WordPress · 1 August 2026 · 9 min
WordPress Core Web Vitals: the speed work that actually protects rankings
WordPress Core Web Vitals: the page speed work on LCP and INP that protects rankings on templates people actually buy from, not a vanity lab score.
WordPress Core Web Vitals fail for ordinary reasons: a page builder nesting sections, a hero stored as a background inside a slider, three font families, and a plugin that prints a script on every URL, including the one you meant to keep light. Rankings are not a prize for a green laboratory badge. Field data on the templates that take the enquiry is the constraint. A slow template caps a page that deserves the click. A fast one will not rescue a page that says nothing a buyer can use.
Decide the budget before the theme is chosen
Agree a weight budget before design sign-off, not after the homepage has been animated into a corner. Set a ceiling for the homepage and for the service or product template: what the document, the CSS, the JavaScript and the LCP image may weigh, and which third parties are allowed on the page at all. If a visual idea needs a heavy script to move a headline a few pixels, it does not ship.
A block theme and a short plugin list beat a child theme of a multipurpose demo almost every time. Refusing the demo is cheaper than spending a quarter deleting it. Write the budget as checks you can see in a trace: a ceiling for script on the service template, a ceiling for the hero image, and a named list of third parties whose absence is not an emergency. "As fast as possible" is how the slider returns at the next redesign.
Also decide who may add a plugin. Tags arrive as a harmless pixel and then listen to every click. New tags go through the person who owns page speed, not through whoever opened the vendor email on a Friday. In that sense WordPress SEO is restraint.
Core Web Vitals on WordPress, measured where the enquiry starts
A homepage laboratory score is a vanity metric when the enquiry starts on a service page opened from a phone. Measure that template. Where Search Console has enough field data, use the URL groups, not one Lighthouse run of the front page. If the site is too small for field data, test the real template on a mid-range phone profile, cold cache and warm, and say so. Do not pretend a lab figure is field data.
Read Largest Contentful Paint, Interaction to Next Paint and Cumulative Layout Shift on the template that sells. Ignore a perfect thank-you page. Ignore a failing campaign URL that should have been redirected already. A technical SEO audit should already have named the canonical URLs. Speeding up a duplicate flatters the chart and does nothing for the page that ought to rank.
Page speed plugins are not the first move. Caching, honest image sizes and compression are hygiene. A plugin that defers every script often breaks the form, the menu, or the feature you deferred to turn a gauge green. Remove the script before you guess which delay attribute will survive the next update. If a host cache and a plain theme already meet the budget, do not add a plugin to celebrate it.
LCP is the hero, the font, or a builder habit
On a WordPress marketing site, Largest Contentful Paint is usually the first large image, implemented in the most expensive way on offer. A CSS background will not take priority the way a real image element can. A slider waits for its script before it paints. A builder wraps the hero in nested rows. An upload several thousand pixels wide is sent to a phone because nobody generated the size the layout uses.
Find the LCP element in the field or in a trace. Serve it as an image, compressed, at the size displayed, with width and height set. Preload it only when you are sure it is the LCP, and only on that template. Two weights of one family are enough. Do not lazy-load that first image. Later images can wait. The one in the first viewport cannot.
Hosting matters once time to first byte is the bottleneck on a page that is otherwise lean: no object cache, or a PHP version that should have been retired. Move host for that reason. Do not move host instead of deleting the slider. A quicker box still has to parse the same builder script on the handset.
INP is the tax on every script you left in the theme
Interaction to Next Paint is where a site that already "passed" LCP still feels broken. The main thread is busy when someone opens the menu, focuses the form or taps a control. The sources are familiar: the builder's front-end script, a chat widget, an A/B tool on every URL, a slider, a lightbox, and a tag manager firing tags nobody can name.
Trace the service template. Do not audit speed from the plugin screen. Plugins that run only in wp-admin are not the same thing as plugins that print on the public site. A form used on one URL should not enqueue everywhere. Shop styles should not load on a brochure page that sells nothing from a catalogue. Load conditionally. A global switch that delays all JavaScript is what teams try first and remove after the form stops submitting.
Banners and review carousels also shift the layout when they arrive late. Reserve the space, or keep them out of the first screen. A banner over the button people came to press is a conversion fault even if the chart looks calmer. INP and layout shift often share a vendor, and removing that vendor repairs both.
What the monthly speed note should contain
The note names the template, the field metric that is poor, the element or script responsible, the change, and the date you will read the field data again. Field data lags. A green lab run the same afternoon is not a result.
Do not lead with a single page-speed score. A blend hides a failing service template behind a homepage that passes. A gauge screenshot is not a finding. The finding is a sentence: the hero is a CSS background inside a slider, and it is being replaced with a sized image while that script comes off the template.
Copy is not a vitals task, but a fast page that hides the offer will not hold the query. Service page structure is the companion. A content programme writes the page. Speed work stops the template punishing it. Where the same URLs are bought in paid search, the landing template owes the same budget. PPC and SEO on one map is how the ad is stopped from pointing at a heavier variant than the organic URL.
What we will not install, and what we leave
We will not add a second optimisation plugin on top of one that already concatenates, defers and breaks the editor. We will not lazy-load the LCP image. We will not turn the site into critical CSS that nobody can rebuild after the next theme update. We will not chase a perfect laboratory score while the contact form depends on a script the delay feature removed.
We leave a small shift from a footer image once the enquiry template is stable. We leave a blog template that is already lean. We leave wp-admin slow when the public site is not. Editor lag is a care issue, not a ranking one, unless people have started pasting from a builder every week and shipping a new layout each time.
SEO Jackers Global will name the templates and do the work on a staging link you can open. It sits with WordPress design and care and, when the pages themselves are the gap, an SEO retainer that does not pretend a plugin is the strategy. Bring the theme, the plugin list and Search Console.