The Vulnerable 50

Everything-PR's quarterly ranking of the 50 publishers, networks, and affiliate businesses with the highest structural exposure to AI search substitution.

By EPR Editorial Team · June 2, 2026 · 14 min read

Executive Summary

Whose business model breaks first when AI absorbs informational traffic?

The Vulnerable 50 ranks 50 named publishers, affiliate networks, and content businesses on a 0–100 composite vulnerability score. Volume I covers the publishing, affiliate, and consumer-information categories where AI substitution is now measurable in public filings.

Public-market exposure is concentrated. Chegg, NerdWallet, Ziff Davis, Future PLC, IAC/People Inc., and TripAdvisor either occupy multiple positions on the list or carry portfolios containing them.

The Insulated 10 — the structural winners — is included at Section 4.

Key Findings

• Chegg leads the list at 96. First publicly traded company to attribute revenue decline to AI. Stock down approximately 99% from its February 2021 peak.
• Five parents own 31 of the 50 named entities — 62% concentration. People Inc., Ziff Davis, Future PLC, Red Ventures, and Internet Brands.
• Personal finance, health, and recipes are the most exposed sectors. Every named entity in those categories scored above 75.
• Citation share is now the leading indicator. Traffic is the lagging indicator.
• The Anthropic $1.5B settlement (September 2025) established a $3,000-per-work floor that changes the negotiating math for every unlicensed publisher.

Related: Chegg: The Canonical AI Casualty (the Vulnerable 50 case study at #1) · The Insulated 10: Who Wins When AI Eats Search (the other side of the framework — who's gaining ground)

1. The Ranking

Tier 1 — Critical (90+)

Tier 2 — Severe (80–89)

The top 10 share four characteristics: high dependency on Google organic, high substitutability inside AI answer surfaces, monetization concentrated in traffic-fed revenue streams, and limited defensibility outside brand recognition.

Tier 3 — Elevated (70–79)

Tier 4 — Significant (60–69)

Tier 5 — Moderate (50–59)

Tier 6 — Watch (40–49)

2. Methodology

The Composite Score

Vulnerability = (TD + AS + (100 − CC) + RC + (100 − DF)) / 5

Higher composite score = higher vulnerability.

Input Definitions

Data Inputs

• Similarweb and SEMrush traffic estimates (12-month rolling)
• Public company filings (10-K, 10-Q, 8-K) and earnings call transcripts
• Everything-PR Citation Share testing across ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews on 200+ category prompts
• Industry studies from Reuters Institute, Pew Research, Bain & Company, IAB Tech Lab, Digital Content Next, Detailed.com
• Press disclosure via Bloomberg, Press Gazette, Digiday, AdExchanger

Inclusion Criteria

• Material dependency on Google organic referral traffic (estimated >40% of traffic from search at peak)
• Monetization via programmatic ads, affiliate commerce, lead-gen, or traffic-fed subscription
• Identifiable at the company, brand, or division level

Scores are directional estimates. They are not audited financial metrics. Limitations in Section 8.

Substitutability — High vs. Low

Defensibility — The Nine Layers

1. Proprietary data
2. Active community
3. Transaction ownership
4. Brand trust / institutional authority
5. Live information
6. Experiential content
7. Personality / creator-led
8. Paywalls and subscription
9. Original reporting

Vulnerability score is functionally the substitution rate net of stacked defensibility layers. Companies with three or more stacked layers cluster in the lower tiers. Companies with one or zero cluster at the top.

3. Tier-by-Tier Datacards

Tier 1 — Critical (90+)

1. Chegg (Parent: Chegg Inc., NYSE: CHGG | Score: 96)

• Traffic dependency: historically very high; built on Google referral for homework-help queries.
• Substitutability: very high — homework-help and study queries are textbook AI-substituted content.
• Citation capture: low.
• Revenue concentration: very high — subscription model fed almost entirely by SEO funnel.
• Defensibility layers: minimal — no proprietary data, no community moat, no transaction layer beyond payment processing.
• Disclosure: search visits down 51.2% YoY (Q4 2025); stock down approximately 99% from its February 2021 peak ($113.51); two layoff rounds in 2025 cutting more than half the workforce; the first publicly traded company to officially attribute revenue decline to AI. See full Chegg case study.
• Recoverable ground: institutional / B2B partnerships; AI-assisted tutoring repositioning — underway but late.

