AI: how publishers survive

Person

Who is Malte Landwehr?

Malte Landwehr is an SEO manager with more than twenty years of experience, formerly Head of SEO at Idealo and today in the team of Peec AI.

By his own account, Landwehr had the decisive LLM aha moment in South Korea, when the speed of change in search became apparent to him for the first time.

Career

How did Malte Landwehr come to Peec AI?

Landwehr moved to Peec AI because he recognised AI-driven search as the next big change in SEO and wanted to act fast.

Peec AI is a tool for LLM visibility: it measures how often ChatGPT, Perplexity and Gemini mention brands or name them as a source, and where publishers end up in these answers.

Effects

What consequences does AI have for publisher SEO?

AI changes Publisher SEO through traffic decline from Google, falling domain authority and increasing automation pressure on standard content formats in newsrooms.

Which content types are particularly at risk from AI?

Particularly affected is content that machines can produce similarly well: service tables, overview articles and simple explainers. Here the value creation shifts away from mass output towards independent research.

Protected, on the other hand, is content with a local angle, own interviews and exclusive data. What an AI cannot reproduce without live research retains editorial value.

How does AI change the domain authority of publishers?

Domain authority loses weight as a blanket trust anchor: when users get answers directly in the LLM, what counts is no longer the brand as a collective term, but whether the LLM cites.

Instead of a trust score at domain level, a trust score at article level emerges. LLMs decide anew per query which source they classify as citable.

Pillar channel

Does Google Discover still have a future for publishers?

Google Discover remains the highest-reach channel for publishers in the Google cosmos, but is increasingly endangered by AI summaries and Google's own content.

According to Landwehr, around seventy percent of Google traffic at German publishers currently comes from Discover. The feed is thus the most important lever for reach, but at the same time the biggest dependency if Google plays out AI answers directly in the feed.

Which levers work in Google Discover despite AI?

Discover traffic today arises from anticipating news situations, high-quality images, mobile-first layout and personal brands. Content with its own journalistic signature escapes AI summaries more often than standard texts.

The biggest operational lever is refresh and republish: newsrooms update existing articles in substance and play them out again, so Discover traffic carries for weeks instead of just hours.

Why is evergreen content so important in the AI age?

Evergreen content provides the stable base when news traffic breaks away through AI Overviews. It is the basis for regular republishing and for visibility in ChatGPT, Perplexity and Google AI Mode.

AI bias

How do AI biases influence news consumption?

AI biases distort news consumption because LLM answers weight sources differently, spread hallucinations and undermine trust in research.

Which sources an LLM cites depends on training data, crawling frequencies and model updates. This fragments opinion formation: users no longer see a selection of headlines, but a pre-digested answer in which a small group of sources dominates.

Strategy

How should publishers adapt their strategy?

Publishers adapt their strategy by strengthening human content, establishing journalists as personal brands and considering licensing deals with AI providers early.

The strategic question is no longer "How do we rank?", but "What can AI not replace us for?". Own research, local knowledge and personal brands are the areas in which publishers maintain differentiation.

Daily business

Which measures matter most in daily business?

In daily business, three levers decide the AI robustness of a newsroom: human content, personal brands of the authors and a solid SEO setup for publishers.

Which content must always come from humans?

Human duties remain research, reportage, local news and investigative work. These are the areas in which eyewitness presence, contact with sources and legal responsibility cannot be delegated.

Newsrooms label these formats visibly: in the article itself, in the byline and in the structured data. Readers and LLMs thus recognise the difference to generated content.

How do you build journalists as personal brands?

Personal brands arise via own profile pages, consistent author schemas and a visible byline strategy. Authors who regularly appear in LLMs gain visibility independently of the publisher.

Profiles must link, stay current and position themselves thematically. Generalists have a harder time in the AI age than sharply positioned specialist journalists.

What does solid SEO for publishers look like in the AI era?

Solid SEO for publishers means: fast indexing, structured data, a clean crawl path and consistent refresh and republish for evergreen topics.

Best practices in editorial SEO matter more than individual hacks: continuity beats tricks in daily business.

