WorkLabWritingAboutLet's talk →
← Back to selected work
Search Strategy·SEO + AI Visibility

When rankings improved but clicks fell

How I diagnosed a zero-click search shift, led a broader search-market audit, and began turning the findings into an operating model across owned content, external influence, and measurement.

Audit complete · Execution underway
search-performance
RANKINGS
↑ improving
CLICKS
↓ declining
Impressions
→ steady
AI visibility
emerging
Environment
In-house B2B cybersecurity marketing
My role
Search strategy, vendor/audit leadership, performance analysis, program design, measurement, cross-functional coordination
Tools
GA4, GSC, SEMrush, Looker Studio, ChatGPT, Claude, Perplexity, Google AI Mode
Focus
Organic search, zero-click behavior, AI visibility, brand presence, measurement
The Problem

The SEO dashboard was telling two different stories

The signals that usually indicate SEO progress were moving in the right direction. Average position improved and impressions held roughly steady. Clicks fell sharply.

That made the usual explanation — "we lost rankings" — hard to defend. The gap was happening somewhere between earning visibility and receiving the click.

At the same time, AI-generated answers and richer search experiences were changing what a buyer could see before ever reaching an organic result. That meant traffic alone could no longer describe the whole search experience.

Something had changed between earning visibility and receiving the click.

The rankings were improving. The search experience around them was changing.
The Shift

From measuring position to measuring actual presence

Before
Better rankings were treated as a reliable indicator of stronger search performance.
Impressions, clicks, CTR, and average position carried most of the story.
SEO tools were expected to give a reasonably consistent answer about where a page ranked.
Traffic decline usually pointed to lost rankings, weaker content, or a technical problem.
Search success was primarily measured through visits to the website.
After
Rankings had to be evaluated alongside the search features surrounding them.
AI Overviews, paid ads, and featured results could push an organic listing much farther down the page.
Visibility could no longer be understood through one platform or metric.
Brand mentions inside AI-generated answers became part of the discovery picture.
Traffic remained important, but it was no longer the only evidence of brand presence.

One query made the problem especially clear.

SEMrush, Google Search Console, and the live search experience all told materially different versions of where the same page appeared. AI-generated results and paid placements changed what the buyer actually encountered before reaching the organic listing.

Each measurement was explainable on its own. None described the full experience by itself.

The Measurement Model

A broader way to evaluate search performance

I did not discard the traditional SEO scorecard. I expanded it.

I began using a broader working model so rankings could be interpreted in context rather than treated as the final outcome.

1

Traditional search

Rankings, impressions, clicks, CTR, landing-page traffic, and query movement.

2

Actual search experience

Where and how the brand appears once AI answers, ads, featured results, and other SERP features are considered.

3

Commercial AI visibility

Whether the brand appears, is meaningfully recommended, is described accurately, and has credible sources supporting it across priority non-branded buyer questions.

4

Business + execution signals

What we changed, which priority opportunities are covered, and whether search activity contributes to qualified traffic, conversion, opportunity, or other downstream signals.

Together, these layers provide a more useful answer than "Where do we rank?" They show whether the company is visible, how that visibility appears, and whether it is contributing to meaningful business activity.

How It Works

Turning a traffic anomaly into a strategic roadmap

1
Spot the divergence
2
Validate across sources
3
Inspect the real search experience
4
Lead a broader search-market audit
5
Translate gaps into workstreams
6
Prioritize + execute

The external audit extended the diagnosis across search demand, AI prompts, competitors, citation sources, and existing content. My job was to turn that much larger body of evidence into decisions: what we could improve directly, what depended on external authority, what needed another owner, and what was worth doing first.

The goal was not to replace SEO with a new acronym. It was to understand how buyers were finding answers now and build a practical way to act on what the data showed.

Showing the Work

The diagnosis had to be visible, not theoretical

search performance — directional
Average position
Visibility
Clicks

The initial signal: traditional visibility metrics and click behavior were moving in different directions.

SEO tool
Search Console
Live SERP
✦ AI Overview
Ad
Ad
← your organic result

Different tools could all be technically correct while describing very different parts of the buyer experience.

search-market audit
Search demand
AI prompts
Competitors
Citation sources
Existing content
Opportunity / gap analysis

A specialist audit broadened the analysis beyond the website into the wider search market.

audit → operating backlog
Opportunity
Diagnosis
Intervention
Owner
Evidence
Result
Owned content
External influence
Technical + entity
Measurement

The work became useful once a large research dataset turned into a smaller operating backlog with owners, dependencies, and measurable next steps.

