What Core Web Vitals are, briefly
Three metrics Google publishes as ranking signals: Largest Contentful Paint (LCP), Interaction to Next Paint (INP) and Cumulative Layout Shift (CLS). All three are measured on real users, not in a lab, through Chrome's User Experience Report.
Are they really ranking factors?
Yes, but the relationship isn't linear. Crossing the "Good" threshold matters far more than incremental improvements within "Good". If your LCP is at 4.5 seconds, getting it to 2.4 will move rankings. Getting it from 1.8 to 1.4 mostly won't.
LCP, the one to fix first
Largest Contentful Paint is the time it takes for the largest visible element to render. In practice, this is almost always your hero image or hero block of text. The fixes are usually infrastructure-level:
- Use a real CDN. Not "Cloudflare on free tier proxying everything through a single origin in Toronto", an actual edge CDN with proper cache rules.
- Serve images in modern formats. AVIF where supported, WebP as fallback, JPEG as last resort.
- Preload your hero image and your largest font file.
- Server response time matters. If your TTFB is over 600ms, fix that first, no amount of CDN tuning recovers a slow origin.
INP, the one most sites get wrong
Interaction to Next Paint replaced First Input Delay in 2024. It measures how responsive the page feels when a user actually interacts with it. The bad news: most sites fail INP because of third-party scripts they forgot they installed.
The audit we run on every engagement:
- Open the page in Chrome DevTools, Performance tab, throttled to 4× CPU.
- Record an interaction (click a button, open a menu).
- Look at the "Long Tasks" list. Anything over 50ms blocks interaction.
- The worst offenders are almost always: tag managers loading too many things, chat widgets, video embeds, A/B testing tools and analytics that should have been a Pixel.
CLS, mostly hygiene
Cumulative Layout Shift is the metric most teams over-index on because PageSpeed Insights makes it look loud. In ranking terms, it is the least impactful of the three. Fix it because it improves UX, not because it will transform your traffic.
The fixes are mechanical:
- Set explicit
widthandheighton every image and iframe. - Reserve space for ads, embeds and consent banners before they load.
- Never inject content above existing content after first paint.
Where to spend your performance budget
For a typical Canadian SMB website, ordered by ROI:
- Get LCP under 2.5s. This is usually one to two days of work with hosting and image-pipeline changes.
- Audit and remove third-party scripts until INP is under 200ms. This is usually one day of work and an argument with your marketing team.
- Fix the worst CLS offenders. This is usually a few hours of CSS work.
Tools we actually use
- PageSpeed Insights for headline scores and field data.
- WebPageTest.org for filmstrip and waterfall analysis.
- Chrome DevTools for INP tracing.
- Cloudflare Real User Insights or Vercel Analytics for ongoing monitoring.
Where this fits with your SEO work
Performance is a baseline, not a strategy. Once you are in the green on all three, the next big rankings work is structural, site architecture, internal linking, and content depth. Our SEO service covers both layers; our hosting service takes the performance baseline off your plate entirely.