Core Web Vitals shifted again in 2026. INP fully replaced FID as the interactivity metric, mobile LCP thresholds tightened by 200 milliseconds, and Google began surfacing field data more aggressively in Search Console. For service businesses with WordPress sites and a few plugins doing too much, that means the rankings dropped without any content changing.
Here's what actually matters and what to do about it.
The three metrics, current thresholds
| Metric | Good | Needs work | Poor |
|---|---|---|---|
| LCP (Largest Contentful Paint) | ≤2.3s | ≤4.0s | >4.0s |
| INP (Interaction to Next Paint) | ≤200ms | ≤500ms | >500ms |
| CLS (Cumulative Layout Shift) | ≤0.1 | ≤0.25 | >0.25 |
You need all three in the green for the "good" CWV pass — one yellow drops you out.
Field data beats lab data, always
Lab data — what PageSpeed Insights gives you when you run a one-off test — measures performance under idealized conditions. Field data — what Search Console surfaces from real Chrome users — measures what actually happens to your visitors.
Google ranks on field data. Always start there.
If your field data shows poor INP but lab data is green, your real users are hitting a JavaScript path that your test runs don't trigger. That's the bug to fix.
The three highest-leverage fixes
In our experience optimizing over 100 service business sites, three fixes move the needle more than anything else.
1. Image strategy
- Serve images in modern formats (AVIF first, WebP fallback)
- Define explicit
widthandheightto eliminate CLS - Lazy-load everything below the fold; eager-load the hero image
- Self-host or use an edge-cached CDN — never hotlink
2. JavaScript discipline
- Audit your script tags. Most service business sites carry 6–10 third-party scripts they don't need
- Defer or async every script that isn't render-critical
- Replace heavy analytics suites with lightweight alternatives where possible
3. Server response time
- TTFB (Time to First Byte) under 600ms is the floor
- Static site generation or full-page caching does more than any plugin
- Edge deployment cuts TTFB by 200–400ms for clients geographically distant from your origin server
What's not worth fixing
- Minor CLS on cookie banners — Google treats banner CLS leniently if it happens before user interaction; don't burn a sprint on this
- PageSpeed opportunities flagging Google services — Tag Manager, Maps, Fonts; defer them, can't eliminate them
- Animation jank that isn't on the critical path — looks bad in DevTools, doesn't affect ranking
The case for a rebuild
Sometimes the right answer isn't to fix the existing site. If your stack is WordPress + 30 plugins + a heavy theme + shared hosting, the fastest path to good CWV is often migrating to a static-site-generated build on edge infrastructure.
This isn't always the answer — content velocity matters, and a CMS your team can edit beats a static site nobody updates. But when CWV failures are blocking rankings and the platform isn't fixable, the rebuild math gets simple fast.
Measure, fix, monitor
CWV isn't a one-time project. Set up monthly field-data reviews in Search Console. When you ship a new feature, measure before and after. The sites that hold green status are the ones that treat performance as ongoing maintenance, not a one-quarter sprint.
Get this right and you get a quiet but durable ranking lift — across every page, on every query.