The Pharma DTC vs. AI Citation Gap 2026

Executive Summary

The top 15 DTC pharma brands spent an estimated $4.2 billion on U.S. television advertising in 2025.¹ Five engines — ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews — now answer a growing share of consumer drug-research questions before the patient ever sees a TV spot.² Across the indications those brands compete in, AI engines cite the molecule, the clinical source, and the cost-comparison site more often than the brand. The bigger the ad budget, the bigger the gap.

The Pharma DTC vs. AI Citation Gap Index ranks 15 of the highest-DTC-spending brands by the distance between their TV ad rank and their modeled AI citation rank. Five brands — Skyrizi, Rinvoq, Tremfya, Vraylar, and Caplyta — sit at the worst end of the curve. Each spends more than $175 million a year on consumer advertising and is cited inside AI answer engines at single-digit-to-low-double-digit rates within its own category. Four brands — Ozempic, Mounjaro/Zepbound, Entresto, and Farxiga — sit at the other end. They earned AI citation share their ad budgets did not buy.

AbbVie owns three of the five worst gaps. Eli Lilly and Novo Nordisk own three of the four biggest wins. The pattern is not random. It is structural — and it is the new front in pharma marketing.

Methodology

• Engines: ChatGPT, Claude, Perplexity, Gemini, Google AI Overviews. Standard 40 / 20 / 20 / 15 / 5 weighting.
• Prompt set: 75 buyer-intent and patient-intent prompts per indication — diagnosis, treatment options, cost, side effects, alternatives, insurance coverage.
• Brands: Top 15 U.S. DTC TV ad spenders by estimated 2025 outlay.
• Citation Share: Percentage of category-relevant AI answers that mention the brand name (not the molecule, not a competitor, not a class).
• Gap: Spend rank minus citation rank. Positive = ad-rich, citation-poor. Negative = citation-rich, ad-modest.
• Sources: U.S. DTC television advertising estimates triangulated from publicly tracked industry sources including MediaRadar, iSpot.tv, and Kantar Media reporting. Citation Share figures are modeled by the 5W AI Visibility Index methodology and intended for category-level comparison only.
• Version: June 2026 directional benchmark.

The Spend Side

Pharma is the second-largest TV advertising category in the United States, behind auto.³ The top 15 DTC drug brands now spend roughly $4 billion a year on television alone — before digital, before influencer, before unbranded condition campaigns. Spending is concentrated in immunology, diabetes, cardiovascular, weight management, and central nervous system.

The top of the table has been stable for three years. Dupixent has held the #1 spot since 2022. AbbVie’s immunology portfolio — Skyrizi, Rinvoq, and Vraylar — collectively spent more than $1.1 billion on consumer TV in 2025. That is more than Pepsi. More than Chevrolet. More than McDonald’s.

Spending of this magnitude rests on a single assumption: brand recall drives the conversation with the doctor. When the patient asks for the drug by name, the prescription follows. The DTC era was built on that loop.

The loop is being repriced. A growing share of U.S. consumers now begin a health-research session inside an AI assistant — before the doctor visit, before the pharmacist, before the insurance portal. The TV ad never enters the chat.

Top 15 DTC Pharma TV Ad Spenders, 2025 (Modeled)

The Citation Side

AI answer engines do not watch television. They read. They cite. They summarize. And they default to the sources that earn trust at the molecule layer — peer-reviewed literature, FDA labels, Mayo Clinic, NIH, UpToDate, Drugs.com, GoodRx, payer formularies, and patient communities on Reddit and Inspire.

When a patient types “what is the best drug for severe eczema in adults?” into ChatGPT, the answer is structured around the indication, the mechanism, the side-effect profile, and the cost-and-access reality. Dupixent is named. So is dupilumab. So are topical steroids. So is methotrexate. So is the JAK inhibitor class. The 60-second TV spot — the one with the woman swimming, the voice-over warning about conjunctivitis — has no role in that answer.

This pattern repeats across every category in the table above. The molecule and the class compete with the brand for the citation slot. In immunology, where biosimilars are now reshaping the answer surface, the gap widens: AI engines describe what the drug does, name the class, mention the cheapest option, and only sometimes cite the brand-name leader.

Modeled AI Citation Share — Top 15 DTC Brands

The Gap

The Gap Index sorts each brand by the distance between its ad-spend rank and its AI citation rank. Positive numbers mean a brand is paying for an attention surface where AI engines do not amplify it. Negative numbers mean a brand is being cited at a rate its media plan did not earn.

