How to make your website faster (and why rankings follow)
Page speed advice usually arrives as a list of thirty things, which is why most of it never gets done. In practice, four things account for the large majority of slow pages, and fixing them in order gets you most of the benefit for a fraction of the effort.
Before that, though, the honest framing of why it matters.
Speed is a conversion problem that happens to help rankings
Google does use Core Web Vitals as a ranking signal. It is real, and it is modest. A slower page with a better answer will routinely outrank a faster page with a worse one, and any consultant who tells you speed alone will move you up the results is selling the easy part of the job.
The stronger reason to care is what happens before ranking matters at all. People leave slow pages. On an Indian mobile connection, a page that takes six seconds has lost a meaningful share of its visitors before they have read a word — including visitors you paid to acquire.
That is the argument. You are not optimising for an algorithm. You are stopping traffic you already have from leaking away.
What to measure, and where
Ignore any single score from a one-off test on office wifi. It measures a laboratory, not your visitors.
Use Google Search Console → Core Web Vitals. It reports what real users on real devices and real networks actually experienced, at the 75th percentile — meaning three in four visits were at least this good.
Three numbers:
- Largest Contentful Paint (LCP) — how long until the main content appears. Target under 2.5 seconds. This is the one to fix first.
- Interaction to Next Paint (INP) — how quickly the page responds when someone taps. Target under 200 milliseconds. Usually a JavaScript problem.
- Cumulative Layout Shift (CLS) — how much the page jumps about while loading. Target under 0.1. Usually images or ads without reserved space.
PageSpeed Insights is still useful for diagnosis — it tells you what is slow. Search Console tells you whether it matters.
Fix one: images
On most sites this is more than half the problem, and it is the least technical thing on the list.
Serve modern formats. WebP or AVIF instead of JPEG and PNG. Typically 25–50% smaller at the same visual quality.
Serve the right size. A 4000-pixel-wide photograph displayed in a 600-pixel column is downloading roughly forty times more data than it needs. This is the most common single mistake on business websites, and it is usually made by someone uploading straight from a camera or a designer’s export.
Set width and height on every image. This costs nothing and eliminates most layout shift, because the browser can reserve the space before the image arrives.
Lazy-load everything below the fold — but never the main image at the top, which is usually your LCP element. Lazy-loading that one makes the score worse.
Fix two: third-party scripts
The second biggest cause, and the least examined, because each one was added by someone with a good reason.
Chat widget. Analytics. Meta pixel. Google Ads tag. A heatmap tool someone trialled in 2023. Embedded fonts. A review widget. A cookie banner. Each looks small. Collectively they are frequently heavier than the entire site.
The exercise, once a year:
- List every third-party script on the site
- For each, name the person who looks at its data and when they last did
- Remove the ones with no answer
You will remove more than you expect. Nothing that matters will break.
For the survivors: load them after the page renders rather than before, and load chat widgets on interaction rather than on page load. A chat widget that appears half a second later costs you nothing and saves everyone else a great deal.
Fix three: fonts
Custom fonts are worth having and cheap to get wrong.
Self-host them rather than calling Google Fonts, which removes a connection to another domain. Use font-display: swap so text appears immediately in a fallback rather than staying invisible while the font downloads. Subset to the characters you actually use. And load two weights, not six — most sites use two and download five.
Fix four: hosting and platform
The floor under everything else.
Server response time — how long before the server sends anything at all. Cheap shared hosting frequently takes 800 milliseconds or more, and no front-end optimisation can recover that. This is worth paying to fix.
Caching — pages served ready-made rather than assembled per visitor. Most platforms handle this; some need a plugin.
A CDN — worth it if your visitors are spread across the country or internationally. Modest if they are all local and your server is too.
Static where possible. A site that ships pre-built HTML rather than assembling each page on request removes an entire category of problem. It is not right for every project, but for marketing sites and blogs it is usually both faster and cheaper.
The order to do this in
- Fix images — the largest gain for the least skill
- Audit third-party scripts and delete what nobody reads
- Sort out fonts
- Fix server response time, upgrading hosting if that is what it takes
- Re-measure in Search Console after 28 days, because that is the window it reports on
Stop when the numbers are green. Perfect scores are a hobby, not a business objective.
When optimisation is the wrong answer
Some sites cannot be rescued. A heavy page builder generating deeply nested markup, a theme bundling a dozen libraries, thirty plugins each adding scripts — the weight is structural, and every fix is fighting the platform.
The test: could a competent developer get this site to good Core Web Vitals without removing features you actually need? If not, you have a rebuild, and continuing to optimise is spending money to stay in the same place.
That is an unwelcome answer, and it is often the true one.
What speed will not fix
A fast page with a weak offer converts slightly better than a slow page with a weak offer. Speed removes a reason to leave; it does not supply a reason to stay.
If your bounce rate is high and your pages load in two seconds, the problem is not speed. It is the message, the offer, or the fact that the people arriving were never the right people.
Page speed is like a clean shop floor. Nobody buys because of it. Plenty of people leave because of its absence.
Frequently asked questions
Directly, only a little. Core Web Vitals are a genuine ranking signal but a modest one, and a slow page with the best answer will usually outrank a fast page with a worse one. The stronger argument is commercial: slow pages lose visitors before they read anything, so the traffic you already have converts less. Fix speed for conversion, and take the ranking benefit as a bonus.
Judge it by Core Web Vitals on real users rather than a single number. Largest Contentful Paint under 2.5 seconds, Interaction to Next Paint under 200 milliseconds, and Cumulative Layout Shift under 0.1, measured at the 75th percentile of real visits. That last part matters — your visitors are on Indian mobile networks, not your office connection.
Almost always some combination of four things: unoptimised images, too many plugins each adding scripts, a page builder generating heavy markup, and cheap shared hosting with slow server response. In that order of likelihood. A site with thirty plugins is rarely rescued by a caching plugin, which is what most people try first.
If your visitors are all in one region and your hosting is in that region, the benefit is modest. If you serve visitors across India or internationally, a CDN meaningfully reduces the time to first byte for anyone far from your server. Most modern hosting platforms include one, which makes it a question you often do not have to answer separately.
Fix it if the foundations are sound and the problems are images, scripts and hosting. Rebuild if the platform itself generates the weight — a heavy page builder, a theme with a dozen bundled libraries, or a stack that cannot serve static HTML. The test is whether an expert can get you to a good score without removing features you need.
Want a second opinion on your marketing?
A discovery call costs nothing and you leave with something useful either way.
Book a discovery call