Earning press coverage does not guarantee that a brand is named when artificial intelligence summarizes the story. An October 2026 Connective3 study of more than 3,500 digital PR links found that 43% of brand mentions were not retained in AI summaries, and that brands given a prominent role in an article were far more likely to be kept.
Updated: November 3, 2026
November 3, 2026. Public relations agencies spend enormous amounts of time securing media coverage for their clients. New research asks a timely question: what happens when artificial intelligence summarizes that coverage and leaves the client out? The answer points to a clear opportunity for communicators who want their earned media to work harder.
Table of contents
What Are the Key Takeaways?
- Connective3 found that 43% of brand mentions were not retained when earned-media articles were summarized by AI.
- The study covered more than 3,500 digital PR links, 170 brands, 224 campaigns and over 2,300 publisher domains.
- Articles with the strongest brand signals were about 4.75 times as likely to keep the brand in an AI summary as those with the weakest signals.
- Even the highest-scoring articles lost the brand about 30% of the time, so prominence helps but does not guarantee retention.
- The research tested summarization, not AI search recommendations, and it was conducted by the company that built the scoring system.
What Is the 43% Problem?
An October 2026 study by digital marketing agency Connective3 found that 43% of brand mentions were not retained when earned-media articles were summarized by AI. The research examined more than 3,500 digital PR links involving 170 brands, 224 campaigns and over 2,300 publisher domains.
The finding highlights a growing challenge for public relations professionals. A brand can secure coverage in a publication, receive a backlink and appear in a media monitoring report, yet still be absent from an AI-generated summary of the article. As AI-powered search becomes another way audiences discover companies, products and services, that distinction could become increasingly important.
Traditional public relations measurement often emphasizes media placements, publication authority, estimated audience reach, backlinks and share of voice. These metrics remain useful. They do not, however, show whether AI systems retain a brand's identity when processing published coverage.
Consider a technology company mentioned briefly near the end of an industry article. The company receives coverage and may earn a valuable backlink. But when an AI system summarizes the article, the company's name may disappear because it is peripheral to the central story. By contrast, an article built around a company's original research, featuring its executives and explaining its findings, may give the brand a far more prominent role. Two placements that both count as wins in a coverage report can produce very different outcomes in AI-generated content.
How Does Brand Prominence Change AI Retention?
Connective3 developed a proprietary measurement system called Brand Citation Score, or BCS, to evaluate how prominently brands appear within earned-media content. The system assigns articles scores from 0 to 100 using factors including brand prominence, contextual relevance, entity salience and narrative framing.
The study found substantial differences in AI retention rates across scoring categories.
| Brand Citation Score | Brand retention in AI summaries |
|---|
| 0-25 | 14.72% |
| 25-50 | 29.46% |
| 50-75 | 55.20% |
| 75-100 | 69.95% |
Articles with the strongest brand signals were approximately 4.75 times as likely to retain the brand in AI summaries as those with the weakest signals. That is an encouraging result for communicators, because prominence within an article is something a thoughtful pitch and a strong story angle can influence.
Even among the highest-scoring articles, however, brand retention was not guaranteed. Approximately 30% of brand mentions in the top category still failed to appear in AI-generated summaries. The findings suggest that earning coverage and earning prominence within coverage are different objectives.
Are PR Agencies Measuring the Wrong Things?
For decades, agencies have demonstrated results through media coverage reports. A successful campaign might generate dozens of articles, backlinks from established publications and millions of estimated media impressions. Those measurements do not necessarily reveal whether a brand will be named when an AI system processes the coverage.
The emerging issue is not whether media coverage still matters. It does. The question is whether conventional reporting captures the full value of that coverage in an environment where audiences increasingly encounter information through AI-generated answers.
A placement that mentions a brand once may have a different effect from an article in which the brand's research, expertise or executives are central to the reporting. For agencies, this raises a practical question: should media coverage reports begin distinguishing between simple mentions and substantive brand visibility? The answer may depend on how clients use AI search and whether their communications programs are designed to influence brand discovery, reputation or purchasing decisions.
Will Original Research Become More Valuable?
One implication of the study is the potential importance of original research and expert commentary. When a company's proprietary data forms the basis of a news story, its identity may be more difficult to separate from the article's central message. Similarly, an executive who provides substantial analysis may create a stronger association between the brand and the subject being discussed.
Connective3 identifies several factors associated with stronger brand signals:
- Brand prominence: whether the brand appears in the headline, opening paragraphs or central discussion.
- Expert attribution: whether executives or specialists contribute meaningful commentary.
- Contextual relevance: whether the article connects the brand to subjects relevant to its business.
- Narrative importance: whether the story substantially depends on the company's research, expertise or actions.
These factors are not entirely new to public relations. Strong attribution, substantive reporting and credible expert commentary have long been valuable. What is changing is the ability to examine how those characteristics relate to AI-generated outputs.
What Does This Mean for PR Agency Clients?
The findings could influence how companies evaluate public relations performance. A campaign generating 100 media placements may look successful in a conventional report. If many of those placements contain only passing brand references, however, AI-generated summaries may omit the company from a substantial share of them.
That does not mean those placements have no value. They may still reach readers, generate referral traffic, improve awareness or support traditional search visibility. It means that AI retention is a separate outcome worth measuring. For brands investing heavily in earned media, the distinction may become commercially significant, and clients who ask for it early will be better positioned to direct their programs toward stories where the brand plays a leading role.
What Five Questions Should PR Agencies Ask?
As AI-generated answers become part of the communications environment, agencies may need to expand their measurement practices. These five questions are a practical starting point.
- Is the client central to the coverage? A passing mention and a story built around a company's research should not automatically be treated as equivalent outcomes.
- Does the coverage connect the brand to relevant subjects? A company's presence in an article matters, but so does the context in which it appears.
- Are executives and experts properly attributed? Substantive commentary can strengthen the relationship between a brand and its areas of expertise.
- Does the brand survive AI summarization? Agencies can test representative earned-media placements to determine whether AI-generated summaries retain the client's name.
- Does coverage translate into broader AI search visibility? Article summarization and appearance in AI search answers are not identical, so both require separate measurement.
Are Summaries the Same as Search Results?
The Connective3 findings require an important qualification. The study tested whether brands were retained when AI systems summarized earned-media articles. That is not the same as measuring whether ChatGPT, Gemini or Google AI Overviews recommend those brands in response to general user questions.
A brand omitted from one article summary might still appear in other AI-generated answers. Likewise, a brand retained in a summary is not necessarily recommended by an AI engine. The research demonstrates a relationship between brand prominence and retention in AI summarization. It does not establish that a particular media placement causes greater brand visibility across AI search.
Connective3 also developed the proprietary scoring system used in the analysis. The published findings should therefore be understood as company-conducted research rather than an independently audited industry benchmark.
What Is the Next Challenge for PR Measurement?
The study raises a broader question about the future of earned media. If a brand earns coverage but disappears when AI summarizes that coverage, how should the placement be valued? Traditional media measurement remains important. Articles can influence audiences directly, build credibility and generate business outcomes independent of AI systems.
But as AI-generated summaries and search answers become more prominent, PR professionals may need additional ways to evaluate how their clients are represented. That could include tracking brand retention, contextual prominence, AI citations and whether earned-media coverage contributes to relevant AI-generated answers.
For public relations agencies, the next competitive advantage may involve demonstrating not only where clients receive coverage, but how effectively that coverage communicates who they are and what they do. The Connective3 findings provide an early indication of why that distinction deserves attention.
