousel from a third-party service, an Instagram feed and an embedded map can add a megabyte of JavaScript from four different companies. Load them after the page is interactive, or replace them with a static version: a screenshot of the map that links to Google Maps, three reviews as plain text with a link to the rest.
5. Caching
Caching lets the server hand a visitor a ready-made copy of the page instead of building it from the database every time. On WordPress, one good caching plugin (WP Rocket, or the one your host provides) is worth having. This is the plugin your web guy meant, and it helps. It just can't fix the four problems above, which is why it moved the score from 34 to 41 and not to 90.
Browser caching, set by the same plugin or the host, means repeat visitors don't re-download your logo and stylesheet. Cheap, and worth turning on.
"A caching plugin is a bandage. It's a fine bandage. It just doesn't work on a site that's bleeding 4 MB images."
6. Preload the hero image
Once images are the right size, tell the browser which one matters. The hero or featured image should be a plain `<img>` tag, not a CSS background, with `fetchpriority="high"` and a preload hint in the page head. Google's LCP optimization guide walks through it. This is developer work, and it typically buys a few hundred milliseconds on the metric Google cares about most.
7. Defer render-blocking CSS and JavaScript
By default the browser stops drawing the page while it downloads every stylesheet and script in the head. Deferring the scripts that aren't needed for the first screen, and inlining the small amount of CSS that is, lets the page draw first and load the rest after. Most caching plugins have a "delay JavaScript" option that does a rough version of this. A developer does it properly in an hour or two.
8. Replace hero videos and sliders
An autoplaying background video in the hero is the single heaviest thing a business site can have, and on a phone it's often a 10 MB download before the visitor sees your headline. Sliders are close behind: five full-width images loading at once so that one can be shown.
Replace the video with a good still photo and a play button if you must have the video. Replace the slider with one image and one sentence. You'll lose nothing. Nobody watches slide three.
What each fix costs
| Fix | Do it yourself | Freelancer or agency |
|---|---|---|
| Images (1) | An afternoon, free | $150 to $500 |
| Plugins and scripts (2) | An hour, free | $100 to $300 |
| Hosting and CDN (3) | $25 to $60 a month, a few hours to move | $200 to $600 for the migration |
| Fonts and widgets (4) | Not really | $150 to $400 |
| Caching (5) | One plugin, $0 to $60 a year | Included with most work |
| Preload, defer, redirects (6, 7, 9) | No | $300 to $800 together |
| Database and PHP (10) | Ask your host, free | $100 to $300 |
| New theme (11) | Weeks of your time | $1,500 to $5,000 |
| Rebuild (12) | No | $2,500 and up |
A full speed cleanup of an existing site, fixes 1 through 10, usually lands between $500 and $1,500 with a competent developer. That's the range our own audit and fix service sits in, and it's the number to compare against a rebuild quote.
Already paid for "speed optimization" and the site still crawls? We'll read the field data and show you what was and wasn't done, free. Get a free website and SEO audit.
Fix it or rebuild it?
Here's the rule we use. Do fixes 1 through 7. If mobile LCP in the field data is still above 4 seconds after that, the platform is the ceiling, and you're paying to polish something that can't get there.
Signs the platform is the problem:
- A page builder like Elementor or Divi with hundreds of nested elements on every page.
- A theme that loads its entire demo's assets whether you use them or not.
- A first byte over one second on managed hosting, which means the code itself is slow.
- A site you're afraid to update because the last update broke something.
At that point, a rebuild on a static or Next.js stack is usually cheaper than another year of patches, and the speed comes with the build rather than being bolted on. We covered when that trade-off pays in why custom Next.js builds beat generic templates. It's the core of what our custom web development team does.
If you're rebuilding anyway, that's the moment to fix the design as well, because a fast site that doesn't convert is just a fast site. Our web design services plan both at once.
What speed is worth to a business
You don't need a study to know a six-second site loses customers, but two numbers help make the case internally.
Google's own research found that more than half of mobile visitors leave a page that takes over three seconds. And when Vodafone cut LCP by 31%, sales went up 8% with no other change to the pages.
For a local service business, the math is simpler. If your quote form gets 20 submissions a month and a third of mobile visitors never see the page, the fixes above are worth roughly ten more quote requests a month. That pays for the whole list in the first month.
Speed also feeds rankings, both because Core Web Vitals is a ranking signal and because a faster site gets crawled more completely. That's why our technical SEO services start with the same twelve fixes.
A speed checklist you can run today
- Test your busiest service page on PageSpeed Insights, mobile. Write down the field LCP.
- Find the LCP element. If it's an image, resize it to 1600 pixels and under 150 KB.
- Resize the ten heaviest images on the site the same way.
- Deactivate every plugin you can't name a use for.
