← Blog
Web Development
November 13, 2026 · 8 min read

Why a slow site loses customers: understanding and improving Core Web Vitals

A page that opens late, reacts slowly to a tap or jumps while being read loses visitors before they see what you offer. Here is what Core Web Vitals measure, how to check your own site and what to fix first.

Web Development

When a visitor clicks through to your site, the first thing they meet is not your design or your copy but the wait. If the page opens late, reacts slowly to taps or shifts while they read, they usually go back without seeing what you offer. In this post we explain why speed affects your business, what Core Web Vitals measure, how to check your site with free tools and where to start improving.

Why does a slow site lose customers?

To a visitor, a slow site is like a shop with a door that sticks: what is inside stops mattering. For someone arriving from a search result or an advert, the back button is a thumb away and the next result is often your competitor. If a visitor you paid to bring in leaves before the page opens, you have paid for that click and received nothing in return.

The second effect is search visibility. Google counts page experience among the factors it considers in ranking, and measures that experience largely through Core Web Vitals. But speed alone will not carry your site to the top; how well the content answers the question being searched, and how strong the competition is, matter far more. Speed is only one factor among many. What matters most is that it directly affects whether the visitor who does arrive will stay.

Core Web Vitals: three measures, plainly

Core Web Vitals is the shared name for three measurements Google chose to capture how a page feels to the person using it. Each answers a different question in the visitor's mind: Has the page opened? Does it react when I tap? Does the content stay put?

  • LCP (Largest Contentful Paint) measures when the main content appears. When the largest element on the page arrives, usually the big image or headline at the top, the visitor feels the page has opened.
  • INP (Interaction to Next Paint) measures how quickly the page reacts to a tap. However long it takes for something to change on screen after you touch the menu, a button or a form field, that is INP.
  • CLS (Cumulative Layout Shift) measures how much the layout moves while the page loads. If you were about to tap a link when an image or advert appeared above it and pushed everything down, you have experienced a layout shift.

Google's thresholds for good

LCP should be 2.5 seconds or less, INP 200 milliseconds or less, and CLS 0.1 or less. A page counts as good when these values hold for the large majority of visits to your site.

Measure your own site with free tools

The first tool is PageSpeed Insights. You type your page's address into this free Google tool and it returns separate reports for mobile and desktop. The report shows Core Web Vitals collected from your real users, if there are enough visits, along with the results of a trial load the tool runs on the spot and a list of suggested improvements. Test not just the home page but each page that brings in sales or enquiries; a product page can behave very differently from the home page.

The second tool is the Core Web Vitals report in Search Console. Search Console is the free panel Google offers to site owners. This report looks at the whole site rather than a single page: it groups your pages, separately for mobile and desktop, as good, needing improvement or poor. That shows you whether the problem sits on one page or, say, on every product page.

In these tools you will see two kinds of data, and they often disagree. Lab data is a single trial the tool runs at that moment, imitating a particular device and connection speed. Field data is collected from real visitors who browsed your site with Chrome over recent weeks; old phones and weak connections are part of it. So base decisions on field data, and use the lab test to find the cause of a problem and to see at once whether a change has worked. If your site is new or has few visits, field data may not appear. That is not an error; there is simply not enough data yet.

The most common causes of slowness

On slow sites the problem is usually not one big mistake but small loads that have built up over time.

  • Oversized images. A photo taken on a phone and uploaded as it is carries many times more data than the size it is shown at. This is the most frequent reason for a poor LCP.
  • Too many third-party scripts. These are small programs added to your site from other companies' servers: analytics tools, advertising pixels, live chat boxes. Each looks harmless; together they keep the browser busy and delay its reaction to taps.
  • Heavy fonts. Every font family and every weight is a separate file. As the number grows, text appears late or lines jump the moment the font loads.
  • Slow hosting. If the server is late with its first response, everything else is added on top of that delay.
  • Images, adverts and embedded content without reserved space. If the browser does not know in advance how much room an image or advert will take, it inserts it afterwards and everything below moves.

What should you fix first?

Trying to fix everything at once is expensive and stops you seeing which change made the difference. We follow the order below, because the early steps usually bring the biggest gain for the least effort.

  1. 1

    Measure, and choose the pages that matter most. Note the field data for the few pages that draw the most visits and bring in revenue; without a starting value you cannot show an improvement.

  2. 2

    Shrink the images. Resize them to the size they are displayed at, serve them in modern compressed formats, and have those below the visible screen load only when the visitor scrolls to them.

  3. 3

    Reserve space for images, adverts and embedded content. When width and height are declared in advance, layout shifts largely disappear.

  4. 4

    Weed out third-party scripts. List them all and ask of each one: is anyone still using this? Remove the unused ones and have the rest load after the main content.

  5. 5

    Simplify the fonts. Reduce the number of families and weights, and have the text appear in a fallback font until the real one arrives.

  6. 6

    Deal with hosting and caching. If the server response is still slow, move to a stronger hosting plan and cache the pages, which means serving a ready copy rather than rebuilding it on every visit.

  7. 7

    Measure again. The lab test shows the change immediately; field data takes weeks to catch up.

A fast site is not built by one big optimisation, but by asking “do we really need this?” of every new thing added to the page.

What to ask your developer or agency

You do not need to know the technical detail; asking the right questions is enough. You should be able to get clear answers to these:

  • How do our Core Web Vitals look in field data, on mobile? Which groups of pages fall outside the thresholds?
  • Are the images we upload through the admin panel resized and compressed automatically, or do we have to do that ourselves every time?
  • Is speed part of the acceptance criteria? Which pages will be tested before launch, and under which device conditions?
  • After launch, who will monitor these values, how often, and who steps in when they slip?

If the answers are vague, or all you are shown is a screenshot of the lab score, it is worth pressing the point. On web projects we treat performance and accessibility tuning as an ordinary part of the job. When we take over an existing site, we likewise measure where it stands first and write up what we found in a short report.

Keeping the site fast after launch

Sites are usually at their fastest on the day they go live. Then a new campaign pixel is added, a video goes on the home page, photos taken on a phone are uploaded to blog posts. None of these is a problem on its own; but a few months later the site has slowed down without anyone noticing.

A few simple habits are enough to prevent this. Open the Search Console report once a month and check whether the number of pages in good standing has fallen. Before adding a new tool or script, write down what it is for and when it will be removed; when the campaign ends, take the pixel out too. After a major design change, measure the most important pages again. If you have a maintenance agreement, ask whether speed monitoring is in scope.

Site speed is not a job done once and finished; it is part of maintaining the site. Measure first, start with the most visited pages and the easiest fixes, then measure again. Speed alone does not guarantee your search ranking, but it does keep the visitor around long enough to see what you offer. The rest is up to your content and your product.

Work with us

Have a project in mind?

Describe your idea in a few sentences. We'll review it and come back shortly with scope, timeline and budget.