Speed and Core Web Vitals for a clinic site

Why speed matters here

Two reasons, and the second is larger than the first. Speed is a ranking consideration, modestly. And a slow page loses readers before they read anything, which matters much more, particularly on a phone on mobile data, which is how a large proportion of clinic traffic arrives.

The practical framing for a clinic is that speed work is worth doing once, properly, and then protecting. It is not worth doing continuously, and the industry habit of chasing a score is a distraction from the modules that produce more.

The three metrics

Largest Contentful Paint measures how long until the largest thing on the screen appears. On a clinic page that is nearly always the hero image or the first heading. Slow LCP is almost always an image problem.

Interaction to Next Paint measures how long the page takes to respond when somebody taps something. Slow INP is almost always a script problem: too much JavaScript running on the main thread, usually from third parties.

Cumulative Layout Shift measures how much the page moves about while loading. It is the metric that produces the experience of tapping a button and hitting an advertisement that appeared underneath your thumb. It is caused by media without declared dimensions, by fonts swapping, and by content injected after load.

Field data against laboratory data

A laboratory test runs your page once, on a simulated device, on a simulated connection. Field data records what actual visitors experienced. They frequently disagree, and when they do, the field data is the one that matters.

Search Console's Core Web Vitals report shows field data grouped by page type, which is the most useful view available to a clinic because it tells you whether the problem is your treatment template or your homepage. Use the laboratory tools for diagnosis, after the field data has told you where to look.

Six fixes, in order

1. Resize images before uploading. A photograph straight from a camera is several times larger than any web page needs. This single change fixes more clinic sites than the other five combined.
2. Serve modern image formats and compress properly. Most platforms will do this if asked once.
3. Set width and height on every image and video. This is what prevents layout shift, and it takes minutes.
4. Load the hero image eagerly and everything else lazily. Lazy-loading the image at the top of the page is a common mistake that makes LCP worse.
5. Load fonts without blocking. Use a display strategy that shows text immediately, and do not load six weights of two families when you use three weights of one.
6. Remove third-party scripts you cannot justify. Covered next, because it is the one that requires a decision rather than a technique.

web.dev's guidance on optimising Largest Contentful Paint is the reference for the first four, and it is written for people who are not specialists.

Third-party scripts

A typical clinic site accumulates: an analytics tag, a tag manager, a chat widget, a booking widget, a review widget, a social feed, a cookie banner, a heat-mapping tool somebody installed during a trial, and a remarketing pixel from an advertising campaign that ended two years ago.

Each was added for a reason and most of the reasons have expired. Every one of them costs loading time and several of them cost interaction responsiveness. The exercise below is an audit with a simple rule: anything nobody can name a current use for is removed.

Every third-party script is a permanent tax on every page load, paid by every visitor, in exchange for a benefit somebody assessed once and never revisited.

Chat widgets deserve a specific note. They are heavy, they frequently sit in front of the content on a phone, and they are only worth their cost if somebody is actually answering them quickly. A chat widget nobody staffs is worse than no chat widget, a point Module 30, Enquiry forms, phone and messaging returns to.

When to stop

When all three metrics are in the good band on field data for your main page types. That is the finish line, and there is no prize for going further.

Protect it rather than improving it: check the field data monthly as part of the review in Module 31, What to measure, and what to ignore, and check it specifically after adding any new tool. Most regressions on clinic sites are caused by somebody adding a widget, which is why the audit is worth repeating annually rather than the optimisation.