Case Study 2 — The Informational Publisher and the Vanishing Click: A Contested Decline, Honestly Told
A complementary angle. Case Study 1 looked at the answer machine's accuracy failure — a real, public episode. This one looks at the harder, contested question the chapter refuses to dodge: what happens to a good site whose traffic is the most exposed kind, and can we even measure the AI answer's effect?
This is a labeled composite, built from a well-documented public pattern of 2024–2025: independent informational publishers reporting meaningful declines in Google organic traffic, attributing much of it to zero-click behavior and AI Overviews — a claim Google has partly disputed, arguing its methodologies are flawed and that AI features still send clicks. The pattern (informational publishers under click pressure; a genuine measurement dispute) is real and public. The specific site, its numbers, and its timeline are constructed and illustrative — the mechanism and the uncertainty are what is real. No figure here is presented as a measured fact about a real company.
Background: a good site, built on the most exposed traffic
Call it a mid-size independent guide site — a genuinely good one. Two knowledgeable people spent a decade writing careful, well-researched informational content: "how does X work," "what is the difference between Y and Z," "how long does W last." No spam, no scraping — real, useful explainers, many of which earned #1 rankings and, for years, a healthy stream of organic traffic that paid the bills through advertising and affiliate links.
Read that business model against Chapter 36 and the vulnerability is immediate. Its entire traffic base was simple and mid-complexity informational search — the single most AI-exposed intent category (§36.3). Every strength that made it successful in the ten-blue-links era (comprehensive, clear, quotable answers to common questions) made it exactly the kind of content an AI Overview can summarize most completely. The site had done nothing wrong. It had simply built its house on the part of the beach the tide was now coming for.
FIGURE C36.2 — "Same #1 rankings, falling clicks" [labeled composite — illustrative]
YEAR 0 Dozens of #1 informational rankings; steady organic traffic; ad + affiliate revenue healthy.
YEAR 1 AI Overviews expand across the site's query set. Rankings HOLD. Impressions HOLD. Clicks soften.
YEAR 2 Clicks down materially year over year while rankings barely move. Revenue follows the clicks down.
(All figures illustrative. The SHAPE — stable rankings, stable impressions, falling clicks on informational
queries — is the reported pattern; the exact magnitude is contested and unmeasurable from outside.)
The issue: ranking and traffic came apart — and nobody can fully measure it
This is the anchor from Chapter 2 and §36.2, lived from the publisher's seat: the page ranks #1 and loses clicks anyway. The decoupling of ranking from traffic is the defining commercial problem of AI search, and this site is where it hurts most, because it has no local pack to fall back on, no transaction to complete, no product to sell — only the informational click that the AI answer is best at absorbing.
But now the chapter's hardest honesty, the reason this is a limits case: the publisher cannot prove how much of the decline the AI Overview caused. Search traffic moves for many reasons at once — core updates (Chapter 6), shifting search behavior, seasonality, competition, changes in how Search Console attributes data. Disentangling "AI Overview took the click" from every other force is genuinely hard from the outside, and this is exactly where the field's confident statistics come from and exactly why Chapter 36 refuses to quote them.
⚖️ Evidence Check Claim: "AI Overviews caused this publisher's traffic to fall by [X]%." — What is defensible: the direction and mechanism — a stable-ranking, falling-click pattern concentrated on informational queries is consistent with zero-click behavior, and multiple independent publishers have reported it. Something real is happening to informational click-through. — What is not defensible: any precise attribution figure. The publisher sees clicks fall but cannot isolate the AI Overview from core updates, behavior shifts, and measurement changes; third-party studies estimating the effect use limited samples and clashing methods; and Google has publicly disputed the gloomier analyses, saying AI features still drive clicks and that some studies are flawed. The truth is a contested range, not a number. — The honest one-liner: we can say informational click-through is under real pressure and this site is squarely in its path; we cannot honestly hand the owner a precise percentage, and anyone who does is guessing.
