LLM Industry Benchmark 2026: Which Sources Shape Media & Publishing Recommendations

Note: This article is an excerpt from the LLM Industry Benchmark. The complete analysis, including results for all ten industries, is available in the benchmark itself.

In short:

• In the benchmark, comparison pages lead the retrieved source types for all four search intents in media and publishing.
• Across all ten industries, how-to guides lead the advice mix. Media and publishing follows a different pattern.
• Mentions are concentrated differently from one industry to another. In media and publishing, three entities account for 32 per cent of the top 20 share of voice.
• All findings refer to this prompt set, this time period (August 2026, German market), and this setup.

Methodology and data corpus

The benchmark is based on a clearly defined data corpus. The following sections describe its structure and the key figures derived from it.

Prompt set and structure: The prompt set comprises 1,000 standardised bottom-of-funnel prompts, with 100 commercially relevant prompts per industry. These are spread across ten topic clusters per industry. The prompts are brand-neutral and geared towards purchasing decisions, meaning questions people ask shortly before subscribing, buying, or booking.

Search intent: Each industry covers four search intents with 25 prompts each:

• Best/Top/Test winner: Recommendation questions, such as “Which streaming service is best for…?”
• Comparisons: Contrasts, such as “Which is better, A or B?”
• Prices/Where to buy: Questions about costs and sources, such as “Where can I subscribe to…?”
• Criteria/Guidance: Questions about selection criteria, such as “What should I look out for when choosing…?”

Market, language, and time period: The prompts are in German and geared towards the German market. Answers from 1–31 August 2026 were analysed.

Systems and repetitions: Gemini, ChatGPT, and Google AI Overviews were observed. Each prompt was executed 15 times, five times per system. Individual responses were treated as samples, which allows fluctuations to be classified more reliably.

Data source: The data comes from Peec AI exports. Fifty brands per industry were monitored, and the report focuses on the top 20 brands by visibility.

Key figures:

• Visibility: How frequently an entity appears in the responses to the prompt set.
• Share of Voice (SOV): The proportion of an entity within the top 20, normalised to 100 per cent.
• Sentiment: The tone of the responses, rated on a scale of 0 to 100.

Retrieval is not citation: The source analysis measures retrieval, meaning the pages a system accessed as an information base. A retrieved source does not automatically appear in the final answer, and a brand mention does not depend on it.

Search intent and source patterns

Search intent appears to influence the type of source that shapes a response, and that pattern differs by industry. The same type of question draws on different source mixes depending on the sector. In Health & Wellness, for example, articles top three of the four intent lists. In Media & Publishing, comparison pages top all four.

The industry therefore serves as the lens, not the scoreboard. The useful question is which source types matter for which search intent within one’s own product and service field.

Source types in media and publishing

The table below compares the leading source type per intent in media and publishing with the cross-industry picture across all ten industries.

These figures are aggregated. The cross-industry percentages do not refer to media and publishing alone. “Other” is a residual category and not a specific format.

Four patterns stand out in the media and publishing data:

• Comparisons lead across intents. Comparison pages lead all four intent lists, not only prompts that explicitly ask for a comparison. This may indicate that comparison content shapes answers to recommendation, pricing, and guidance questions as well.
• Guidance follows a different path. Across all industries, how-to guides lead the criteria and guidance mix with 22.33 per cent of retrieval. In media and publishing, comparison pages hold that position, with how-to guides ranking second.
• Product pages stay relevant. Product pages feature in every list and rank second for recommendations and pricing.
• The pattern is not universal. Other industries show different leaders, so media and publishing findings should not be transferred directly to other sectors.

AI visibility focus

An average value does not show whether mentions are spread across many brands or concentrated on a few. The benchmark therefore looks at two concentration metrics within the top 20 of each industry:

• HHI: Describes the concentration of SOV shares. Higher values indicate a more uneven distribution.
• Top 3 share: The combined SOV share of the three largest contributors.

For media and publishing, the benchmark reports an HHI of 723 and a top 3 share of 32 per cent. Almost a third of the SOV within the top 20 falls on three entities. This is a relatively low concentration signal for this setup. It describes the distribution of mentions, not market shares or market opportunities.

