There is a specific kind of confusion that only becomes visible in retrospect. You were measuring something real. The numbers were accurate. The methodology was sound. And yet the picture you had of the market was systematically incomplete — not because your tools were broken, but because they were looking at the wrong thing.

This is where most B2B analytics teams stand today. Not failing. Just looking at the surface of a problem that now runs deeper.

What the Classic Stack Actually Measures

Keyword rankings, organic traffic, CTR, conversion rate — these are measures of access. They tell you whether a user found a path to your content, and whether they walked down it. What they cannot tell you is what happened in the user's mind before the click, during the read, or after they left.

Intent was always the thing underneath. The keyword was just the observable trace of an intent — a crude proxy for something far more specific: an expectation, a decision stage, a mental model of what the right answer should look like.

"A keyword is a shadow cast by an intent. We spent twenty years optimizing for shadows." — The measurement gap in one sentence

The classic stack worked well enough as long as the proxy held. And for a long time, it did. If you ranked for the right keywords, you reached the right people. The gap between proxy and reality was small enough to ignore.

That gap has now widened significantly. And a new layer has inserted itself between query and click — one that the classic stack cannot see at all.

The Layer You Cannot See

When a user asks an LLM a question, the system does not return a list of ranked documents. It constructs an answer. That answer reflects a specific framing of the question, a specific selection of what matters, and a specific weighting of which providers, criteria, and trade-offs are relevant.

None of that process appears in your Search Console. None of it shows up in your ranking tracker. It happens above the layer your analytics stack was built to observe.

The Measurement Stack — What Each Layer Sees
Conversions
Who completed an action — but not why, or what they understood first blind to intent
Clicks / Traffic
Who arrived — but not what they expected to find, or whether you delivered it blind to expectation
Rankings / CTR
Where you appeared in a list — not what the list communicated about you blind to framing
LLM Answer Layer
What the market actually hears — framing, criteria, provider positioning, decision logic now measurable
Semantic Engine
Why certain content functions or fails — intent structure, journey gaps, function coverage now measurable

The classic stack measures from the click upward. The new layer sits below the click — at the point where the market forms its expectations before it ever decides to search, compare, or contact.

How LLMs Frame Differently Than Users Ask

This is where the measurement gap becomes concrete. A user asks a question with a specific intent. The LLM answers with a framing that may — or may not — match what the user actually needed. And your content may be entirely present in the training data, entirely absent from the answer.

Example — B2B Consulting, Mid-Market Decision
User asks
"Which consulting firms are right for a B2B company with €15–30M revenue and project-based engagements?"
LLM frames
A general comparison of full-service vs. boutique consulting, with three large firm categories named. No mention of specialization depth, no criteria for project-based structures, no mid-market specialists.
The gap
The user's actual decision criteria — revenue band, invoice structure, risk transfer — never surfaced. The answer was technically accurate and substantively useless for this decision.

Your ranking tracker saw nothing. Your content was indexed. Your keyword coverage was complete. And the market heard something that did not help it choose you.

Two Different Diagnoses for the Same Symptom

This is why the triangulation from Episode 01 matters so much in practice. When organic visibility is strong but LLM presence is weak, you are facing a framing problem — not a content gap. The answer exists. It is not being surfaced in the context where decisions are forming.

Classic diagnosis — Traffic drops
Build more content
Target more keywords
Improve technical SEO
Increase link authority
Optimize for featured snippets
Semantic diagnosis — Same symptom
Reframe decision-layer content
Make criteria explicit, not implicit
Build comparison and trade-off structures
Ground claims with verifiable specifics
Close journey bridges, not keyword gaps

The same traffic drop. Two completely different root causes. Two completely different sets of actions. The classic stack cannot distinguish between them.

The Diagnostic Shift

Moving from keyword-layer measurement to answer-layer measurement is not an upgrade to your existing stack. It is a different question entirely: not "where did we appear?" but "what did the market hear when it asked?"

That shift requires different instrumentation, different analytical categories, and — critically — a different definition of what counts as a gap worth fixing. The next episode addresses exactly that: how intent is forensically reconstructed from LLM response patterns, and what that reconstruction actually reveals.

Sources & Further Reading

Additional sources and methodology notes available on request.