Cision's 564 Million-Post Study Finds Brands Losing AI Answer Visibility

Cision analyzed 564.61 million social media posts and 4,678 search queries and AI prompts, and its central finding is uncomfortable for communications teams: the brands AI answers recommend are often not the sources those answers cite.

The report, The State of Search and Social 2027, was announced on November 9. It spans health and wellness, food and drink, automotive, artificial intelligence and online culture, with its principal analysis covering August 2025 through August 2026 using Brandwatch Consumer Research and Trajaan data.

Where do owned websites lose visibility?

The automotive example carries the sharpest numbers. Toyota held a 13.8% share of brand mentions in a sample of car-related AI answers. Yet Toyota's own website represented only a fraction of the cited material, while YouTube supplied more than 4,000 cited sources.

The gap matters more than any ranking. A brand can appear frequently in an answer without supplying, or controlling, the sources behind it. Those sources include reviewers, journalists and creators whose incentives and editorial standards differ from a corporate press office.

A caveat follows the promise. This is a measurement lead, not proof that a press campaign caused an AI system to cite a particular story. One category's source mix also says nothing about how AI answers behave in healthcare, finance or B2B software.

Do search demand and public sentiment move together?

No, according to Cision, and that is the report's second working premise: a spike in searches, a surge in social conversation and an AI-generated summary can tell three different stories about one topic.

Take AI itself. Rising searches coexist with social discussion about data-center impacts and a backlash against low-quality AI content. High curiosity is not automatically favorable sentiment. In another example, social discussion of magnesium glycinate emphasized stress relief, while sampled AI answers gave greater weight to other health-related claims.

These are different data surfaces, not interchangeable indicators. Search queries are not a representative survey of buyers. Social volume counts posting activity, not unique people. An AI answer is the output of one system and one set of prompts.

What should a PR measurement brief now include?

The immediate change is operational. An earned-media report that ends with clips, impressions and sentiment scores will miss whether the same themes appear in consumer questions and in the third-party sources AI systems reuse. Equally, a dashboard showing rising AI mentions is not proof of improved reputation.

A practical audit, as Cision frames it, runs in steps:

• Start with a fixed list of category questions relevant to buyers.
• Record which brands appear, which domains are cited, and whether sources are owned, journalistic, creator-led or community-produced.
• Repeat the same prompts across defined platforms and dates, and store the answers rather than relying on an isolated screenshot.
• Compare findings with existing coverage, audience questions and social themes, looking for disagreements: a feature in coverage but absent from answers, a safety concern visible socially but missing from the press narrative, or a creator review surfacing where the company site does not.

Cision chief marketing officer Amy Jones said some of the most useful insights emerge in "the space between channels." That is a fair description of the reporting opportunity, provided teams retain the underlying source lists, time periods and sampling assumptions behind each comparison.

What are the study's limits?

They are significant. The social-post count and the prompt count describe two different evidence pools; they should not be added together, converted into a number of consumers, or read as a representative global panel.

The official report page sits behind a registration form. The public announcement offers selected findings but permits no independent check of every chart, category sample, platform mix or query methodology. Cision also sells media monitoring, search intelligence and AI visibility products — a commercial interest readers should weigh when evaluating its conclusions.

The sensible reading is narrower than an industry-wide prediction. The study gives PR teams a reason to audit the outside sources shaping AI answers about their brands and to reconcile those with search and social research. Whether changing a communications program subsequently changes AI recommendations is a separate question, and one that needs repeated observation and stronger evidence than a vendor report provides.