2. Investopedia (Parent: People Inc., NASDAQ: IAC | Score: 91)

• Traffic dependency: high; legacy SEO authority on financial definitions.
• Substitutability: very high — definitions and "what is" queries are the canonical AI substitution case.
• Citation capture: moderate — still cited by ChatGPT and Perplexity on niche definitional queries.
• Revenue concentration: high — programmatic and affiliate.
• Defensibility layers: brand authority (partial); minimal proprietary data.
• Disclosure: monthly visits down from 40.3M to 26.9M across 2025, a 33.2% decline (Loopex Digital).
• Recoverable ground: original research, market commentary, brand-authority anchoring.

3. WebMD (Parent: Internet Brands / MH Sub I, KKR-owned | Score: 90)

• Traffic dependency: very high.
• Substitutability: very high — symptom-input queries are perfectly substituted by AI engines.
• Citation capture: moderate — still surfaces on condition-specific queries.
• Revenue concentration: high — pharmaceutical display advertising dependent on traffic volume.
• Defensibility layers: institutional authority (partial); minimal community or transaction layer.
• Disclosure: monthly visits down from 122M to 69.5M across 2025, a 43.1% decline (Loopex Digital); the second-largest absolute consumer-web traffic decline of the year.
• Recoverable ground: provider funnels, telehealth, brand-anchored Rx education.

Tier 2 — Severe (80–89)

4. Healthline (Parent: Red Ventures | Score: 88)

High substitutability plus minimal defensibility outside category brand. Symptom queries substituted directly by ChatGPT, Claude, and Gemini. Affiliate and lead-gen tied to first-line health research — the exact intent most substituted. Recoverable: condition communities, treatment-funded programs.

5. Bankrate (Parent: Red Ventures | Score: 87)

AI Overviews now answer rate, APR, and product-comparison queries directly on the search page. Lead-gen affiliate model intact; the traffic feeding it is not. Recoverable: tools tied to actual financial offers and applications.

6. The Balance (Parent: People Inc. | Score: 85)

Personal finance education; partially overlapping Investopedia. Named by People Inc. as among titles most affected by AI Overviews. Recoverable: limited; consolidation pressure inside the People Inc. portfolio.

7. Tom's Guide (Parent: Future PLC, LSE: FUTR | Score: 84)

Monthly audience down 36% in the 18 months to December 2025, falling to 1.97M (Ipsos iris). Part of Future's tech B2C division, where sessions fell 10% and e-commerce revenue 24%; 45 editorial redundancies announced January 2026. Recoverable: hands-on testing, original review video, brand-anchored buying guides.

8. NerdWallet (Parent: NerdWallet Inc., NASDAQ: NRDS | Score: 83)

Public-market pure-play on SEO-fed financial lead-gen. Flagged by Barclays and multiple Wall Street analysts as a structural AI-search loser. Listed by CNBC alongside eBay, Wayfair, Booking Holdings, and Chegg. Recoverable: direct app, deposit and brokerage relationships, advisor matching.

9. Allrecipes (Parent: People Inc. | Score: 82)

The largest recipe property; declines mirror the wider category collapse. Independent recipe creators reporting traffic losses of 30–80% during 2025. People Inc.'s scale offers more diversified distribution than smaller creators, but the SEO-dependent legacy base is severely exposed. Recoverable: branded recipe video, app, member community.

10. CNET (Parent: Ziff Davis, NASDAQ: ZD | Score: 81)

Sold by Red Ventures to Ziff Davis October 2024 for over $100M. Multiple layoff rounds. Tech-review category now competitor to AI engines on most consumer-electronics buying decisions. Recoverable: testing labs, exclusive review access, video execution.