That uncertainty is not an academic footnote. It is the publisher's actual predicament: they are quite sure they are being squeezed, they cannot prove by how much, and they have to make survival decisions anyway — which is the real texture of operating in the AI transition, and why §36.6's "reason under uncertainty" is a practitioner skill, not a philosophy seminar.
What it shows — and the limit it teaches
The clean lesson and the hard limit sit together.
The clean lesson: informational traffic is the exposed traffic, quality does not exempt you from the tide, and a business whose only channel is Google informational clicks has a single point of failure it did not choose. Everything §36.3 and §36.5 argue is visible here.
The limit it teaches: being good is necessary but not sufficient, and the effect is not cleanly measurable. This is the AI-era echo of the hard truth from Chapter 13's Case Study 2 — some genuinely good sites get hurt by forces they did not deserve and cannot fully quantify. It does not mean quality was pointless (the site's authority is exactly what keeps it in the AI's retrieval set and what makes its brand worth building). It means quality was never a guarantee of traffic on a platform you do not own — which is the whole argument for diversification, arriving now not as prudent advice but as the difference between surviving the squeeze and not.
🔗 Connection The counter-intuitive discipline this site needs — stop depending on the exposed channel; build what you own — is the traffic diversification imperative of §36.5, first raised for algorithm resilience in Chapter 6 (Google Updates) and deepened for publishers in Chapter 34 (Publisher and Media SEO). The measurement humility is the evidence-tier habit from Chapter 2 (The Ranking Algorithm).
Outcome
The composite site's realistic path is neither collapse nor easy rescue — it is the slow, unglamorous work Chapter 36 actually prescribes. It leaned into the parts of its informational content an AI answer cannot fully replace: original testing, first-hand assessment, "here's how to tell on your specific situation" — the information gain (Chapter 13) that both keeps it in the AI's citations (§36.4) and gives readers a reason to click past the summary. More importantly, it moved its center of gravity onto owned channels: it finally built and nurtured an email list (the readers who now return regardless of the SERP), invested in becoming a named destination people seek directly rather than an anonymous answer they stumble onto, and diversified its revenue away from pure search-dependent ad clicks. The click pressure did not vanish. The business's dependence on that click did.
Some publishers in this real pattern have navigated the transition this way and stabilized. Others have not, or have not yet. That range — not a tidy triumph, not a tidy tragedy — is the honest outcome, and pretending otherwise would betray the chapter's whole posture.
The lesson
A #1 ranking is no longer a promise of traffic, the AI answer's exact bite cannot be measured from outside, and the only durable response to both is to stop depending on the channel you don't own. This case teaches the chapter's limit as sharply as Case Study 1 taught its mechanism: quality keeps you in the game (in the retrieval set, in the citations, in the brand) but does not guarantee the click, and the effect resists clean measurement — so you plan around the direction, diversify against the uncertainty, and build the owned audience that makes you robust to a future no one can forecast. The tool is new. The verdict is the one this book has delivered since Chapter 1, now with the highest stakes yet: be genuinely worth finding, and never bet the whole business on a single algorithm's willingness to send you the click.
Discussion questions
- This site did "everything right" and still lost clicks. Explain, using intent (§36.3), why its specific traffic base made it uniquely exposed — and why a local service business with the same rankings would not face the same squeeze.
- The publisher cannot prove how much of its decline the AI Overview caused. Why is that measurement problem real rather than an excuse, and how should an honest strategist talk to the owner about a decline they can't cleanly attribute?
- Google disputes the gloomier third-party studies of AI-Overview click impact. Using Chapter 2's evidence tiers, explain how you would weigh the publisher's lived experience, the third-party studies, and Google's rebuttal without landing on a fabricated number.
- Design the diversification plan for this site: what do you build off-platform, what do you change about the informational content itself, and how does each move map to a section of Chapter 36?
- Compare this site's plight to Rivertown's (the Strategy File). Both face AI Overviews — why is the local service business so much better positioned, and what, if anything, does the publisher have that Rivertown should envy?