And yet, over the same 17-month period, click-through rate collapsed 79%. Monthly leads dropped by an estimated 1,200. Organic sessions fell 37% at the exact moment the content was performing its best.

This is a data story I watched unfold at UAGC — and the pattern it reveals is one of the most important things happening in search right now.

The Dataset​

17 months of Google Search Console and GA4 data, October 2024 through February 2026. University blog content targeting primarily informational, educational queries.

The complete interactive report is published here: The Search Ecosystem Evolved — UAGC Zero-Click Search Impact Report

What follows is the strategic analysis.

What the Data Shows​

Rankings and traffic are now decoupled​

At the start of the tracking period, average position was 17.7 — buried on page two. Monthly impressions: 3.5M. Clicks: 28,700. CTR: 1.07%.

Seventeen months of sustained SEO work delivered a 51% position improvement — average position reached 8.7, comfortably on page one. Impressions surged to 6.6M/month — nearly double.

By every traditional SEO metric, this was exactly what was supposed to happen.

Except CTR dropped from 1.07% to 0.22% — a 79% decline. And monthly clicks fell from 28,700 to 14,300 — halved.

The scissors effect: rankings going up, CTR going down. The two lines that used to correlate now diverge sharply.

February 2026 was the clearest proof point: best-ever average position (8.7), worst-ever CTR (0.22%). The execution succeeded. The ecosystem changed the outcome.

The lead pipeline narrowed — even as conversion rate improved​

Fewer clicks produce fewer organic sessions (136K → 86K, down 37%), which feeds fewer leads into the enrollment funnel.

Here is the part that deserves careful attention: organic conversion rate actually improved — from 1.55% to 2.40%, a 55% gain. The visitors who still click through are more qualified. The snippet pre-filtered out low-intent traffic. What remains is higher intent.

But the math does not close. At October 2024 traffic volumes with today's conversion rate, organic would generate approximately 3,275 leads/month. At current volumes, it generates approximately 2,064. That is an estimated ~1,200 leads lost monthly from zero-click search alone.

The ecosystem is doing better qualification work than we were — but it is also keeping more of those users for itself.

Why This Happened​

Three structural shifts converged in this dataset:

AI Overviews now synthesize answers for approximately 30% of informational queries directly on the SERP. When someone searches "what is a capstone project," Google answers it with content from your page. The user's need is met. The click never happens.

AI answer engines — ChatGPT, Gemini, Perplexity — satisfy queries entirely outside the search index. These sessions never appear in GSC at all. They are invisible to the dataset.

User behavior shifted in response to both. Searchers now expect answers without visiting a page. The act of clicking has become optional for a growing share of informational queries.

The result: 27.2% of U.S. searches now end without a click, up from 24.4%. For educational content — "what does PhD stand for," "what is a business plan" — the rate is substantially higher.

What Content Is Resistant vs. Vulnerable​

The pattern that emerges from segmenting by content type is the most actionable finding in the dataset.

The highest-volume query in the dataset — "business plan" — generates 7.7M impressions and 0.02% CTR. The AI answers it completely. Meanwhile, "strategic planning healthcare" achieves 6.04% CTR — 300 times higher — because the answer requires professional context that a snippet cannot fully satisfy.

The principle: AI systems cannibalize content that resolves entirely in a sentence. They do not cannibalize content where the resolution requires the full experience of reading, comparing, deciding, or acting.

The Strategic Shift This Requires​

The content strategy that built this visibility was the right strategy for the ecosystem that existed. It's not the strategy for the ecosystem that exists now.

Stop​

Investing in generic definitional content ("what is X") where the query resolves fully in a snippet or AI Overview. These pages still earn impressions — sometimes millions — but they contribute effectively zero clicks or downstream value.

Shift​

Reframe existing high-impression definitional pages with decision hooks, comparison angles, and calls-to-action that a featured snippet cannot replicate. "What is a capstone project" → "Capstone vs thesis: which satisfies your program's research requirement?" The impression count stays. The CTR recovers.

Build​

Interactive tools, program-specific decision guides, comparison frameworks, and brand-owned content where the full experience matters. Treat every impression from a cannibalized query as a brand touchpoint — the goal is recognition and trust, not the immediate click.

