Citation Share Replaces Impression Share: the 90-Day Plan to Dominate AI Search

Your Brand Is Invisible on ChatGPT, Yet You Still Optimize for Blue Link CTR

I look at 15 sites a week. They all still measure SEO like it’s 2019. Blue link CTR. Average positions. Organic clicks from 10-link SERPs.

Then I ask the one question that matters.
Not to them. To ChatGPT.
« Name the three best spare parts suppliers for industry X in France. »

Silence.
Their brand doesn’t appear.
None of my clients do either.

One of my clients, a B2B marketplace, spent €2.3 million on Google Ads in 2025. Hyper-optimized campaigns, precisely tuned bids, controlled landing pages. Paid search ran like clockwork.

But in January 2026, I did the audit.
36 strategic queries — precise questions their buyers were asking Perplexity, Gemini, ChatGPT.
Zero citations.
Nothing.
Not a single mention.

€2.3 million spent to buy clicks, not to build presence in AI.
That’s when everything shifted.

Citation Share: the Metric That Replaced Impression Share

Jason Shafton spent years at Google scaling the Ads platform.
Billions of dollars in spend.
He knows how an auction works.

In his article for Search Engine Journal — « I Helped Scale Google Ads To Billions – Here’s How I’d Build An AI Search Strategy Today » — he stakes a claim PPC agencies should sit with:
« Impression share ran the auction years. Citation share runs this one. »

Impression Share was the percentage of times your ad showed for a given query.
The platform measured, we optimized.
Simple.

Today, the game has changed.
AI assistants don’t rank pages.
They generate answers.
And in those answers, they cite sources.
Or they don’t.

Citation Share is the percentage of AI responses in which your brand is cited, across a defined corpus of queries.
A metric almost nobody measures yet.
And it’s the one that determines your real visibility in 2026.

Three lessons Jason Shafton draws from the Google Ads era:

• Platforms reward signals they can measure, not effort.
• Budget must follow facts, not habits.
• Every technology shift transfers value to the fastest adapters.

I’ve observed the same thing in SEO since 2016.
Those who understand the signals AI models use to cite gain an advantage nothing can close.

Why the DOSE Framework Applies Perfectly to AI Search

I’ve used the DOSE framework since 2018.
Guillaume Attias at BMO Academy taught it to me.
Discovery, Organization, Semanticization, Expansion.
A process for building semantic architectures that engines understand.

Back then, it was for Google.
Clusters.
Pillar pages, supporting pages, deep internal linking.
Result: entire clusters of queries ranking without additional backlinks.

But the arrival of AI Overviews, ChatGPT, Perplexity validated this framework like never before.
Why?
Because these models don’t hunt for 10 blue links.
They hunt for a reference entity on a given topic.

And they detect that entity by reading the depth, coherence, and semantic redundancy of content.
That’s exactly what DOSE structures.

Phase D: I discover missing entities on the site.
Phase O: I organize content into semantic silos.
Phase S: I create ultra-complete pillar pages covering every facet.
Phase E: I link everything so no gaps remain.

AI models love this.
They see a site structured like a well-written book, not a patchwork of content.
And they cite.

La structure du sprint 90j repose sur un pipeline strict : audit → architecture → production → mesure. Chaque phase cible les 36 requêtes stratégiques avec une logique de flux pondéré, montrant où la valeur se cristallise.

The 90-Day Sprint in 4 Phases: What I Deployed for the B2B Client

No betting on randomness.
No "flood it with content."
Each supporting page reinforces a pillar.
Each pillar points to a cluster of AI queries.

The editorial budget?
Cut 40% from peripheral content.
Concentrated on the 30 cluster pages.
Where AI models actually read.

I also added co-citation signals.
A few guest posts on solid industry sites, citing our pillar.
Not for PageRank.
So models see third-party entities pointing to ours.

Le résultat du client B2B ne sort pas de nulle part. Voici comment 13 semaines d'architecture sémantique rigoureuse ont transformé zéro citation initiale en 23 citations réparties sur 36 requêtes stratégiques — et surtout, en +820% d'impressions mensuelles.

En 13 semaines, trois métriques ont basculé : du néant à la domination. Citations brutes, Citation Share et impressions mensuelles — les trois leviers qui prouvent que la stratégie a marché.

Real Result: From Zero Citations to 23 Citations Across 36 Queries in 13 Weeks

13 weeks after launch, I redid the audit.
Same 36 queries.
Same prompts.

January 2026: 0 citations.
April 2026: 23 citations.

Citation Share moved from 0% to 64%.

But the number that really impressed the client was impressions.
A citation is a mention.
But a citation in an AI response read by 400 buyers a month is raw visibility.

Monthly impressions from AI citations went from 1,200 to 11,040.
A +820% increase.

And all without spending one more euro on Google Ads.
Just architecture.
Semantics.
Patience.

The best part?
Conversions from these citations — tracked via a discrete UTM parameter in cited articles — generated revenue equal to 17% of what paid search delivered, with an editorial investment 4 times smaller.

Stop Creating Content at Random. Structure to Become the Reference Entity.

The trap I see everywhere: brands fill their blogs with articles.
150 articles.
300.
600.
Each team pushes their topics with no master plan.
Result: isolated articles, never cited by models.

AI Search doesn't reward volume.
It rewards organization.

At my B2B client, we killed 60% of ongoing production.
We restructured.
We put the €12,000 editorial effort in the right place.
The semantic cluster became a citation machine.

You can't bid on a citation.
You can't pay for placement in ChatGPT.
But you can build the most structured site in your market.
The one models will recognize as the authority.

And when Gemini searches "who writes the best customer segmentation guide for logistics"?
It finds your pillar.
Not your competitor's 47 scattered articles.

So, are you still measuring SEO by blue link CTR?
Or do you start tomorrow, tracking your Citation Share?

Stéphane Jambu

SEO & AI Engineer

I build growth systems / AI / Neuroscience | 650+ clients · 80 LinkedIn testimonials · 30 years of expertise · 15 years of systems running without me.

Follow on LinkedIn