The 2026 Pharma DTC vs. AI Citation Gap Index

The Five Worst Gaps

Five brands sit at the steepest end of the curve. Each one spends nine figures on consumer television. Each one is cited inside AI answers at single-digit-to-low-double-digit rates within its own category. The combined ad spend across these five brands exceeds $1.5 billion.

1. Skyrizi (AbbVie) — Gap +7

Skyrizi is the #2 DTC TV spender in the United States. It is the #9 most-cited drug among the brands in this index. AI engines describe plaque psoriasis treatment by mechanism — IL-23 inhibitors, IL-17 inhibitors, TNF blockers — and cite the class before the brand. Humira biosimilars dominate the cost layer of the answer. Skyrizi appears, but it does not lead.

2. Rinvoq (AbbVie) — Gap +8

Rinvoq carries an additional AI-era headwind: the JAK inhibitor class warning. When AI engines summarize rheumatoid arthritis or atopic dermatitis treatment, the class-level safety conversation surfaces before the brand. Patient-community content on Reddit and Inspire frequently leads with concern, not endorsement. The $425 million ad budget does not change the source mix the AI engines read.

3. Tremfya (Johnson & Johnson) — Gap +8

Tremfya competes in the same psoriasis answer surface as Skyrizi and Cosentyx, against Humira biosimilars, and against a wave of newer IL-23 entrants. AI engines treat the category as crowded and price-sensitive. Tremfya’s $310 million in DTC spend buys television presence. It does not buy citation share.

4. Vraylar (AbbVie) — Gap +6

Vraylar advertises across bipolar, schizophrenia, and adjunctive depression. AI answers in these categories lean heavily on generic comparators — lithium, lamotrigine, generic SSRIs — and on payer-cost guidance. Brand-name citations for newer atypical antipsychotics are sparse. AbbVie is paying TV-era rates for a citation surface that defaults to generics.

5. Caplyta (Johnson & Johnson, post-Intra-Cellular acquisition) — Gap +2

Caplyta joined the J&J neuroscience portfolio when the $14.6 billion Intra-Cellular Therapies acquisition closed in April 2025.⁵ Its gap is smaller in absolute terms but more painful in context: it is a newer brand with a smaller installed prescribing base, and it spends aggressively to build category awareness. AI engines have not absorbed Caplyta into the canonical schizophrenia or bipolar-depression answer. The TV strategy is doing work the AI strategy is not — and the post-acquisition integration is the moment to fix it.

Four Biggest Wins + One Pattern

Four brands earn more AI citation share than their ad budgets predict. Each one offers a different lesson. The fifth slot is the pattern that ties them together.

1. Ozempic (Novo Nordisk) — Gap −14

Ozempic is the largest negative gap in the index — and the most instructive. Novo Nordisk did not buy this position. The brand became a cultural phenomenon through earned media, celebrity adoption, social commentary, and continuous news coverage. Wikipedia, Reddit, mainstream press, and medical news all stack to a single conclusion inside the AI engines: Ozempic is the canonical answer for GLP-1, semaglutide, and weight loss — even where Wegovy is the on-label drug. Earned media compounded into citation infrastructure.

2. Mounjaro / Zepbound (Eli Lilly) — Gap −5

Lilly’s tirzepatide franchise rides the same earned-media wave. Mounjaro and Zepbound are referenced together as the principal Ozempic comparison in nearly every weight-loss and Type 2 diabetes prompt. Coverage of head-to-head clinical performance — tirzepatide vs. semaglutide — gives the brand a citation surface that TV alone could not build.

3. Entresto (Novartis) — Gap −6

Entresto’s win is quieter. Heart-failure guidelines from the American Heart Association and the American College of Cardiology have reshaped around sacubitril-valsartan. AI engines read those guidelines and surface Entresto by name. Clinical authority, not advertising, is doing the work.

4. Farxiga (AstraZeneca) — Gap −6

Farxiga benefits from cross-indication citation: diabetes, heart failure, and chronic kidney disease pull the brand into three different answer surfaces. AI engines describe SGLT2 inhibitors as a class, then cite Farxiga and Jardiance as the two leaders. The CKD label expansion was the catalyst — and CKD is exactly the kind of guideline-defined indication AI engines parse cleanly.

The pattern — what every winner has in common

Each of the four winners earned citation share at a layer above advertising. Ozempic and Mounjaro through cultural and earned media. Entresto through clinical guidelines. Farxiga through cross-indication label expansion. None of them paid their way in. All of them built a source-mix outside the TV plan that AI engines now read. That is the pattern, and it does not reverse.