What This Means for SEO Strategy in 2026​

This dataset is a proof point for something I have been arguing in client work for the past 18 months: the KPIs for organic search need to update alongside the ecosystem.

Average position is a valid proxy metric when it correlates with traffic. When a site can achieve its best-ever position and its worst-ever CTR simultaneously, position alone is not an outcome metric — it is a distribution signal. You are in the auction, but the auction's prize has changed.

The metrics that matter now:

• Impression-to-click ratio by content type — distinguishes healthy traffic from cannibalized queries
• Brand vs non-brand split — brand queries are structurally resistant to zero-click erosion
• Organic conversion rate — the quality signal in a volume-compressed environment
• AI citation presence — whether your content appears in AI Overviews and answer engines even when clicks do not follow

The ~1,200 leads this site is losing monthly will not come back through better rankings alone. They will come back through content the ecosystem cannot consume on behalf of the user — content that demands the full experience.

Read the Full Report​

The complete interactive report includes month-by-month data visualizations, the full CTR segmentation by content type, and the specific query-level analysis.

The Search Ecosystem Evolved — Full Case Study →

Working through the same pattern in your organic data? Let's look at it together.

The architecture is correct. The execution misses why it works.

What the Pillar-Cluster Model Is Actually Doing​

The conceptual model is well-known: a broad "pillar" page covers a topic at high level, with "cluster" pages covering sub-topics in depth and internally linking back to the pillar.

The original logic was mechanical — internal link equity flows through the cluster to the pillar, concentrating authority on the most important page. This is partially true, but the more important effect is topical coverage signaling.

When Google's systems evaluate whether a site is authoritative on a topic, one of the signals is comprehensiveness. A site that has 15 well-developed pages covering different aspects of a topic signals deeper expertise than a site with a single comprehensive page — even if that single page is excellent.

The pillar-cluster model works not primarily because of internal linking (though that helps) but because it creates the topical surface area that demonstrates genuine depth.

Where Most Implementations Go Wrong​

Wrong: Building structure before building substance​

The most common mistake is treating the cluster as an architecture exercise. Teams map out a pillar, identify 8–12 cluster topics, assign writers, publish the content, and call the cluster "complete."

The problem: if the cluster pages are thin, generic, or nearly identical in content depth to what competitors already have, the cluster signals average coverage at best. You are not demonstrating expertise — you are matching it.

A topical cluster that demonstrates authority needs each cluster page to be the authoritative resource on its sub-topic. Not comprehensive. Not thorough. The resource a practitioner would bookmark and return to.

Wrong: Choosing cluster topics based on keyword volume alone​

Keyword volume is a useful input but a poor primary decision driver for cluster architecture. High-volume sub-topics are where competition is highest and where it is hardest to differentiate.

The better approach: map sub-topics to the actual questions and decisions your target audience faces, whether or not those questions have large standalone search volumes. A cluster page that answers a real question no one else has answered well is more valuable than a page targeting a high-volume keyword already covered by 30 competing resources.

Wrong: Internal linking as an afterthought​

Internal links in a content cluster are not decoration. They are navigational signals to both users and crawlers. A cluster where the pillar links to cluster pages but cluster pages do not link to each other or back to the pillar is not a cluster — it is a set of independent pages with a shared topic.

Every cluster page should:

• Link back to the pillar page (with relevant anchor text)
• Link to 2–4 other cluster pages where the content connects
• Receive links from the pillar page and from related cluster pages

The link graph should reflect the conceptual relationships between sub-topics, not just the hierarchy.

What Actually Builds Topical Authority in 2026​

Depth that is demonstrably better, not just longer​

The era of "10x content" (making content 10x longer than competitors) is largely over. AI can generate long content. Length is not differentiation.

Depth that builds authority is:

• First-hand experience and data that competitors cannot replicate
• Coverage of failure modes, edge cases, and nuance that generic guides skip
• Updated information that reflects the current state of the topic, not its state two years ago
• Specific enough to be useful to a practitioner, not vague enough to apply to everyone

Content that earns external references​

A content cluster that exists only on your site is not building authority — it is organizing information. Topical authority requires external validation: other sites linking to your cluster pages because they are the best resource on that sub-topic.