Tier 3 — Elevated (70–79)

The pattern repeats — high substitutability, partial brand defensibility, programmatic or affiliate exposure. Notable callouts:

• PCMag and ZDNet carry the same dynamic as CNET, with PCMag more consumer-search dependent.
• Verywell Health and Verywell Mind are People Inc.'s most-affected vertical inside the portfolio.
• IGN has community and real-time content as partial offsets to programmatic exposure.
• Lonely Planet and The Points Guy anchor Red Ventures' travel and credit-card affiliate exposure; AI engines now recommend cards and itineraries directly.

Tier 4 — Significant (60–69)

People Inc. lifestyle brands (The Spruce, Real Simple, Byrdie, Brides) and Future PLC gaming/science properties (Live Science, Space.com, PC Gamer, GamesRadar). Programmatic-dependent, partial brand moats, mid-substitutability content.

Tier 5 — Moderate (50–59)

Coupon and deals businesses (Honey, Capital One Shopping), and Q&A platforms (Quora, Stack Overflow) where the substitution vector is AI assistants directly answering the question. Quora monthly visits down 28.1% across 2025; Stack Overflow platform visits down 35.5%.

Tier 6 — Watch (40–49)

Local services and transactional businesses (Yelp, Angi, Houzz, LegalZoom-adjacent) where the transaction floor offers partial defensibility. Hollywood Reporter included as a structural signal — Penske Media became the first news publisher to sue Google over AI Overviews.

4. The Insulated 10 — Surprise-Resilient Companies

The Vulnerable 50 is one side of the structural shift. The other side is the set of companies whose business models gain ground as informational traffic compresses. The pattern is consistent: stacked defensibility layers, ownership of either the transaction or the citation, and either a paywall or a community moat at the floor. See the full Insulated 10 hub for the framework, the ranking, and the playbook.

The Insulated 10 is not a fixed list. It is a structural pattern. Any company carrying three or more of the nine defensibility layers — and operating in a category with high transactional intent or low substitutability — clusters in this set.

5. Ownership Concentration

Five parents account for 31 of the 50 named entities — 62% of the list.

The four public parents have responded differently.

IAC / People Inc. rebranded July 31, 2025 and shifted to a 19-brand core focus, scaling back 21 brands. Licensing revenue from OpenAI grew 30% YoY in Q1 2025 — a buffer, not a replacement.

Ziff Davis is acquiring its way out: Lifehacker (2023), CNET ($100M+, October 2024), theSkimm (March 2025), Dwell / Domino / Bob Vila / Field & Stream from Recurrent Ventures (May 2026), and a $1.2B sale of its Connectivity division to Accenture in March 2026.

Future PLC is restructuring under CEO Kevin Li Ying (March 2025). Stock down sharply through 2025–2026. Barclays cut its price target from 640p to 285p. Recent acquisitions include SheerLuxe and BLUSH Talent for £39.9M.

Red Ventures went the opposite direction — selling CNET, narrowing portfolio focus toward verticals it views as more defensible.

The structural question is which parent ends up holding the most defensible portfolio. The most likely path is further consolidation, as private equity and strategic acquirers buy distressed informational properties at compressed valuations and bundle them with subscription, transactional, or community assets.

6. Sector Concentration

Personal finance, health, and recipes are the most structurally exposed — every named entity in those categories scored above 75. The pattern matches the Casualty List finding: information-dense, transaction-thin categories compress fastest. Categories with a transaction floor (legal services, local services) score lower despite high content-layer substitutability.

7. Recoverable Ground

Four structural moves correlate with reduced vulnerability over the next 18 months.

AI licensing agreements. Companies with executed deals: People Inc. (OpenAI, since May 2024 — licensing revenue up 30% YoY); Penske Media (settled with OpenAI March 2026); Axel Springer (OpenAI, approximately $13M per year); Financial Times (OpenAI, $5–10M per year); News Corp (OpenAI, $250M over 5 years; Meta, $150M over 3 years); The New York Times (Amazon, May 2025); Reddit (Google approximately $60M per year, OpenAI approximately $70M per year — $203M total in 2024). Licensing converts content into recurring revenue and improves citation share inside the licensed engine. It does not restore traffic.

