The AEO Tooling Wars: Profound, Otterly, Peec, Ahrefs — Honest 2026 Comparison

According to a Reuters Institute Digital News Report published in June 2025, 51% of news consumers in the United States now get information from AI-assisted search tools at least weekly — up from 14% in 2023. The same report documents that click-through rates from AI-generated answers to publisher websites are running at 3-7% of equivalent Google organic traffic volumes. That is not a rounding error. That is a structural collapse of the economic model that funded digital publishing for two decades.

The numbers from individual publishers confirm the trend. Condé Nast reported in its Q4 2025 earnings call that organic search traffic was down 38% year-over-year. Dotdash Meredith, which operates over 40 brands including People, Investopedia, and Allrecipes, reported search-driven page views down 44% from their 2022 peak. G/O Media's portfolio of former Gawker properties — Gizmodo, Lifehacker, Jalopnik — saw Google-referred traffic fall by more than half, precipitating layoffs in two successive rounds. The pattern is consistent across verticals, with technology, health, finance, and how-to content hit hardest — precisely the categories where AI assistants provide the most confident direct answers.

This is the zero-click world publishers have been warned about since 2023. It is fully operational now. And the question that matters for everyone still operating a publishing business is not whether it is real. The question is: given the new reality, what revenue models actually work?

The Zero-Click Crisis in Numbers

Before moving to the playbook, it is worth being precise about the damage, because the averages obscure important variation that shapes strategy.

The table reveals the strategic principle underneath the chaos: publishers with generic, answerable content are hardest hit. Publishers with local specificity, practitioner depth, or exclusive access are insulated. The damage is not random. It maps precisely onto what AI can and cannot substitute.

The ad revenue implications are worse than the traffic numbers alone suggest. CPMs in news content categories have compressed by 25-40% since 2023 as brand advertisers have consolidated spend into fewer, higher-trust placements and away from the mid-tier that traffic-dependent publishers occupy. A publisher experiencing a 40% traffic decline and a 30% CPM decline simultaneously is looking at advertising revenue down roughly 60% — a number from which recovery through incremental optimization is not possible.

The publishers adapting are not trying to grow their way back to 2022 traffic levels. They are building different businesses.

Why Traffic-Dependent Revenue Models Are Breaking

The implicit assumption behind all advertising-driven publishing is that audiences arrive, consume, and can be shown ads. AI search breaks this assumption at the point of arrival. When a user asks ChatGPT a question and receives a direct answer, no page view occurs. No ad impression is delivered. No cookie is set. The publisher whose content informed that answer receives nothing.

This is fundamentally different from the Google SEO era, where even a zero-click featured snippet would occasionally drive branded searches, where the SERP itself showed ten blue links that could attract clicks. AI search is a terminal destination in a way that Google's SERP rarely was. The user gets the answer; the session ends. The publisher is structurally cut out of the economic transaction.

The secondary effect is equally damaging. Even for the traffic that does arrive via AI-adjacent paths, the user's need is already partially satisfied. They arrive knowing approximately what the article will say — the AI gave them the summary. They read for confirmation or depth, not for discovery. Session duration falls. Pages per session falls. The engaged-reader metrics that programmatic advertisers use to justify premium CPMs decline alongside raw traffic volume.

The AI SEO apocalypse analysis from Signal's zero-click search deep-dive documents that the publishers most exposed are those whose editorial model was built around content volume rather than content authority. They produced high quantities of good-enough SEO content at scale, distributed it through Google, and monetized the resulting traffic programmatically. All three legs of that stool are now compromised: the content no longer differentiates from AI summaries, Google is directing fewer users to publisher pages, and programmatic CPMs have declined along with the traffic.

The survivors are building on a different asset: their authority as a credible, citable source that the AI itself depends on.

Model One: Licensing Training Data to AI Labs

The most talked-about adaptation is also the most opaque: direct commercial agreements with AI labs for access to publisher archives as training data or as retrieval-augmented generation (RAG) source pools.

The market is real. By mid-2025, OpenAI had signed content licensing agreements with more than 20 media organizations, including The Atlantic, Vox Media, The Financial Times, and News Corp. Google reached agreements with Reddit (a $60 million annual deal, disclosed in a February 2024 SEC filing), the Associated Press, and multiple Springer Nature properties. Meta's agreement pool for Llama 3 and subsequent models is less publicly documented but reportedly includes several European newspaper groups.