This means your cluster strategy needs to be paired with a link acquisition strategy. Identify which cluster pages are most likely to attract natural links (original data, unique perspectives, comprehensive guides on under-covered sub-topics) and invest disproportionately in those.

Consistent coverage updates​

Topical authority is not a one-time build. Search landscape, industry practices, and best-in-class examples change. A content cluster that was authoritative in 2023 and has not been updated since is gradually becoming less so as competitors publish fresher content.

Build a review cadence into your cluster strategy: audit the highest-traffic cluster pages annually, update statistics and examples, and add coverage for sub-topics that have emerged since the initial build.

Answer real questions, not just search queries​

The best cluster content is built around what practitioners and buyers actually need to know, not around what they type into a search box. These are related but not identical.

Talk to your sales team about what questions prospects ask. Review customer support tickets. Read forum threads and community discussions. The questions people ask in private (to their colleagues, in Slack communities, in support tickets) are often higher-value cluster topics than the questions people type into Google — and they are usually underserved.

A Practical Cluster Audit​

If you have an existing content cluster and want to evaluate its authority:

1. Map the cluster — list every page that is meant to be part of the cluster
2. Audit internal links — verify bidirectional linking between cluster pages and pillar
3. Assess depth — for each cluster page, ask: is this the resource a practitioner would bookmark, or is it comparable to every other resource on this sub-topic?
4. Check external links — which cluster pages have meaningful inbound links? Which have none?
5. Identify coverage gaps — what sub-topics does your target audience care about that your cluster does not cover?
6. Check freshness — which pages have outdated statistics, examples, or recommendations?

Most clusters I audit fail at steps 3, 4, and 5 — the pages are structurally organized but not authoritative, they have not attracted external links, and there are meaningful gaps.

The Bottom Line​

The pillar-cluster model is correct. The execution that makes it work is harder and slower than most teams anticipate. It requires:

• Sub-topic pages that are genuinely the best resource on each sub-topic
• External link acquisition paired with the content build
• Regular updates to maintain freshness and relevance
• Topic selection driven by real audience questions, not just keyword volume

Done correctly, topical clusters are the most durable organic growth strategy available. Done superficially, they are a lot of content that does not rank.

Want to audit your existing cluster structure or plan a new one? Let's work on it together.

This post covers what INP actually measures, why it is harder to fix than FID, and what a realistic optimization process looks like.

What INP Measures (and Why It Is Harder Than FID)​

FID measured the delay before the browser could begin processing a user's first interaction. INP measures the full responsiveness of any interaction throughout the page's lifecycle — click, tap, keyboard input — and reports the worst one (or close to it).

The difference matters enormously in practice:

• FID was measured at first interaction, often a click before the page was fully loaded. Optimizing FID mostly meant reducing main-thread blocking during initial load.
• INP is measured across the entire session. A user who clicks a filter on your product page, expands an accordion, or submits a form — any of those interactions can tank your INP score if they cause long tasks on the main thread.

A page can have excellent initial load performance and terrible INP. I see this regularly on sites that have invested heavily in LCP and CLS while leaving JavaScript interaction handlers unoptimized.

The INP Thresholds​

The 200ms threshold is achievable but not trivial. The common mistake is assuming that a 400ms INP is "close enough." It is not — Google's field data from Chrome UX Report (CrUX) confirms that pages in the "needs improvement" range see measurable ranking disadvantage compared to "good" pages in competitive queries.

Where INP Problems Come From​

After auditing INP across dozens of client sites, the culprits break into three categories:

1. Long JavaScript tasks blocking the main thread​

The most common cause. When a user clicks a button and your site executes a 600ms JavaScript task before rendering the response, INP is 600ms.

Tools to find these:

• Chrome DevTools Performance tab → look for long tasks (red bars in the main thread timeline)
• PerformanceObserver with longtask type in production monitoring
• Google's Lighthouse in CI

Common sources: analytics batching, A/B test framework evaluation, poorly structured React re-renders.

2. Input delay from timer-heavy event handlers​

If your click handlers do expensive synchronous work (DOM queries across large trees, unoptimized loops, synchronous storage reads), the delay accumulates.