For context, media and publishing reaches the following averages:

• Visibility: 9 per cent (across all ten industries: 6.94 per cent)
• SOV: 3.15 per cent (across all ten industries: 3.72 per cent)
• Sentiment: 54.0 (across all ten industries: 54.58)

Three entities show the highest visibility in this dataset: Spotify (24 per cent), Apple Music (13 per cent), and WordPress (12 per cent). Spotify and Apple Music are alternatives within the same decision, namely which streaming service to choose. WordPress plays a similar role in a different category, namely which platform to use for a website. What links the three is their role as products users can select. This may indicate that the prompt set covers media and publishing as a subscription and production ecosystem, where familiar category names surface for purchase-near questions. This reflects the benchmark snapshot. It is not a statement about overall market leadership or brand strength.

Concentration scores help explain how mentions are distributed. They do not explain why a particular brand appears. For that, it is worth looking at the questions asked and the pages accessed behind your own results.

What does this mean for SEO teams and marketers?

The data reveals a clear pattern: depending on intent and industry, responses draw on different types of sources. That can help teams position themselves more strategically. The goal is not to be visible everywhere, which is hardly realistic. Instead, teams should understand what they offer, where they have well-founded answers, and for which questions visibility is genuinely valuable. The benchmark points to several areas to consider:

• Handle answer types separately. A comparison should explain alternatives, advantages, and trade-offs. A guide should clarify requirements or support a task. Practical implication: a single general article rarely does both jobs well.
• Keep prices and terms consistent. Comparisons carry weight in purchasing decisions. Practical implication: check that plans, price components, contract durations, and cancellation terms are described identically on product pages and in relevant comparisons.
• Explain the criteria and link them to offers. Articles and guides dominate consultation questions. Practical implication: explain selection criteria, such as use cases, capabilities, and pricing, and link them directly to suitable offers so that advice and offer do not diverge.
• Prioritise digital PR. Identify the most important third-party sources per intent and check what they say about your brand. Not every publication is equally relevant. Articles that clearly highlight your company, services, or products matter most, but topics higher up the funnel also count.
• Keep information consistent. Product names, features, prices, and terms should match across your website and external sources. Outdated or conflicting information increases the risk of inaccurate statements in generated answers.
• Treat monitoring as ongoing work. Track brand mentions and retrieved sources separately, ideally with a consistent prompt set. Analysis by category, intent, and system complements traditional search and conversion metrics.

Limits of expressive power

The results apply only to this prompt set, this time period, and this setup. They do not represent a complete measure of an industry’s or brand’s visibility in AI systems. In particular:

• The prompt set shapes the outcome. Different questions or wording can lead to different answers.
• Individual responses are samples. Small differences should not be overinterpreted.
• Three systems are not one system. Gemini, ChatGPT, and Google AI Overviews cannot simply be treated as equivalent.
• Knowledge and search are not separate. The source mix alone does not reveal causal mechanisms.
• Dialogue and personalisation are missing. Typical purchasing journeys cannot be reliably described from this data.
• Sentiment is not a quality assessment. It describes the tone of the responses in the dataset, not customer satisfaction with a brand.

Read the full benchmark

This article provides an excerpt. The complete LLM Industry Benchmark includes, among other things:

• The source mix for each of the ten industries and all four intent types
• Visibility, SOV, and sentiment across the benchmark
• HHI and top 3 share within the top 20 of each industry
• The leading entities observed per industry
• The cross-industry shares of the source types
• The industry scorecards and the complete methodology

Download the full LLM Industry Benchmark.

Looking to strengthen your brand’s presence in AI answers? Get in touch with our LLMO and GEO team.

FAQ: Media & Publishing in AI answers

What does the LLM Industry Benchmark measure?

It captures visibility, SOV, sentiment, concentration, and retrieval patterns for a defined prompt set, market, and time period. Results are aggregated across Gemini, ChatGPT, and Google AI Overviews.

Which sources do AI systems retrieve for media and publishing questions?

In the benchmark, comparison pages lead the retrieved source types for all four intents in media and publishing. Product pages, how-to guides, and listicles also appear for selected intents. These patterns apply only to the examined prompt set and time period.

Which media and publishing brands are most visible in AI answers?

This cannot be answered in general terms. Mentions in AI answers vary with the model, timing, prompt phrasing, and context. The benchmark can only state which entities appeared most often in this dataset during the observation period. Spotify, Apple Music, and WordPress are examples from this snapshot, not a measure of overall market leadership.

Does AI visibility depend solely on one’s own website?

Comparisons, articles, guides, and other third-party formats play a significant role. It is therefore worth checking the specific URLs that appear in answers, so you can distinguish your own content from relevant third-party sources.

How can brands test whether their measures are working?

In three steps: audit your sources and content, check that the information is current and consistent, and then re-measure with the same test questions under the same conditions.