The economics vary widely by publisher:

Tier 1 (major archives, global brands): $10-50 million annually. News Corp, Financial Times, The New York Times (which is suing OpenAI rather than licensing), major wire services.

Tier 2 (strong niche archives, high-quality vertical content): $1-10 million annually. Established B2B trade publications, specialist media with long archives.

Tier 3 (mid-tier with niche depth): $100,000-$1 million annually. Vertical-specific publishers with well-curated archives in categories AI labs want more training data for.

RAG licensing (real-time access): $0.002-0.01 per retrieval query, or monthly access fees of $10,000-$500,000 depending on volume and exclusivity.

The strategic complication is timing. The window for high-value training data deals was 2023-2024, when model training for the current LLM generation was ongoing. By 2025, the major labs had largely assembled their training corpora, and the most aggressive licensing negotiations were complete. Publishers entering negotiations now are more often discussing RAG access than training data — a different commercial arrangement with different economics.

The negotiating leverage remaining sits with publishers who own unique content types: legal databases, scientific literature, proprietary financial data, and primary-source journalism with named-source exclusivity. Generic information content has declining licensing value as the models themselves become capable of generating plausible substitutes.

For publishers in the negotiation window, the crawler permission economy analysis is essential context — the choice between blocking AI crawlers and licensing access is not binary, and the most sophisticated publishers are structuring tiered access arrangements that preserve AEO visibility (by allowing retrieval-augmented access) while monetizing the broader archive (through training data fees).

Model Two: Subscription Built on Exclusive Access, Not Volume

The second working model is direct subscription, but the approach that succeeds in 2026 is structurally different from the paywalls publishers tried during the 2016-2020 metered access era.

The metered paywall approach said: we have valuable content, and after a free article limit you must pay to continue reading. In a zero-click world, AI provides unlimited reading equivalent for commodity information, so the value proposition of a metered paywall against commodity content has evaporated. Users who previously bumped into the paywall and converted had a pain point — wanting to finish reading an article they had begun. That pain point no longer drives conversion when the alternative is asking ChatGPT.

The subscription model that works instead is built on three things AI cannot provide:

Original reporting with named sources. Perplexity and ChatGPT can summarize publicly available information. They cannot produce reporting that required a journalist to spend three weeks cultivating a source inside a company and then publish that source's specific claims. The Information, which has held $599/year pricing since 2013 and never cut it, has built its entire editorial identity on this distinction. It is not a publication about technology; it is a publication that publishes things other publications do not know. Subscribers pay for access to a reporting product that has no AI substitute.

Proprietary data and intelligence. Bloomberg Terminal is the oldest and most successful subscription product in media, and its durability through every platform shift — print to web, web to mobile, SEO to AI — comes from the fact that it is the source of market data, not a publisher of market commentary. AI assistants cite Bloomberg data. They cannot replace it. Publishers who own proprietary data — their own survey panels, their own tracking of specific markets, their own databases of industry activity — have a subscription value proposition that compounds rather than erodes under AI pressure.

Community and access. The fastest-growing subscription publications in the B2B media space are those that have built community infrastructure alongside content. Semafor, The Ankler, Puck News — the category-leading subscriptions in media, entertainment, and political verticals — all sell access to a community of practitioners as much as they sell content. Weekly calls with editors, private Slack channels, member-only events, direct line to reporters. An AI assistant cannot replicate the experience of being in a room (real or virtual) with the people who run your industry.

Model Three: Newsletter as the Owned-Channel Hedge

Newsletters represent the cleanest hedge against AI-driven traffic loss because they operate on a fundamentally different distribution architecture. An email delivered to a subscriber's inbox does not depend on Google, Perplexity, or any AI intermediary. The distribution is direct, owned, and unaffected by algorithm changes, AI adoption rates, or platform policy shifts.

The newsletter economics story is well-established at this point. Substack disclosed in early 2025 that it had more than 50 million active readers and that its top 10 publishers collectively earned more than $25 million in subscription revenue in 2024. Beehiiv reported surpassing $100 million in gross merchandise volume (subscription revenue processed) in its 2025 annual report. The category is growing even as broader publishing is contracting.

What is less appreciated is how newsletters function as a zero-click hedge for publishers who were primarily web-first. A publisher who builds an email list of 50,000 subscribers has a direct-distribution asset that is immune to search traffic collapse. The monetization options at that list size are substantial:

Sponsorship revenue. At an open rate of 35-45% and a highly qualified audience, a newsletter with 50,000 subscribers can command $2,000-$8,000 per sponsored issue, two to four issues per week, across 48 weeks per year — $200,000-$1.5 million annually depending on audience vertical and sponsor demand.