Paywall and subscription transitions. Companies that built subscription floors before substitution accelerated — NYT / Wirecutter, FT, WSJ, Bloomberg, Consumer Reports — are reporting digital subscription revenue growth at double-digit rates. NYT added 1.4M digital subscribers across 2025 with digital subscription revenue up approximately 14% YoY each quarter. Subscription is the highest-defensibility business model for content publishers. Most companies on this list do not have one.

Citation share investment. Companies actively investing in being cited inside AI answers — through structured data, source authority, primary research, and prompt-relevant content architecture — are outperforming peers in the same vertical. Citation share is now the leading indicator. Traffic is the lagging indicator.

Lawsuits in progress. Penske Media v. Google (first such suit); NYT and Chicago Tribune v. Perplexity; Ziff Davis v. OpenAI (April 2025); Encyclopedia Britannica and Merriam-Webster v. OpenAI; Reddit v. Anthropic (June 2025); Chegg v. Google (Q1 2025). The Anthropic $1.5B copyright settlement in September 2025 established a $3,000-per-work floor that materially changes negotiating leverage for unlicensed content.

8. Limitations

Traffic estimates are drawn from third-party measurement tools (Similarweb, SEMrush, Ipsos iris, Loopex Digital), not server logs at the named companies — direction is reliable; absolute magnitudes are directional. Vulnerability scores are composite estimates, not audited financial metrics. Where companies have provided more granular disclosure, scoring leans on that disclosure. Where they have not, scoring relies on third-party estimates. Citation Share testing draws on a fixed prompt set of approximately 200 consumer-intent queries; different prompt sets produce different distributions; no queries were logged or retained. The list excludes B2B SaaS publishers, news organizations, government sources, nonprofit reference (Wikipedia), and platforms (YouTube, Reddit, Pinterest) — each warrants its own list. Inclusion is an assessment of structural exposure, not a forecast of imminent failure.

9. The Standing List

The Vulnerable 50 is updated quarterly.

Volume II — Q3 2026 expands the universe to adjacent categories under consideration: tutoring and test preparation, legal information services, local services, translation and language reference, real estate listings, and B2B SaaS content marketing publishers. The Insulated 10 will be updated alongside.

Submissions, corrections, and parent-company filings for Volume II: research@everything-pr.com

FAQ

What is The Vulnerable 50? Everything-PR's quarterly ranking of the 50 publishers, affiliate networks, and content businesses with the highest structural exposure to AI search substitution. Each entity is scored 0–100 on a composite of traffic dependency, AI substitutability, citation capture, revenue concentration, and defensibility.

How is the score calculated? Vulnerability = (TD + AS + (100 − CC) + RC + (100 − DF)) / 5. Higher score = higher vulnerability. The full input definitions are in the Methodology section.

Who tops the list and why? Chegg, at 96. See the Chegg case study for the full breakdown — the combination of very high Google traffic dependency, very high substitutability (homework help is canonical AI substitution territory), low citation capture, and minimal defensibility layers. Chegg is also the first publicly traded company to formally attribute revenue decline to AI.

Which sectors are most exposed? Personal finance (median 84), health and wellness (median 81), and recipes (median 75). Information-dense, transaction-thin categories compress fastest.

Who is gaining ground? The Insulated 10 — led by Reddit, Wikipedia, and the major paywalled publishers (NYT, WSJ, FT, Bloomberg). Stacked defensibility layers — community, paywall, original reporting, proprietary data, transaction — are the common pattern.

What is "citation share" and why does it matter? Citation share is the share of voice a brand or publisher captures inside AI engine answers. As informational traffic compresses, being cited inside the answer is the new visibility. It is now the leading indicator; traffic is the lagging indicator.

How often is the list updated? Quarterly. Volume II ships Q3 2026 and expands to adjacent categories.

Who can submit corrections or filings? research@everything-pr.com