Topical authority in AI search: Semrush data

Some brands can afford broad topical coverage. Forbes, for example, ranks for topics ranging from bitcoin prices to neck pillows. For most companies, however, visibility grows when the content strategy clearly focuses on core topics. Topical authority is one of the most used concepts in SEO and explains why teams build clusters instead of isolated keyword pages. Swinging too far from the base has even led to algorithmic demotions in the past, for example at HubSpot and ClickUp. For a long time it remained open whether this principle also applies in AI search – or whether large language models open a new playing field where brands can swing thematically wide.

An analysis based on the Semrush AI Visibility Toolkit provides solid evidence. More than 50,000 brands, over 220,000 domains, and around 600,000 citations were evaluated. The focus was on 1,094 US categories with five representative prompts each and monthly snapshots from January through June 2026 in ChatGPT (USA only). The central finding: topical authority still matters in AI search – especially through durability. Once a brand earns an outsized share of brand mentions in a category, it usually keeps that position.

Method, definitions, and study limitations

The study measures which brands ChatGPT mentions most often across five prompts per category. It does not capture sentiment, recommendation quality, user trust, or purchase impact. It also lacks organic rankings, classic SEO visibility, and share of voice at topic level. Semrush only records whether a brand appears in an answer – positive, neutral, or negative.

• Owner: brand with the highest share of mentions, named in at least four of five prompts and at least five percentage points ahead of the runner-up.
• Emerging leader: most-mentioned brand in at least three prompts without meeting the owner threshold.
• Unsettled category: no leading brand in at least three prompts.

Among cited pages, product and service landing pages dominate, followed by editorial content. Homepages were cited in only about four percent of cases. Almost half of the URLs were hard to classify.

Most categories have a leader, but rarely an owner

In June 2026, only 15.2 percent of analyzed categories had a clear owner. 53.7 percent were open fields with several contenders. More than half of the 1,000-plus categories therefore have a leader that a challenger can dethrone. Compared with classic Google search, the field looks clearly more open.

Especially relevant: the highest-volume topics least often have an owner. The top half of categories ranked by AI search volume holds around 98 percent of estimated demand but shows an owner rate of only 11.3 percent – versus 19.0 percent in the bottom half. In total, 89.3 percent of estimated AI search demand sits in categories with no clear owner. Demand concentrates, ownership does not.

Brand mentions beat citations

An owner here is the brand with the largest share of mentions in the tested prompts – not the one with the most citations. Earlier user studies show that users pay little attention to citations and weigh brand mentions heavily. In shortlists, about 75 percent choose the top brand. In AI Mode, users rarely build their own comparison lists: only eight of 147 codeable tasks produced a self-built shortlist. 64 percent clicked nothing during the task. 74 percent chose the first-named result as their favorite.

Interesting finding: the relationship between citations and brand mentions is slightly negative (−0.229). The most cited domain was also the most mentioned brand in only 20.8 percent of cases. Conversely, the most mentioned brand was cited at least once in 69.9 percent of cases. Citations shape answers; brand mentions decide what users see and select.

Traits of category owners and chances to switch

Compared with the runner-up, owners more often show higher brand search volume (55.7 percent), higher organic traffic (48.4 percent), and higher authority scores (52.5 percent). These are clues, not a causal formula. Brand strength can ease entry into the answer set; ownership depends more on topic-specific relevance, content quality, and prompt fit.

Clear owners retained first place in 90.4 percent of month-over-month comparisons. Switches concentrated on narrow leads: the median lead of switching leaders was 1.3 points, versus 2.9 points for retained leads. Narrow leads are contestable; wide, sustained ownership is hard to displace.

How teams compete for category ownership

Topical ownership should guide where teams invest – not automatically justify more content. A practical process looks like this:

• Watchlist: define 10 to 20 commercially important categories and track at least five prompts each (definition, comparison, alternatives, use case, buying decision).
• Classify: separate owner, emerging leader, or unsettled category.
• Invest: prioritize valuable categories with no owner or a narrow lead; defend your own leads; attack established owner categories selectively.
• Close gaps: address prompt gaps with clearer positioning, comparison pages, documentation, expert guidance, and third-party sources.
• Measure: brand mentions as the ownership metric, citations as source authority, pipeline and revenue as the business metric.

Topic focus pays off because ownership compounds over time. Brands that spread content across many adjacent topics rarely reach the mention share that makes a lead stick. According to Semrush data, the window is still open – and it will shrink. Choosing the wrong five categories today will cost more later. Brands should pick the categories where they want to be the answer, protect existing leads, and skip fields where an established brand already appears in every prompt.