Subscription conversion. Email subscribers convert to paid subscriptions at 2-8x the rate of website visitors, and they do so without any algorithmic intermediary. The newsletter is the funnel for the subscription business.

Event revenue. A newsletter audience is the most efficient possible event marketing channel. Publishers running annual or quarterly events monetize their email list into in-person revenue that has nothing to do with traffic.

The operational implication for web-first publishers is urgent: build the email list now, before the traffic that populates that list finishes declining. Publishers who waited until their traffic had fully collapsed to build email capture found themselves with nothing to convert. The time to build the owned channel is while you still have the traffic to funnel into it.

Model Four: Branded Research and Intelligence Products

The fourth working model is the most sophisticated and the one with the highest margin ceiling: branded research products sold directly to enterprise buyers as annual intelligence subscriptions.

The model works because it reframes the publisher's core asset from content to intelligence. Every publisher with editorial authority in a vertical — media, technology, healthcare, finance, retail — has accumulated something AI labs, enterprise buyers, and strategy teams value but cannot generate themselves: a continuous observation of that vertical's evolution, an expert editorial judgment about what matters, and a database of original reporting and primary sources.

The research product crystallizes that asset into something a corporate buyer can pay for on a contract, charge to a departmental budget, and justify as professional intelligence rather than media spend.

The economics are compelling. A publisher with genuine authority in a B2B vertical — say, enterprise software, healthcare technology, or sustainable supply chain — can sell an annual research license for $15,000-$150,000 per corporate account. With 50 enterprise accounts, that is $750,000-$7.5 million in high-margin recurring revenue that does not depend on traffic, CPMs, or Google rankings.

The publishing organizations moving fastest in this direction are often former trade publications that had strong subscriber bases among industry practitioners. IDC, Forrester, and Gartner built their entire businesses on this model before "media" and "research" were understood as separate categories. The smaller, verticalized publishers adapting to zero-click are rediscovering what those organizations learned decades ago: enterprise buyers will pay high prices for intelligence that is genuinely exclusive, methodologically credible, and delivered in formats that fit their internal workflows.

Model Five: Events and Community Revenue

Events are the oldest hedge against digital distribution volatility. When display advertising collapsed after 2008, media companies rushed into events. When the 2020 pandemic shut events down, they pivoted to virtual. When in-person events came back in 2022-2023, they came back at premium pricing. The cycle continues because events deliver something that neither AI nor programmatic advertising can: verified professional community in a specific time and place, with the publisher's editorial authority as the trust infrastructure.

The event economics for a well-positioned niche publisher in 2026 are strong:

Annual conference (500-2,000 attendees): $800-$3,000 per ticket, plus $100,000-$500,000 in sponsorship revenue, yields $700,000-$6 million gross per event, at 40-60% margins.

Executive roundtable series (20-40 participants per event, 6-12 per year): $2,000-$10,000 per attendee, limited sponsorship, high margin per event. Marketed as exclusive access rather than general conference.

Virtual briefings and webinar series: $500-$5,000 per attendee depending on format and exclusivity. Lower per-event revenue but scalable with marginal production cost.

The community infrastructure that makes events possible — the Slack groups, Discord servers, peer learning cohorts, and practitioner networks that publishers have built around their editorial authority — also generates recurring revenue independently. Annual community memberships sold to practitioners in a specific vertical at $500-$2,000 per year are a model several newsletter-first publishers have deployed successfully.

Case Studies: Publishers Adapting Well

The distance between abstract strategy and actual execution is large. A handful of publishers provide clear case studies of adaptation working at scale.

The Information. Founded in 2013 by Jessica Lessin as a premium subscription publication, The Information has never relied on SEO traffic for revenue. Its $599/year product is built entirely on original reporting inaccessible elsewhere, and it targets a small, high-value readership (technology industry insiders) who have the professional need and the budget to pay for it. In 2025, The Information reported 40,000 paid subscribers — a $24 million annual revenue run rate from subscriptions alone, before events and research sales. Traffic from AI search changes are irrelevant to its business model.

Bloomberg Media. Bloomberg is the clearest case of proprietary data as the foundation of a subscription empire. Its Terminal product carries roughly 300,000 subscribers at $27,000/year — a $8 billion revenue base that AI assistants depend on as a data source rather than compete with. Bloomberg's consumer-facing media properties face the same zero-click pressures as any publisher, but the Terminal business is structurally insulated and actually benefits from AI citations to Bloomberg data.

