SEO Isn't Dead, It Changed: What to Do Differently

"SEO is dead" gets declared wrongly about once a year. It's still wrong now — but one part of the SEO conversation genuinely did change enough in the past two years that ignoring it is now a measurable competitive disadvantage, not just a missed nicety: technical site performance, specifically a metric called Interaction to Next Paint, has become a far harder bar to clear than most blog owners realize, and it now gates visibility in both traditional search and AI answer engines at once.

What actually changed: a much stricter responsiveness metric

In March 2024, Google formally retired First Input Delay as a Core Web Vital and replaced it with Interaction to Next Paint. The difference matters more than it sounds: FID only measured how quickly a page responded to a visitor's very first click or tap. A page could pass easily if that first interaction was fast, even if every subsequent click, form submission, or menu open lagged badly. INP measures every interaction across an entire visit and reports the worst one, scored at the 75th percentile of real visitors — a far more honest, and considerably harder, test of whether a page actually feels responsive to use (a 2026 technical breakdown of why INP replaced FID).

The current 2026 thresholds across all three Core Web Vitals: Largest Contentful Paint (how fast your main content renders) under 2.5 seconds is good, over 4 seconds is poor; Interaction to Next Paint under 200 milliseconds is good, over 500 milliseconds is poor; and Cumulative Layout Shift (how much content jumps around while loading) under 0.1 is good, over 0.25 is poor (Enterno's 2026 Core Web Vitals guide with current thresholds).

Most sites are still failing this — which is the actual opportunity

Here's the part worth paying real attention to: as of 2026, only around 42% of mobile sites pass all three Core Web Vitals thresholds simultaneously (Dotcom-Monitor's 2026 analysis of website speed and SEO). A separate, more granular benchmark of the 500 most-visited sites in the Russian-language web found a similar picture: only 38% passed all three metrics, with a median Interaction to Next Paint of 218 milliseconds — just over the 200ms "good" threshold — and results varying enormously by underlying platform, from 28% of sites on one popular CMS passing, up to 67% of sites built on modern JavaScript frameworks (Enterno's 2026 Core Web Vitals benchmark of 500 high-traffic sites). Two independent studies, in different markets, converge on roughly the same number: a clear majority of sites are still failing this bar. For a blogger willing to actually fix it, that's a real, measurable, and currently underexploited advantage over competitors who haven't caught up.

The March 2026 core update raised the stakes specifically

Google's March 2026 core update formally increased the weight Core Web Vitals carry in the ranking algorithm — sites that already passed the thresholds saw rankings climb, while sites that didn't saw drops, in some verticals described as dramatic (the same 2026 speed-and-SEO analysis). This isn't a new, isolated signal bolted onto an otherwise unchanged system — it's an existing signal that quietly became considerably more consequential in the same update cycle that reshaped scaled content enforcement, discussed elsewhere in this series.

This now gates AI citation too, not just traditional rankings

Here's the detail that connects this directly to the rest of this series' focus on AI search visibility: AI answer engines — ChatGPT, Perplexity, Google's AI Overviews, Gemini, and Copilot — now actively deprioritize slow or error-prone sources when selecting which pages to cite in a generated answer (the same source's analysis of AI search engine behavior). This series has covered extensively what makes content structurally citable — standalone paragraphs, clear headings, genuine originality. All of that becomes irrelevant if the retrieval system times out, gets an error, or deprioritizes your page before it ever evaluates the content itself. Technical performance is the floor everything else in this series sits on top of, for both search surfaces at once.

A concrete before-and-after, at real brand scale

The revenue impact of this isn't abstract. Vodafone documented improving its Largest Contentful Paint from 4 seconds down to 2 seconds and reported an 8% increase in sales alongside a 15% improvement in site-visit conversion rate as a direct result (a 2026 compilation of documented site-speed-to-revenue case studies). A large company's specific percentages won't map exactly onto a smaller blog, but the underlying mechanism — faster response translating directly into fewer abandoned visits and more completed actions — applies regardless of scale.

A practical technical audit for a typical blog

• Check your real-world field data, not just a lab test. Google's Core Web Vitals are measured from actual Chrome users via the Chrome User Experience Report (CrUX), not a simulated single-load test — check your site's real CrUX data (available free via PageSpeed Insights or Search Console) rather than relying only on a single Lighthouse score, which can look better than what real visitors actually experience.
• Compress and modernize images first. Images typically account for roughly half of a typical page's total weight — converting to modern formats and adding proper compression is usually the highest-leverage single fix for Largest Contentful Paint specifically.
• Remove render-blocking CSS and JavaScript where possible, and inline only the critical CSS needed for the visible portion of the page.
• Audit your plugins and embedded widgets — for WordPress and similar platforms specifically, an accumulation of plugins is one of the most common causes of poor Interaction to Next Paint scores, since each one can add its own JavaScript execution time to every interaction.
• Use a CDN and proper caching to reduce server response time, which affects both LCP directly and every subsequent interaction indirectly.

What hasn't changed — and why the fundamentals still matter

None of this replaces the core arguments made elsewhere in this series: relevance to real search intent, genuine backlinks, and structurally citable, original content remain the foundation of visibility in both traditional search and AI answer engines. What changed is that technical performance moved from a minor tiebreaker to an active, heavily-weighted gate sitting underneath all of it — a well-written, perfectly structured, genuinely original post that loads slowly or responds sluggishly is now at a real, measurable disadvantage before an algorithm or an AI retrieval system ever gets to evaluate the writing itself.

Frequently asked questions

Do I need to hire a developer to fix Core Web Vitals issues?
Not necessarily for a first pass — image compression, reducing unnecessary plugins, and enabling caching (often through a single plugin on WordPress-style platforms) address a large share of common issues without custom development. More stubborn Interaction to Next Paint problems, often caused by heavy JavaScript, are more likely to require developer help.

How do I check my own site's actual Core Web Vitals scores?
Google's free PageSpeed Insights tool and the Core Web Vitals report in Search Console both show real, field-collected data from actual visitors (via CrUX), rather than only a simulated lab test — check both, since they can occasionally disagree.

Is this really as important as backlinks or content quality?
It's better understood as a gate than a competing priority — Google's own March 2026 update increased its weight specifically because passing or failing now measurably moves rankings on its own, and AI answer engines apply a similar filter before even evaluating your content's citability. Strong content behind a slow site is increasingly working with one hand tied behind its back.

Why are so many sites still failing this if it's been a ranking factor since 2021?
Largely because INP, introduced in 2024, is a meaningfully harder bar than the FID metric it replaced — many sites that comfortably passed under the old metric are failing the new one simply because it measures every interaction rather than just the first, catching problems the earlier standard never surfaced at all.