How is AI Search advertising different from search engine advertising?

Search engine advertising buys ranked placements alongside a list of blue links. AI Search advertising places brands inside the synthesized answer itself, in formats native to AI engines, and compensates the publishers whose content was cited to ground that answer. The unit of value shifts from click to citation.

Is pay-per-crawl enough to keep publishers funded?

No. Pay-per-crawl is a useful floor because it makes extraction transactional, but it prices access, not influence. A page fetched once and cited as the source of a high-value purchase decision is worth far more than a page fetched a thousand times and never surfaced. A citation-based ad layer is what closes that gap.

Should CMOs reallocate SEO and SEA budgets to AI Search?

No. SEO and AI Search visibility share foundational signals — clean structure, entities, authority — and reinforce each other. Most teams should keep SEO stable, hold paid search steady, and add a dedicated AI visibility line item rather than reallocate from existing channels.

What can publishers do today to start monetizing in AI Search?

Three concrete moves: structure articles for citation (definition-first prose, named sources with dates, explicit entities), join emerging citation-based ad networks that compensate for influence rather than fetches, and refuse any AI access deal that pays only for crawling without compensating for citation share inside AI answers.

Pay-Per-Crawl vs Citation Economy: The AI Search Ad Model

For 20 years the contract was simple: create content, attract audience, monetize. The audience is now agentic — and pay-per-crawl is not the ad model the open web needs.

For twenty-five years, the open web ran on a contract every publisher understood: create content, attract an audience, monetize the attention. That contract is breaking — not because the audience disappeared, but because it became agentic. Sam Altman's micropayment idea and Cloudflare's pay-per-crawl are useful first attempts. Neither is the ad model the open web actually needs.

Publisher search traffic from Google fell 33% globally between November 2024 and November 2025, according to Chartbeat data aggregated for the Reuters Institute Trends and Predictions 2026 report (January 2026).
French press net advertising revenue is now 19.5% below 2019 levels (BUMP, 2025), and the 'edition and information' category lost a further 8% in digital ad revenue in 2025 even as the broader market grew 11% (Observatoire de l'e-Pub, 35e édition).
Pay-per-crawl prices what an agent costs to fetch a page. It does not price what that page contributes to a brand showing up in an AI answer — and that gap is the entire opportunity.
The agentic web needs a citation-based ad model: brands pay to appear inside AI answers, publishers are compensated for being the cited source.
Publishers that re-architect for citation rather than traffic recovery will compound an advantage over the next 18 months.

What the citation economy actually is

The citation economy is the value chain in which AI Search engines extract and cite publisher content to generate answers, and in which both brand visibility and publisher revenue are tied to being inside those answers rather than to the click. The unit of value moves from impression and click to mention and citation. Pay-per-crawl prices the input — fetching the page. The citation economy prices the output — being the source that grounds the answer a buyer reads.

The link economy and the citation economy, side by side

The old contract was symmetric. Google ranked publishers, sent traffic, and brands paid CPC for placements next to that traffic. Publishers monetized the attention they earned; brands monetized the intent Google routed. The agentic-web equivalent is asymmetric so far: AI engines extract from publishers, synthesize an answer, sometimes name a source, and rarely send a click. In France, JDN reported in May 2026 that the 'edition and information' category captured €504M in digital display ads in 2025, while the platforms swallowed €9.426 billion. The link economy still runs — it just runs without the publishers it used to fund.

What JDN and Nieman Lab reported, and what we verified

Journal du Net, in its May 22, 2026 analysis, ties the French press crisis — at least 930 jobs cut in recent months — to four converging forces: insufficient digital revenue, ad migration to platforms, the rise of retail media and CTV (CTV/VOD ad revenue grew 90% in France in 2025), and the collapse of Google referral traffic. We confirmed the 33% global Google traffic drop against the Reuters Institute Trends and Predictions 2026 report (January 2026), and the platform ad share rising from 67% in 2019 to 76% in 2025 against the Observatoire de l'e-Pub. Separately, Nieman Lab (May 2026) reported Sam Altman, on Nicholas Thompson's podcast, endorsing a micropayment model — roughly $0.17 for an agent summary, $1 for a full human read — to compensate publishers for AI access.

Why pay-per-crawl is not the business model the open web needs

The strongest version of the pay-per-crawl case

The credible version of the pay-per-crawl argument — voiced by Cloudflare, Tollbit (now serving close to 7,000 publisher sites) and ProRata — is that the first step toward fixing the agentic-web imbalance is forcing AI bots to pay for fetching content. The Cloudflare Radar data is hard to argue with: Anthropic's ClaudeBot crawled 23,951 pages for every single referral it sent back to publishers in Q1 2026, and Meta-ExternalAgent sent zero. Charging per fetch turns an extractive flow into a transactional one. That is real progress.

Why pricing access is not pricing influence

Pay-per-crawl prices what an agent costs to fetch. It does not price what that fetch contributes to a brand winning a citation in an AI answer a buyer reads. A single niche review by a credible publisher might tip a $50,000 software purchase decision. An aggregator page may be fetched a thousand times for nothing of consequence. Per-fetch pricing collapses both into the same unit. Sam Altman's $0.17 idea has the same flaw at higher resolution: it monetizes access volume, not commercial influence. The open web has always been funded by the latter, not the former.

What this means for brands and for publishers

For CMOs, media buyers and agencies: AI visibility is a media line, not an SEO subtask

Treat AI Search visibility as a discipline distinct from SEO and SEA, with its own budget, KPIs, and owner. eMarketer projects US AI search ad spending to grow from just over $1B in 2025 to nearly $26B by 2029. Heads of media should ringfence a 2026 test budget for native-ad-for-AI-agent inventory; SEA managers should audit how much of their funnel already depends on AI-Search-driven discovery; agencies should build the practice as a service line before clients ask why a competitor is being cited and they are not.

For publishers: stop pricing the future on traffic recovery

The 33% Google traffic loss is not coming back, and Reuters Institute survey data shows news publishers expect a further 43% decline by 2029. Stop pricing the future on traffic recovery and start pricing it on being the source of record AI engines cite for the queries that matter in your category. That means three operational moves: structure content for citation (clear entities, dates, named sources, definition-first prose), join the emerging citation-based ad layer, and refuse any AI access deal that compensates the input without compensating the influence.

Three signals to watch over the next 18 months

First, AI licensing is migrating from one-off lump sums to usage-based and citation-weighted terms — ProRata's revenue-share model, paying 50% of revenue back to 100+ publisher partners across 750+ titles, is the most explicit early example. Second, AI engines are quietly experimenting with sponsored citations inside answers; the format wars start this year. Third, the structural traffic decline forecast by Reuters Institute respondents forces a strategic, not tactical, response from every newsroom and every brand that relies on them. Whoever standardizes the citation-weighted pricing unit first will set the benchmark for everyone else.

Conclusion

Hold on to this: the audience has gone agentic, and the open-web link economy needs a citation economy successor before another wave of newsrooms — 930 jobs already cut in the French press alone in recent months, per JDN — disappears. Pay-per-crawl is a useful floor. It is not the ad model.

Smalk AI is that ad model — a generative engine advertising network that places native ads for AI agents alongside the publisher content those agents cite, so brands pay to show up in AI answers and publishers are compensated for the influence their content carries, not just the page an agent fetched. What to watch next: which network, platform, or coalition standardizes the first citation-weighted pricing benchmark — the moment that benchmark prints, the category prices in.

Pay-Per-Crawl Isn't Enough: Publishers Need an AI Search Ad Model