Axios. Axios has been the most aggressive large publisher in the newsletter-plus-events pivot. Its local news subscription model, Axios Local, sells advertising and subscriptions in individual metro markets based on owned email distribution, not Google traffic. Its pro subscription line — Axios Pro — sells vertical intelligence products to enterprise buyers at $500-$1,000 per user. In March 2026, Axios reported that its pro subscription and events revenue exceeded its advertising revenue for the first time — a structural inversion that signals where the business is heading.

The Athletic. Acquired by The New York Times in 2022 for $550 million, The Athletic is the canonical case of sports journalism moving to direct subscription at scale. Its subscriber base of roughly 3.5 million paying customers generates revenue per reader that is 4-8x what advertising-supported sports coverage achieves. The AI zero-click dynamic is less severe in sports than in other verticals — because sports coverage is highly time-sensitive and local, two properties that insulate it from AI commodity substitution — but the subscription model also means traffic loss would be less damaging regardless.

The AEO-to-Subscription Funnel: Publishers Becoming Citation Sources

The most strategically sophisticated publishers in 2026 are not just surviving zero-click — they are using AI citation to build brand awareness that feeds their subscription and licensing businesses. The insight is that being the most-cited publisher in AI answers about a topic creates brand recognition at the point of intent, even when no click occurs.

This is what we mean by monetizing authority as a citeable source rather than monetizing traffic. When ChatGPT answers a question about media industry economics and cites a specific Signal analysis, the user may not click through to read the full piece. But they learn that Signal covers media industry economics with depth, and the next time they are considering a subscription or looking for a research partner, the brand is already associated with expertise in their mind.

The relationship between AI citation visibility and traffic works differently than traditional SEO — citations build awareness but not immediately measurable traffic. The publishers building the right response are therefore investing in two parallel tracks: AEO infrastructure to maximize citation frequency, and direct conversion mechanisms (newsletter sign-ups, trial subscriptions, content upgrades) that capture the intent signal when a user does arrive directly.

The playbook for becoming an authoritative AI citation source overlaps significantly with subscription conversion strategy:

Original research and data. Publishers that conduct their own surveys, build their own tracking panels, and publish original datasets become cited because they are the primary source — not a secondary summary. The citation drives awareness; the exclusive data drives subscription.

Named-source reporting. AI assistants cannot fabricate named sources, so reporting that includes specific attributable quotes and claimed facts from identified individuals gets cited when the information is unique. That citation signals to readers that the publisher has access that others do not.

Schema-rich, extraction-friendly content. The technical infrastructure of AEO — FAQPage schema, HowTo schema, well-structured H2 headings, answer-shaped passages — enables higher citation rates from the same underlying content investment. The entity context and schema markup framework explains the mechanics; for publishers, the practical implication is that editorial investment in original research should be paired with structural investment in making that research quotable.

The Zero-Click Publisher Playbook: Step-by-Step

1. Audit your current revenue mix against the zero-click reality. Break down your current revenue by source: programmatic advertising, direct advertising, subscriptions, events, licensing, and other. Calculate what percentage of each revenue line depends on search-driven traffic. That percentage is your zero-click exposure. Publishers with more than 60% exposure to search-driven programmatic revenue need to restructure; publishers with less than 30% are already largely hedged.

2. Build the email list immediately, using whatever traffic you have left. Every new reader who arrives via search is a potential email subscriber. Publishers installing aggressive but respectful email capture — exit-intent overlays, content upgrade offers, inline newsletter prompts — are converting traffic into owned-channel assets before that traffic disappears. A subscriber who signed up from search traffic in 2025 will still receive the newsletter in 2028 even after the search traffic is gone.

3. Define your exclusive-access value proposition for subscriptions. What does your publication know or have access to that AI cannot substitute? Named sources in a specific industry? Proprietary data? A community of practitioners who trust your platform for peer exchange? Identify it clearly, build the editorial infrastructure around it, and price it at a level that signals professional value rather than commodity content.

4. Negotiate AI licensing while leverage remains. If you have a content archive with genuine editorial quality and vertical depth, the negotiation window for training data licensing is still partially open — but narrowing. Research which labs are actively seeking content in your vertical. The RAG access market is growing even as training data deals slow down; real-time access arrangements where AI systems query your content as a retrieval source are becoming a recurring revenue line.

