The performance bar for marketing sites has moved, and most of the sites we audit have not noticed. INP replaced FID two years ago, the LCP threshold has tightened in real-world percentiles, and the SEO weight on Core Web Vitals continues to compound through the page-experience signals. The result is the most consequential, most ignored performance shift of the decade. Sites that cleared the 2023 bar are quietly underperforming the 2026 one, and the traffic loss is attributing to other things on the dashboard.
INP Is The Metric Most Teams Are Underestimating
Interaction to Next Paint replaced First Input Delay in March 2024, and it is a meaningfully harder metric to clear. FID measured only the first interaction; INP measures the worst interaction across the entire page lifetime. Google's INP documentation is the canonical reference. The 200ms "good" threshold, taken at the 75th percentile of real-user interactions, exposes long tasks that the FID-era instrumentation never surfaced. Most sites we audit pass FID and fail INP, and the failure is concentrated in third-party tags, hydration costs on JavaScript-heavy frameworks, and long event handlers on navigation.
LCP: The Threshold Did Not Move, But The Practical Bar Did
The LCP threshold of 2.5 seconds at the 75th percentile has not changed, but the practical bar has shifted upward as competitors ship faster sites. The Chrome User Experience Report's aggregate field data shows the median marketing site at the 75th-percentile LCP now sits around 2.0–2.2 seconds for sites in the top quartile of their category, with the gap widening. Clearing the threshold is not enough; the SEO advantage compounds for sites that beat it comfortably.
CLS Is The Quiet One
Cumulative Layout Shift is the metric most teams pass once and forget about. The failure mode is consent banners, late-injected ads, and font-loading layout pops that accumulate over a session. Web.dev's CLS guide walks through the canonical fixes: reserve space for media, preload critical fonts, and constrain consent and third-party injections to fixed regions. The work is not technically hard. The discipline of catching CLS regressions in CI, before they reach the field, is what most teams do not have.
The SEO Weight Is Real And Compounding
Google's page experience documentation is unambiguous: Core Web Vitals are part of the ranking system, alongside HTTPS, mobile-friendliness, and the absence of intrusive interstitials. The weight is modest in isolation and substantial in aggregate, particularly for competitive head terms where multiple results are otherwise comparable. The sites that have prioritized vitals consistently are the sites that hold rankings against newer, better-funded competition.
Where Marketing Sites Actually Lose Performance
The performance regressions on marketing sites cluster in a small number of places. Tag-management bloat, where the cumulative cost of marketing analytics has grown unaccounted-for. Hero images that ship at 2x of the container size on mobile. Web fonts that block rendering on critical paths. Hydration overhead on frameworks chosen for developer ergonomics. And the long tail of third-party embeds that nobody owns. Each is fixable. None gets fixed without a performance-budget discipline.
The Operating Pattern That Works
Sites that clear the 2026 bar share an operating pattern. Performance budgets wired into CI, with budgets that fail the build rather than warn. RUM monitoring of Core Web Vitals at the 75th percentile, segmented by template and device class. A quarterly performance review with explicit owners for tag bloat, image weight, and third-party regressions. And a treatment of vitals as part of the SEO program, not as a separate engineering concern. The sites that do all of this are the sites that gain ranking. The sites that do none of it are the sites that quietly lose traffic that they attribute to algorithm shifts.
A One-Day Audit
A pragmatic self-audit, runnable in a day. Pull the CrUX dashboard for the property and segment by device. Run PageSpeed Insights on the top ten templates and capture INP, LCP, and CLS. Run a Lighthouse audit on the top three landing pages and read the diagnostics. Pull a tag inventory from the tag manager and identify tags that haven't fired in 90 days. Identify the biggest images on the home page and check their served dimensions vs. container dimensions. The output should be a punch list with a clear severity ranking. Most sites have ten to fifteen items on it.
Key Takeaways
- INP replaced FID and is meaningfully harder to clear — most sites pass FID and fail INP
- The LCP threshold did not move, but the practical bar has shifted upward as competitors ship faster
- CLS is the quietly-broken metric — consent banners, late tags, and font pops accumulate over a session
- Core Web Vitals are part of Google's ranking system; the SEO weight compounds on competitive terms
- Performance regressions cluster in tag bloat, oversized images, web fonts, and hydration overhead
- Sites that clear the 2026 bar wire performance budgets into CI and treat RUM monitoring as load-bearing