LLM optimisation

How do you optimise content for LLMs?

Publishers optimise content for LLMs through questions in headings, short summaries, schema markup and well chunkable structures, without manipulation.

LLMs pull answers in bites from the web. Anyone who structures their content into W-questions, concise definitions and short paragraphs ends up more often as a citation in ChatGPT, Perplexity and Google AI Mode.

LLM building blocks

Which building blocks pay directly into LLM visibility?

Three building blocks decide whether a publisher ends up in LLM answers: chunkable structures, matching schema markup and answer-oriented headings.

Which structures make citing easier for LLMs?

W-question headings match the way users formulate in chat. Short summary sentences directly after each heading make a paragraph citable for LLMs, ideally in twenty to twenty-five words.

Important is the independence of the paragraphs: bridging sentences such as "as already mentioned" destroy citability, because an LLM cannot form a meaningful excerpt without context.

Which schema markup do publishers need?

For publishers, Article, NewsArticle, FAQPage, Person and HowTo count above all. Cleanly maintained, they make content machine-readable without manipulation being necessary.

Decisive is the consistency between visible text and schema data. LLMs rate discrepancies as a signal for low-quality content.

How do you chunk content for LLM answers?

Chunking means: every paragraph, every list, every table stands on its own. Anyone who chunks breaks long prose into short blocks with their own focus that an LLM adopts without a break in context.

In practice that means: at most two sentences per paragraph, lists with three to four items and tables with unambiguous column headers.

Monitoring

What role do monitoring and SEO details play?

Monitoring shows when LLMs mention or cite publishers. Tools such as Peec AI make brand visibility measurable.

How do you distinguish brand mention and source citation?

Brand mention means that an LLM names the publisher as a known entity without linking, a branding effect. Source citation means that the LLM plays out a specific article as evidence, with direct reach value.

Both have different levers: branding grows via reach, frequency and personal brands; citability grows via cleanly structured, chunkable content with concise answer sentences.

Which SEO details pay into LLM visibility?

In daily business, LLM visibility is closely linked to classic news optimisation: speed, indexing, structured data and reaction to news situations are the details that decide in Discover as in LLMs.

Anyone who performs strongly in Google Discover often wins in LLMs too. Both systems rate similar signals: freshness, clean structure and brand trust.

Scenarios

Which future scenarios are there for publishers?

Publishers face three scenarios: licensing deals instead of traffic, prompt positioning as a business model and a wave of solo publishers who use AI for scaling.

What does the optimistic licensing scenario look like?

In the optimistic scenario, LLM providers pay publishers for training data and live access to content. Publishers become the curated source that LLMs necessarily need to deliver current and legally sound answers.

In this model, revenue shifts from click-driven to licence-based. Reach is no longer an end in itself, but bargaining power for deals with OpenAI, Google, Anthropic or Perplexity.

What does a pessimistic scenario mean for publishers?

In the pessimistic scenario, AI search cannibalises the ad market without LLM providers compensating publishers. AI Overviews and LLM answers answer the query, the click to the publisher does not happen.

Then direct channels, newsletters, audio and video become vital for survival: reach that the publisher controls itself and does not have to collect via a platform.

Realistic is a mixed model: Large publishers negotiate deals, small solo publishers automate production and position themselves as niche brands that LLMs choose as a source.

Conclusion

Conclusion: how can publishers hold their ground in the AI age?

Publishers hold their ground in the AI age via several channels: direct, newsletters, YouTube, Spotify and Republishing in Google Discover.

Anyone who relies only on Google traffic is too dependent. Anyone who builds reach via own platforms, newsletters, Google News, audio and video and republishes consistently at the same time has the most stable basis, no matter in which direction AI search develops.

Transparency

Sources used

• Own experience from eight years of publisher SEO at publishers such as Axel Springer, Funke, Ippen and t-online.
• Publisher SEO Podcast: "Episode with Malte Landwehr on AI search and publisher strategy on Spotify (in German)", overview of all episodes.
• Malte Landwehr: "Profile and publications".