What changed since 2024
Three things matter in 2026:
- INP (Interaction to Next Paint) fully replaced FID in March 2024 and is now a confirmed ranking signal. The "good" threshold is under 200ms at the 75th percentile.
- LCP (Largest Contentful Paint) still targets under 2.5s, but Google's field data now weighs mobile much more heavily than desktop.
- CLS (Cumulative Layout Shift) target is unchanged at under 0.1, but new ad and consent-banner patterns are quietly breaking it on thousands of small business sites.
Why most small business sites fail INP
INP measures the worst interaction on a page — usually a tap, click, or keypress. The killers we see most often:
- Heavy third-party tags firing on click (chat widgets, marketing pixels, A/B test scripts).
- Hydration spikes from React/Next/Astro sites that ship too much JavaScript to mobile.
- Long click handlers doing analytics work synchronously instead of via `requestIdleCallback`.
The 2026 fix list (in priority order)
1. Audit third-party scripts. Remove anything not actively used. Defer chat widgets until after first user interaction. 2. Use `fetchpriority="high"` on the LCP image. This single attribute often shaves 400–800ms off LCP on mobile. 3. Preconnect to your CDN and font host. Two `<link rel="preconnect">` tags in the head. 4. Self-host fonts with `font-display: swap` and a `size-adjust` fallback to avoid CLS. 5. Move analytics to Partytown or a server-side tag so they don't block the main thread. 6. Reserve space for ads, embeds, and cookie banners with explicit `min-height` to kill CLS.
How to actually measure (for free)
- PageSpeed Insights for one-off checks — uses real Chrome user data (CrUX).
- Google Search Console → Core Web Vitals report for the full site, grouped by issue.
- DebugBear or Treo ($30–$80/mo) if you want trend graphs without building your own.
Don't rely on Lighthouse alone — it's lab data and routinely disagrees with what Google actually uses to rank you.
What this is worth
In our client data, moving a service business site from "Needs Improvement" to "Good" on all three metrics correlates with a 12–28% lift in non-brand organic traffic within 90 days, especially on competitive local queries. It's not magic — it's just that you stop being penalized for being slow.