The Score You Are Chasing Is Not the One Google Uses

On most sites we examine, three things account for the large majority of it. Images uploaded at camera resolution, often several megabytes each, where a properly sized and compressed version would be a fraction of that. Third-party scripts that accumulated over years: a chat widget, a tag manager, two analytics tools, a pixel somebody added for a campaign that ended in 2023, all loading on every page. And a theme that loads assets for features your site does not use, which is common with multipurpose themes sold with dozens of demo layouts. Hosting and server configuration matter too, and people usually assume that is the answer when it rarely is the first one. The only way to know for your site specifically is to look, which is what the diagnosis tier is for.

Google defines good as Largest Contentful Paint of 2.5 seconds or less, Interaction to Next Paint of 200 milliseconds or less, and Cumulative Layout Shift of 0.1 or less. The part that matters more than the numbers is how they are assessed. Google measures at the 75th percentile of page loads, so three quarters of real visits must meet each target. Mobile and desktop are assessed separately, and a page passes only if all three metrics hit their targets, not on average. Crucially this is field data from real users, not a lab test. The 0 to 100 score in PageSpeed Insights is a separate thing: a simulation of one load on a modelled device. It is useful for diagnosis and it is not what Google assesses, which is why a site can show 94 in the lab and fail in the field. Our Core Web Vitals optimization services target the field measurement, because that is the one with consequences attached.

Yes, and less than most people selling speed work imply. We would rather quote Google than characterise it. Google confirms its ranking systems do use Core Web Vitals. It also says there is no single signal, that it always seeks to show the most relevant content even if the page experience is sub-par, and that good Core Web Vitals scores do not guarantee pages will rank at the top. Google goes further and cautions that chasing a perfect score purely for SEO may not be the best use of your time. The fair summary is that speed is a genuine but secondary factor, mattering most as a differentiator when several genuinely useful pages compete for the same query. If your pages are not ranking because the content does not match what people are searching for, speed will not rescue them, and our technical SEO service is the more relevant conversation.

Three to five days for a diagnosis, two to four weeks for a full optimization, and five to eight weeks for theme or build level work. Then there is a part nobody controls: field data updates over time because it reflects accumulated real visits, so the measurement that actually matters moves gradually over weeks rather than the moment the work finishes. This is why we stay with the project through the re-measurement period instead of closing it when a lab score turns green. If a provider shows you an improved result on the same afternoon the work was done, they have shown you a lab simulation, not a change in what Google assesses.

It helps, and it is not a strategy. Caching means the server does not rebuild the page for every visitor, which is a real gain and often the easiest one available. What it does nothing about is the size of what gets delivered once the cache has done its job. A cached page still sends the 4MB hero image, still loads the eleven third-party scripts, and still carries whatever assets the theme insists on. We configure caching properly on every engagement because it would be odd not to, and then we deal with the things caching cannot touch, which is where most of the remaining time is.

The honest risk is real and it is manageable. Aggressive optimisation can break things: deferring the wrong script stops a form working, removing an unused stylesheet turns out to be used on one page nobody checked, and over-eager image conversion damages quality. We work on a staging copy where one exists, change things in stages rather than all at once, and test the pages that matter after each stage, particularly forms and anything involving payment. We also document what was changed, so if something surfaces three weeks later it can be traced rather than guessed at. Any provider who tells you there is no risk has either not done much of this or is not being straight with you.

Because those are two different measurements and the one that matches your experience is the one Google uses. The score is a lab simulation: one load, a modelled device, a modelled connection. Your actual visitors are on real phones with real connections, sometimes poor ones, and the field data captures that spread at the 75th percentile. A site tuned to score well in the lab can still deliver a slow experience to the quarter of visitors on weaker connections or older devices. There is also a specific technical gap worth knowing: lab tools cannot measure Interaction to Next Paint directly, because there is no real user input to measure, so they report Total Blocking Time as a proxy instead. A good proxy is still a proxy.

Yes, and WordPress being slow is almost always a consequence of what has been installed on it rather than the platform. The usual work is a plugin load review to find which ones load on every page when they are needed on one, database cleanup where years of revisions and transients have accumulated, theme asset handling so the site stops loading code for features it does not use, plus the universal image and script work. Where we run into a wall is a heavy multipurpose theme that cannot be made light without replacing it, and in that case we say so rather than billing for incremental gains that cannot reach the target. Our WordPress website design page covers the platform decisions in more detail.

Sometimes, and it is oversold. A content delivery network serves your files from a location near the visitor, which genuinely helps when your audience is geographically spread or far from your server. If your hosting is in India and essentially all of your visitors are in India, the gain is modest and there are usually cheaper wins available first. If you have visitors in the Gulf, Europe or North America, it matters considerably more. We will tell you which situation you are in rather than adding it to every quote, and if you already pay for one we will check it is actually configured to do anything.

We will give you an estimate after the diagnosis and not before, because a credible number requires knowing what is wrong. What we can say in general is that the gains are largest where the starting point is worst, which sounds obvious and has a practical implication: a site failing badly often improves dramatically for a modest fee, while a site already passing Core Web Vitals may cost more to improve further and gain less from it. That is also the situation where Google’s own caution applies most directly, that chasing a perfect score purely for SEO may not be the best use of your time.

Often not, and we would rather say so on the page than during a sales call. If your site receives a few hundred visits a month, making it 40 percent faster affects a small number of people and will not show up in your business. Below a certain traffic level Google may not even have collected enough field data to report on your site, which is itself a signal about priority. In that situation the money usually does more spent on being found at all, and our technical SEO service or content work is the better purchase. Speed optimization earns its place when enough people are already arriving for their experience to be worth improving.

Interaction to Next Paint measures how quickly a page responds visually when someone interacts with it, and the threshold for good is 200 milliseconds or less. It replaced First Input Delay, which only measured the delay before the browser began processing the very first interaction on a page. That made FID relatively easy to pass while still allowing a page to feel sluggish throughout the rest of the visit. INP looks at interactions across the whole visit and reports on the worst of them, which is a much better reflection of how a page actually feels to use, and a noticeably harder bar. If your site passed Core Web Vitals comfortably a couple of years ago and does not now, this change is frequently the reason.

Yes, and Search Console is the right place to be looking, because it reports field data on real visits rather than a lab test. It groups URLs by issue, which usually reveals that the problem is a template rather than individual pages: all your blog posts fail on CLS, or every product page fails on LCP. That is useful, because fixing one template fixes hundreds of URLs. The thing to expect is patience. After the fix, Search Console needs to accumulate new field data before the status changes, so validation takes weeks rather than days and is not a sign nothing happened.

Ask what they will measure. A website performance optimization agency that answers field data at the 75th percentile, mobile and desktop separately, is describing what Google actually assesses. One that answers with the PageSpeed Insights score is going to optimise for a simulation. Then ask whether they will tell you if the work is not worth doing at your traffic level, because anyone willing to say that is not simply selling hours. If you want to hire page speed expert India wide, those two answers will separate the field faster than any portfolio.

Yes. We are based in Vikaspuri in West Delhi, so as a page speed expert Delhi NCR based we can meet in person across Delhi, Gurugram, Noida and Faridabad. Beyond that we work remotely across India, the UAE, the UK and the US. This particular service is almost entirely remote by nature, since the diagnosis comes from field data and crawling, the work happens on the server, and the verification is a measurement rather than a meeting.