5. Launch the research product. Identify the two or three topics in your vertical where your editorial observation is most defensible and most valuable to enterprise buyers. Build a branded research product around those topics — an annual report, a quarterly intelligence briefing, an interactive database. Price it for enterprise budgets, not consumer wallets. Sell it directly to companies who operate in your vertical.

6. Invest in AEO infrastructure to maximize citations. Being cited by AI assistants is now a brand-building channel, not a traffic channel. Invest in the technical and editorial infrastructure that maximizes citation rate: FAQ content at scale, HowTo markup, answer-optimized headings, original data that makes you the primary source. The ROI is not page views — it is brand associations built at the moment of intent. See how publishers are adapting their AEO infrastructure in response to AI Overviews for the technical specifics.

7. Build the event product around your highest-trust audience. Identify the segment of your audience with the highest practitioner density — the readers who are industry insiders, not casual consumers. Design an event or roundtable series specifically for that segment. Price it at a premium. The publisher's editorial authority is the trust infrastructure that makes the event worth attending; you are monetizing the credibility you have already built.

8. Measure the right things. Stop optimizing for page views and organic sessions. The metrics that matter now are email subscriber count and growth rate, subscription trial start rate, brand search volume (as a proxy for AI-driven awareness), citation share across major AI assistants, and direct revenue per subscriber. Publishers optimizing against the old metric set are navigating by a map of a country that no longer exists.

What Doesn't Work

For every adaptation that is working, there are strategies that are clearly failing — worth naming specifically because many publishers are still trying them.

Doubling down on SEO volume. Publishers responding to traffic decline by publishing more SEO-optimized articles faster are accelerating toward a cliff. The bottleneck is not content production; it is the fact that AI answers the queries the content was targeting. More content does not fix this problem. It is the wrong medicine for the disease.

Paywalling commodity content. Erecting a subscription wall in front of content that AI provides freely in answer boxes does not create subscription value — it creates friction without value proposition. Users bounce; they get the answer from the AI instead. The subscription has to be built around exclusive content, not around restricting access to generic content.

Cutting editorial staff to preserve margin. Publishers that are cutting editorial staff to protect margins in the face of revenue decline are accelerating their irrelevance. The exclusive reporting, original research, and practitioner-grade analysis that differentiate them from AI outputs require more editorial investment, not less. The cost-cut path leads to commodity content with no subscription value and no AI citation authority.

Fighting AI companies without a business model replacement. Multiple publishers are pursuing legal action against AI labs for unauthorized training data use — a reasonable position on intellectual property grounds. But legal strategy is not a revenue strategy. The publishers most focused on litigation are in several cases the same publishers most exposed to structural revenue collapse. Winning a copyright case does not rebuild a subscription business.

The Long View: Publishers as AI Infrastructure

The publishers most likely to thrive in the next five years are those that reframe their role from content producers to AI infrastructure. The AI systems that consumers interact with daily — ChatGPT, Perplexity, Claude, Gemini — need credible, original, up-to-date information to provide accurate, trustworthy answers. They need publishers.

The negotiating leverage publishers have is real, but it is time-sensitive. As AI models become more capable of generating plausible synthetic information, the marginal value of real journalism declines in AI training pipelines. The window to monetize editorial authority at scale — through licensing, RAG access, branded research, and citation-authority-driven subscriptions — is open now.

The publishers that recognize this moment clearly, and build revenue structures around their authority rather than their traffic, are the ones that will still be operating five years from now. The publishers that wait for traffic to recover are waiting for something that will not come back.

The AI search cannibalization data by industry confirms that no content-heavy publisher vertical has escaped significant traffic decline — and the trajectory suggests the decline continues through 2027 at minimum. The companies making the transition now have a structural advantage over those that act in 2027: they are building subscription bases, email lists, and enterprise relationships before those assets become expensive to build from scratch.

Takeaway: The publishers surviving zero-click AI search are not fighting the trend — they are building different businesses on top of their existing authority. Training data licensing monetizes their archives. Direct subscriptions built on exclusive reporting, proprietary data, and practitioner community monetize their depth. Newsletter infrastructure monetizes their audience directly, immune to algorithm changes. Branded research products monetize their analytical authority with enterprise buyers at margins programmatic advertising never approached. The structural shift is real, irreversible, and already mostly complete for the hardest-hit content categories. Publishers who move through all five models in the next 18 months — hedging traffic, building owned channels, monetizing authority, and investing in AEO citation infrastructure — are the ones with viable businesses in 2028.