Why the Gap Exists

Four structural forces are widening the gap between pharma DTC ad spend and AI citation share. Each one is addressable. Most pharma marketing organizations are not yet structured to address them.

1. AI engines cite the molecule, not the brand.

Peer-reviewed literature, FDA labels, and medical reference sites refer to drugs by their generic name. AI engines inherit that convention. When a patient asks about treatment for atrial fibrillation, the engine cites apixaban before Eliquis, rivaroxaban before Xarelto, semaglutide before Ozempic. Brand citations win only where the brand has saturated the cultural and editorial layer above the clinical layer. Few brands do that — and almost none do it deliberately.

2. Payer and cost sources outrank brand sites.

GoodRx, Drugs.com, SingleCare, NeedyMeds, and major payer formularies dominate the cost layer of AI answers. These sites lead with generic name, price comparison, savings card eligibility, and insurance coverage. Brand-owned domains rarely appear. Pharma sites are still optimized for FDA-compliant claims, not for citation by an AI engine.

3. Patient communities are an AI source. Pharma is not in them.

Reddit, Inspire, PatientsLikeMe, and condition-specific Facebook groups are heavily cited inside Perplexity and increasingly inside ChatGPT. Patient narratives outweigh brand narratives. Brands that engage these communities responsibly — and that surface their own real-world evidence in citable formats — show up. Brands that treat them as risk surfaces do not.

4. The earned-media flywheel is the only thing that closes the gap.

Ozempic is not a TV story. It is a magazine cover story, a podcast story, a celebrity-disclosure story, and a guideline-update story — all flowing into AI citation infrastructure. Encyclopedia entries — Wikipedia chief among them — are a disproportionately large source layer for ChatGPT in particular.⁴ The brands that lead the citation rankings did not spend their way in. They earned their way in. That is the pattern, and it does not reverse.

The 5W AI-Era Pharma Playbook

Closing the gap is operational. Six moves. Sequenced. Measured.

Move 1 — Map the molecule answer.

Every brand should commission a citation audit at the molecule layer, not just the brand layer. Map every AI answer for the top 75 patient-intent prompts in the indication. Identify where the brand is cited, where the molecule is cited, where competitors are cited, and where the brand is missing entirely. This is the GEO baseline.

Move 2 — Own the indication entity.

A clean, accurate, well-cited Wikipedia entry for the indication — not just for the brand — is one of the highest-leverage retrieval assets in pharma communications. Most brands have invested almost nothing here. See: Wikipedia for Brand Authority.

Move 3 — Get into the guideline conversation.

Clinical society guidelines drive the AI answer in cardiovascular, endocrine, and immunology. Brand-affiliated KOLs, society sponsorships, and continuing-medical-education infrastructure are not just medical-affairs assets. They are AI-visibility assets. Communications and medical affairs should be planning together.

Move 4 — Build a structured real-world-evidence library.

AI engines reward sources that are structured, dated, schema-tagged, and citable. A real-world evidence library published on a brand-adjacent domain — with structured data, plain-language summaries, and primary-source links — earns citations the brand site itself cannot earn under FDA constraints.

Move 5 — Earn the cultural layer.

Long-form podcast appearances, condition-focused journalism, patient-advocacy partnerships, and the disciplined placement of human stories around the indication are the assets that closed the gap for Ozempic. Pharma communications teams need to treat earned media as a citation pipeline — not as a coverage report.

Move 6 — Measure citation share. Report it to the CEO.

The single most important change is the dashboard. Brand teams that measure share-of-voice on TV but not citation share inside ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews are measuring last decade’s market. The CMO does not need another media-mix model. The CMO needs a citation map.

What the Next 90 Days Look Like

The earnings cycle is the forcing function. Q2 and Q3 2026 calls will surface the same question in every pharma analyst exchange: how is the brand performing in AI discovery. Brand teams that walk into that conversation with a citation map will lead. Brand teams that walk in with TRP and reach numbers will not.

AbbVie’s immunology portfolio is the canary. The three brands in this index — Skyrizi, Rinvoq, Vraylar — collectively spent more than $1.1 billion on TV in 2025 and earned an average citation share of 11 percent inside the categories they compete in. The gap is the largest concentrated exposure in pharma marketing. It is also the largest concentrated opportunity.

Citation share is becoming the new demand layer. In categories where AI engines now answer the buyer-research question, brands that do not earn citation share will not enter the consideration set. The DTC era is not over. The DTC era is being repriced.