To improve Core Web Vitals, first find the one metric that fails (LCP, INP or CLS) and the part of it that’s slow, then fix that part only. On most small business sites the fix is one of three things: the main image loads late or too heavy (LCP), scripts from apps and plugins tie up the browser (INP), or something on the page has no reserved space (CLS).
Generic “11 tips” lists skip the diagnosis, so people compress every image, install a caching plugin and still see “poor” in Search Console. This guide is organized by what you see in the report, with the cause and the fix for each. The thresholds and fixes come from Google’s own guides on web.dev and Search Central, checked in November 2026.
Find the failing metric before you fix anything
- Open the Core Web Vitals report in Search Console. It’s under Experience. It uses data from real Chrome visitors and groups pages that share a template, so one slow product template shows up as many URLs. A group takes the status of its worst metric, and mobile and desktop are rated separately, according to Google’s report help page.
- Note the metric and the device. “LCP issue: longer than 2.5s (mobile)” tells you far more than “Poor URLs: 40”.
- Run one example URL from the group through PageSpeed Insights. The top section is field data from real users over the previous 28 days. The lower section is a Lighthouse lab test on one simulated phone. Use the top part to confirm the problem and the lower part to find its cause.
- Read the Insights in the lab section. Lighthouse now groups its findings into performance insights such as “LCP breakdown”, “LCP request discovery”, “Layout shift culprits”, “Render-blocking requests” and “Third parties”. These point at the actual element or script.
- On Shopify, check the web performance dashboard too. Shopify’s web performance dashboard sits on the Themes page under Online Store, rates all three metrics from the past 30 days of real visits, and marks events such as app installs and theme updates on the chart. If a metric got worse the week you added an app, you’ve found your suspect.
Step 2 matters most. Each metric has different causes, and work on the wrong one does nothing for the report. The thresholds Google uses are the same everywhere: LCP is good at 2.5 seconds or less and poor above 4 seconds, INP is good at 200 milliseconds or less and poor above 500, and CLS is good at 0.1 or less and poor above 0.25. All are measured at the 75th percentile, so the slowest quarter of your visits decides the score.
If you want the wider technical picture first (crawling, indexing, rendering), the technical SEO checklist covers speed as one stage of five.
Symptom: LCP longer than 2.5 seconds
Largest Contentful Paint is the time until the biggest image or text block in the first screen appears. On a home page that’s usually the hero photo. Google’s LCP guide splits that time into four parts and says roughly how much of the 2.5 second budget each one should take.
The “LCP breakdown” insight in PageSpeed Insights shows the same four parts for your page. Whichever one is far over its share is the one to fix.
Cause: a slow server response. If time to first byte alone eats most of the budget, image work can’t save you. Google’s guide says a high TTFB can make a 2.5 second LCP “challenging, or even impossible.” Fixes: remove redirect chains (an ad link that goes http, then https, then www costs a round trip each time), turn on full page caching, put the site behind a CDN, and avoid unique URL parameters on links you control, because they stop the CDN serving its cached copy. On cheap shared WordPress hosting, a better host is sometimes the only real fix.
Cause: the browser finds the main image late. This is resource load delay. It happens when the hero image is set as a CSS background, injected by a slider script, or marked loading="lazy". Google’s guide is blunt: “Never lazy-load your LCP image.” Put the image in the HTML as a normal <img> and add fetchpriority="high" to it, and to no more than one or two images, because the guide says a high priority on more than that stops helping.
Cause: the image is too heavy. This is load duration. Export the photo at the size it displays, serve WebP or AVIF, and give the browser a srcset so phones get a smaller file. A 4MB photo straight from a phone camera is the classic case.
Cause: CSS or scripts hold the page back. This is render delay: the image has arrived but can’t be drawn yet. Large stylesheets and synchronous scripts in the <head> are the usual reason, and Google’s guide notes it is “almost never necessary” to add synchronous scripts there. Defer non-essential scripts and remove the ones you no longer use.
On WordPress: since version 6.3, WordPress adds fetchpriority="high" to the image it estimates is the LCP element and leaves likely above-the-fold images out of lazy loading, as the WordPress core team explains. It won’t override attributes a theme or plugin already set, so a page builder or “lazy load everything” plugin can still undo it. If the LCP breakdown shows a long load delay, check the hero image’s HTML for loading="lazy" first.
On Shopify: Shopify’s theme performance guide tells theme developers to mark the LCP image with fetchpriority="high", apply lazy loading only to images outside the first screen, and use the image_tag filter, which adds a srcset and image dimensions for you. If your theme hand-builds image URLs, that’s worth raising with the theme developer.
Symptom: INP longer than 200 milliseconds
Interaction to Next Paint measures how long the page takes to show a response after a tap, click or key press. It replaced First Input Delay as a Core Web Vital on March 12, 2024, so any guide still telling you to fix FID is out of date. Google’s INP guide splits each interaction into three phases.