The fix is usually breaking work into smaller chunks: setTimeout(() => {}, 0) to yield to the browser, requestIdleCallback for non-urgent work, or proper async boundaries.

3. Rendering delay after the processing phase​

Even if the event handler is fast, the browser still needs to repaint. Complex CSS animations, large DOM trees, and forced layout recalculations all increase rendering time after a user interaction.

A Realistic Optimization Process​

The right order of operations:

1. Establish field data — install CrUX monitoring or use a RUM tool (Web Vitals JS library, Sentry Performance, or Datadog RUM). Lab data from Lighthouse does not always capture INP accurately.

2. Identify the interactions — the CrUX data will tell you which interaction types are causing poor INP. It is often a small number of interactions responsible for most of the problem.

3. Profile in DevTools — reproduce the worst interactions in Chrome DevTools with the Performance recorder. Identify the specific tasks causing the delay.

4. Chunk and yield — restructure long synchronous tasks to yield to the browser between chunks. React 18's concurrent features (useTransition, useDeferredValue) are useful here if you are on a React stack.

5. Validate in field data — after deploying fixes, wait 28 days for field data to reflect changes in CrUX (it uses a rolling 28-day window).

What to Prioritize First​

If you are starting from scratch on INP optimization:

• Check CrUX via PageSpeed Insights for your most important landing pages
• Focus on interactions that happen on high-traffic pages (product pages, checkout, search results)
• Look for third-party scripts that fire on interaction — these are a common and often easy win (lazy-load or defer analytics and chat widgets)
• Check if your JavaScript bundle is large and synchronous — code-splitting reduces the probability of long tasks during critical interactions

The Underrated Factor: Device Distribution​

INP is measured on real user devices in the field. If a significant portion of your users are on mid-range Android devices with 2–4 GB RAM, your INP in the field will be worse than what you measure on a developer MacBook.

Always check your CrUX data segmented by device type. Mobile INP is typically 2–4x worse than desktop INP. If you are optimizing for desktop only, you are missing the audience that needs it most.

INP is harder to fix than any previous Core Web Vitals metric because it requires understanding your full JavaScript execution model, not just your initial load path. But the sites that have done the work are seeing both ranking benefits and measurable conversion rate improvements — faster interactions increase task completion.

If you want to walk through your site's INP profile, let's talk.

After a decade in SEO and the last two years integrating AI tools into my workflow, here's what I've learned about optimizing for this new landscape.

The Shift: From Keywords to Context​

Traditional SEO focused on matching keywords. AI-driven search focuses on understanding intent and context. Large language models don't just match strings -- they comprehend meaning, evaluate authority, and synthesize answers from multiple sources.

This means your content strategy needs to evolve:

• Answer questions directly. AI models pull from content that provides clear, authoritative answers.
• Structure for extraction. Use headers, lists, and tables that AI can parse and cite.
• Build topical authority. AI models favor sources that demonstrate comprehensive expertise on a subject.

What's Actually Working​

Based on real client engagements, these are the highest-impact changes:

1. Structured Data Is More Important Than Ever​

Schema markup helps AI systems understand your content's context. FAQPage, HowTo, Article, and Organization schemas are now table stakes, not nice-to-haves.

2. Content Depth Over Content Volume​

AI models evaluate the quality and comprehensiveness of individual pages. One thorough guide outperforms ten thin posts targeting long-tail variations.

3. E-E-A-T Signals Matter for AI Citations​

When ChatGPT or Perplexity cites sources, they lean toward content with clear authorship, demonstrated expertise, and transparent methodology. Author pages, case studies, and first-person expertise signals all contribute.

4. Technical Foundations Still Win​

Fast load times, clean crawlability, mobile optimization, and proper canonicalization remain foundational. AI systems still rely on search engine indexes, and those indexes still prioritize technically sound sites.

The Practical Takeaway​

You don't need to rebuild your SEO strategy from scratch. The fundamentals still apply. But you do need to layer in AI-readiness:

1. Audit your structured data coverage
2. Review content for direct answer formats
3. Strengthen author and expertise signals
4. Monitor AI search platforms for your brand mentions

The organizations that adapt now will have a significant advantage as AI-driven search continues to grow.

Need help assessing your AI search readiness? Get in touch for a strategy consultation.