- Remove every tracking script and widget you don't read.
- Turn on a caching plugin, or your host's cache.
- Replace any hero video or slider with one image.
- Check that the hero image is not lazy-loaded.
- Ask your host what PHP version you're on. If it's below 8, ask them to update it.
- Test again. If field LCP is under 2.5 seconds, stop. If it's over 4, read the rebuild section.
What RankMeHi does about slow sites
- Audit and fix, from $490. We run the twelve fixes above on your existing site, in that order, and report the before and after numbers from field data rather than a lab score. Most WordPress and Wix sites finish under 2.5 seconds on mobile.
- New website, from $2,490. Five pages on Next.js, images sized right from the start, fonts self-hosted, no plugins to bloat. Fast on day one and still fast in year three.
- Ongoing care. Speed decays as people add plugins and upload phone photos. Our website maintenance services include a monthly speed check so it doesn't.
Prices are on our pricing page. When you don't need us: if your field LCP is already under 2.5 seconds on mobile, stop reading speed articles and spend the money on content or photos. Your site is fast enough.
Key takeaways
- Measure field LCP on a service page before touching anything. The colored score is not the metric.
- Four fixes solve most slow sites: images, unused plugins and scripts, hosting with a CDN, and third-party fonts and widgets.
- Images are almost always the LCP element. 1200 to 1600 pixels wide and under 150 KB is the target.
- A caching plugin helps but can't fix the four problems above, which is why it barely moved your score.
- If mobile LCP is still over 4 seconds after fixes 1 to 7, the platform is the ceiling. Rebuild instead of patching.
- A full cleanup costs $500 to $1,500. A rebuild starts at $2,500. Compare those against what ten more quote requests a month is worth.
FAQ
How can I improve website speed for free?
Resize your images to 1200 to 1600 pixels wide and under 150 KB, deactivate plugins you don't use, remove tracking scripts and widgets you don't read, turn on your host's caching, and replace any hero video or slider with a single photo. Those five steps are free and fix most slow business sites.
What is a good website load time?
Google's "good" threshold for Largest Contentful Paint is 2.5 seconds, measured from real visitors on mobile. Under that, speed stops being a ranking or conversion problem. Between 2.5 and 4 seconds needs improvement, and over 4 seconds is costing you visitors and rankings.
Why is my website slow even with a caching plugin?
Caching serves a ready-made copy of the page, but it can't shrink a 4 MB image, remove a chat widget or speed up a slow server. If the site is still slow with caching on, the problem is almost always images, third-party scripts or hosting, in that order.
Does website speed affect SEO?
Yes. Core Web Vitals is a Google ranking signal, and slow pages get crawled less completely. The bigger effect is on people: more than half of mobile visitors leave a page that takes over three seconds, so a slow site loses the customers it does rank for.
How much does website speed optimization cost?
A full cleanup of an existing site, covering images, scripts, hosting, fonts, caching and code fixes, costs $500 to $1,500 with a competent developer. If the platform itself is the ceiling, a rebuild on a modern stack starts around $2,500 and makes the speed permanent instead of patched.
Do the four, then measure
Most slow sites don't need twelve fixes. They need the first four, done properly, and a measurement afterward to prove it. Start with the images this afternoon, and if the site is still slow next week, you'll know exactly which section of this article to come back to.
Want the twelve fixes done for you, with before and after numbers? Send us your site and we'll show you where the seconds are going, free. Request your free website and SEO audit.
Services mentioned in this article
RankMeHi designs and builds business websites, then makes sure they are found and convert. Start with the two pages below.
Keep reading
Core Web Vitals Explained: What Google's Speed Scores Mean for Your Business (and How to Fix Them)
Core Web Vitals explained for business owners: what LCP, INP and CLS mean, the scores Google wants, how to test your site, and the fixes that matter most.
Read article
Website Maintenance Services: What's Included, What It Costs, and What Happens If You Skip It
What do website maintenance services include, and what should they cost? See 2026 prices, a monthly checklist, and how to choose a plan that protects leads.
Read article
Next.js vs WordPress for Business Websites: Which One Actually Gets You More Leads?
If you’re a local business owner and your website isn’t bringing in calls or inquiries, you don’t really care which tech stack is “cooler.” You care about one thing: “Which website platform is going to help me get more leads and customers… without turning into a nightmare to manage?” Two of the most common options …
Read article
Why Custom Web Development in Next.js Beats Generic Templates for Local Businesses
If you’re a local business owner, you’ve probably heard some version of this: “Just grab a template, throw up a WordPress site, and you’re good to go.” You do that.You pay someone (or spend evenings yourself), you launch the site… and nothing really changes. The problem usually isn’t that you “don’t have a website.”The problem …
Read article