Cause: too much JavaScript running when people tap. This is input delay, and on small business sites it’s usually third-party code: chat widgets, review carousels, booking pop-ups, heatmaps and tracking pixels nobody reads any more. Shopify’s guide puts it plainly: JavaScript, “whether your own theme code or scripts from installed apps, runs on the main thread,” where it competes with responding to the visitor. Open the “Third parties” insight in PageSpeed Insights, list every script by who added it, and remove or defer the ones that don’t earn their place. After you uninstall an app or plugin, view the page source and search for its script name to make sure it’s really gone.
Cause: heavy work on click. This is processing time, often a menu, filter or “add to cart” button that recalculates a lot before showing anything. Google’s guide says to “do as little work as possible” in these handlers and to split the rest into smaller tasks, noting that “yielding indiscriminately is better than not yielding at all.” This is developer work. Give your developer the specific button and the INP breakdown.
Cause: a huge page. This is presentation delay. Google warns that “a large DOM can cause rendering updates to be very expensive.” Page builders that wrap every element in several layers of <div> are a common source. The “Optimize DOM size” insight flags it.
One catch: the lab test in PageSpeed Insights doesn’t tap anything, so it can’t show a real INP. It reports Total Blocking Time instead, which Google calls “a reasonable proxy metric for INP for the lab” but not a substitute. Trust the field data at the top for INP itself.
Symptom: CLS above 0.1
Cumulative Layout Shift measures how much the page jumps while people read it. Google’s CLS guide lists the usual causes, and each has a direct fix:
- Images and videos without dimensions. Always set
widthandheight, or a CSSaspect-ratio, so the browser holds the space before the file arrives. Fig 2 shows this change. - Ads, embeds and iframes. Google calls ads “one of the largest contributors to layout shifts on the web.” Reserve space with
min-heightfor the smallest likely size, and keep the slot even when nothing fills it. The same goes for embedded maps, booking widgets and review badges. - Banners added after load. Cookie bars, promo strips and “free shipping” bars that push the page down are a common culprit. Overlay them or reserve their space. Shifts within 500 milliseconds of a tap don’t count, so a “Load more” button that opens instantly is fine.
- Web fonts swapping in. A custom font that arrives late can reflow every line. Preload the main font, and use
font-display: optionalor a closely matched fallback font. - Animations that move layout. Animate with
transform, nottop,left,widthorheight.
The “Layout shift culprits” insight names the element that moved and the reason, which saves a lot of guessing.
Symptom: PageSpeed Insights says fine, Search Console says poor
This is the most common confusion, and nothing is broken. The two tools measure different things. The lab test is one simulated load on one device, while Search Console uses the 75th percentile of real visits over 28 days, split by mobile and desktop. A fast office laptop on fiber will pass a page that fails for customers on older phones and patchy cell signal.
It also lags. Field data covers the previous 28 days, so a fix shipped today takes weeks to show fully in the field numbers. Read the field section of PageSpeed Insights for the real result, and the lab section only to debug. The Search Console vs Google Analytics guide explains why tools that measure real visitors never quite agree with each other either.
Symptom: “Not enough data” or no report at all
Small sites see this a lot. Google’s report needs enough Chrome visits to a group of pages before it rates them. When a page lacks data, both Search Console and PageSpeed Insights fall back to one rating for the whole origin, and if the origin is too quiet as well, you get nothing.
There’s nothing to repair in the report itself. Run your top five pages (home, main service pages, contact) through PageSpeed Insights, fix what the lab Insights flag, and look again in a few months as traffic grows. The technical SEO issues guide covers what else to check in Search Console while you wait.
Confirm the fix with Start tracking
After you ship a fix, open the issue in the Core Web Vitals report and select Start tracking. Google watches the affected URLs for 28 days, and the issue counts as fixed only if no URL shows it during that window. Starting tracking doesn’t make Google recrawl anything, and if any URL still fails, validation fails and you can restart it after the next fix.
Be wary of shortcuts. Shopify’s guide warns against “fake performance apps” that game Lighthouse with tricks such as transparent overlays or different pages for testing tools. They can make the lab score look great while real visitors, and therefore Search Console, see no change.
FAQ
Are Core Web Vitals still a ranking factor?
Yes, as one signal among many. Google says good Core Web Vitals align with “what our core ranking systems seek to reward” and recommends them “for success with Search,” but a fast page won’t outrank a more relevant one (Search Central).
What are LCP, INP and CLS?
LCP is how fast the main content appears (good at 2.5 seconds or less), INP is how fast the page responds to a tap or click (good at 200 milliseconds or less), and CLS is how much the layout jumps (good at 0.1 or less).
How long does Search Console take to show an improvement?
Plan on about a month. Field data is a rolling 28-day window, and validation after Start tracking also runs for 28 days, so the report catches up gradually rather than the day you ship the fix.
Which metric to fix this week
Pick the metric that is red on mobile for your most visited template, find its slowest part with the matching PageSpeed Insights insight, and fix only that. For most small sites it will be the hero image or a script you forgot you installed, and one change there can move a whole group of URLs from poor to good.