Visuals are generalized to protect current-employer and partner work; the process reflects the actual program.

The Baseline

The biggest gap was not traffic. It was how the brand was understood.

The broader audit showed an important distinction: branded questions could make overall visibility look healthier than it was. Non-branded category and evaluation questions told a much weaker story.

The company was not consistently entering the consideration set, and important differentiation was not always represented clearly in the answers and sources buyers were encountering.

That changed the nature of the work.

This was not only a ranking problem or a content-volume problem. It was a search-market problem: owned content, external sources, technical accessibility, and message consistency all shaped what buyers — and the systems answering their questions — could find.

Owned search

Improve high-value existing pages and build missing buyer-evaluation content where the company can directly act.

External influence

Use citation evidence to prioritize third-party publications, reviews, partner surfaces, expert visibility, and other sources the company can influence but does not control.

Technical + entity foundation

Keep important content accessible, structured, current, internally connected, and described consistently across the wider digital footprint.

Measurement + prioritization

Track a stable set of commercially relevant search questions and connect baseline → diagnosis → intervention → next measurement.

The operating plan separates work that can move immediately from work that depends on cross-functional decisions, so one blocked initiative does not stop the wider program.

What Required Judgment

The hardest part was deciding what search success should mean now

The tools did not agree because they were measuring different versions of the search experience.

I had to determine which differences were normal, which represented a genuine performance issue, and which showed that our existing reporting model was no longer sufficient.

I also had to separate what marketing could directly control from what we could only influence. We could improve our content, technical foundation, and message consistency. We could work with communications, subject-matter experts, and employees to increase credible external signals. We could not control whether an AI system cited us for a particular answer.

The final challenge was keeping the reporting honest. The diagnosis and external audit were completed work. The operating model and early execution were real. Measurable AEO business outcomes were not yet proven, so I did not want early activity presented as finished impact.

That same standard applied to technical recommendations. As "AI SEO" accelerated, I checked emerging tactics against current primary documentation instead of treating every new file, schema recommendation, or optimization idea as a requirement. Sometimes the more valuable decision was to keep working on durable search fundamentals instead of chasing the newest tactic.

Ranking or real visibility?

Does a reported position reflect what a buyer actually encounters?

Control or influence?

Is the gap something we can fix directly on owned properties, or does it depend on third-party evidence and other teams?

Useful tactic or AI-search theater?

Is the recommendation supported by current primary evidence, or are we implementing something simply because it has been labeled AEO?

What This Enabled

The work moved beyond the SEO report

The diagnosis gave the marketing organization a clearer explanation for why rankings, clicks, and traffic were no longer moving together.

It also created a practical starting point for AI visibility. Instead of reacting to individual headlines or purchasing another tool, we had a baseline, four program levers, and a prioritized operating backlog.

The work also began expanding beyond the digital marketing team. Other groups requested context on how search was changing and what teams could do to support visibility.

The measurement story expanded

I reframed the search story around the difference between ranking, appearing, being cited or recommended, earning a click, and ultimately contributing to meaningful business activity.

A roadmap took shape

The audit became a prioritized operating backlog spanning owned search, external influence, technical/entity work, measurement, and cross-functional dependencies.

The work moved beyond SEO

The findings now inform conversations and execution across content, product marketing, communications/PR, reviews and external authority, and digital measurement.

This is still an active program. The market-mapping work is complete, the operating backlog and first execution sprints are taking shape, and the next proof point is whether those interventions measurably improve commercial search visibility over time.

What I Learned

Search performance no longer fits in one dashboard

A ranking can improve while the available space for an organic result shrinks.

A brand can appear inside an AI-generated answer without producing a clean referral trail. It can also rank well in traditional search while remaining almost invisible in the systems buyers are beginning to use for research.

I did not stop measuring rankings, clicks, leads, or pipeline. I stopped asking those metrics to explain the entire search experience by themselves.

The question changed from "Where do we rank?" to something more useful: Where does our brand appear when a buyer looks for an answer, what evidence is shaping that answer, and which part of that system can we actually improve?

Next Case Study

Building the foundation layer for marketing AI

How I turned scattered prompting and disconnected source material into a shared context system that marketing workflows could build from.

View case study: Marketing AI Context Layer →