Being named in one ChatGPT answer is not the same as owning a topic. When a buyer researches a technology or application in an AI engine, they ask a sequence of questions: a definition, a comparison, which alternatives exist, and how to choose. If your brand appears in one answer and disappears from the next three, you have not won AI visibility. You showed up once.
A recent SEMrush study on ChatGPT topic authority, run with Kevin Indig of Growth Memo, proves the point. The team tracked 1,094 US topic categories in ChatGPT every month from January to June 2026, covering over 50,000 brands, more than 220,000 domains, and over 600,000 citations. This post explains what the study found, what it means for scientific marketing, and how to map and own your niche.
To own a topic in ChatGPT, a brand must be named in at least 4 of the 5 related prompts AND hold at least a 5-percentage-point lead in share of mentions over the runner-up. Winning a single prompt does not count.
The SEMrush study treats each "category" as a cluster of five representative prompts: a definition, a comparison, alternatives, a use case, and a buying question. It scores a brand across the whole cluster, not on one prompt. Share of mentions means how often a brand is named in ChatGPT's written answer, not how often it is cited in the source links. The bar is consistent presence across related prompts plus a clear margin over the next competitor.
Only 15.2% of topics have a clear owner. About 85% are still open. Across the 1,094 categories the study measured:
For scientific brands, the roughly 85% of topics with no clear owner is the opening. Most technical topics in AI search are still there to be taken.
Niche topics are easier to own because fewer brands compete inside them. The SEMrush study split categories by how often people ask ChatGPT about them. The most-asked-about topics had a clear owner only 11.3% of the time. The less-asked-about half had an owner 19% of the time.
For a scientific brand, this is the case for going narrow. A specialised technical topic that maps closely to your buyers is both more ownable and more relevant to the people you want to reach.
No. Strong SEO helps you show up, but it does not decide who owns the topic. Comparing each topic owner against its runner-up on domain-level metrics, the SEMrush study found the owner was ahead only about half the time: 55.7% on branded search volume (the only statistically significant difference, and a modest one), 52.5% on Authority Score, and 48.4% on organic traffic. In nearly half of topics, the runner-up had at least one of these advantages and still lost.
As Kevin Indig put it, traditional SEO metrics are not enough to explain who owns a topic. Deep coverage of the specific topic is what turns a brand from runner-up into owner.
Usually, but only with a wide lead. Clear owners kept their top spot in 90.4% of month-to-month comparisons. Narrow leads changed hands often. In topics where the leader flipped, the median lead was just 1.3 percentage points. Where the leader held, it was 2.9 points. Once a brand builds a real margin across a topic, ChatGPT tends to keep surfacing it. A thin lead is not safe.
For scientific marketing, it means going deep on one topic rather than broad across many. Scientific purchases are specialised and researched from many angles. A prospect evaluating an analytical instrument or a drug delivery system asks ChatGPT a whole sequence of questions before contacting a vendor.
So the goal is not to rank for one isolated question. The goal is to pick one focused topic and build deep coverage across every related prompt in it. The 5-point rule means you are competing for a subject, not a prompt.
Both work. The rule is to go as narrow as your buyers allow, because the narrower the topic and the closer it sits to your buyers, the easier it is to reach a 5-point lead.
Covalent Bonds follows the same rule. For the past several months, all of our content has doubled down on one topic: AI and AEO (Answer Engine Optimisation) for scientific marketing. We did not spread across general marketing terms. We chose one subject, mapped the related prompts, and are working to own the whole cluster. That is how you build authority where it counts for your audience, and it is exactly what we do for the scientific brands we work with.
You map a topic by starting from one narrow subject, building out the five core buyer prompts, and expanding each into the real follow-up questions your buyers ask. ChatGPT judges you across that whole set, so map it before you write. Here is a repeatable method.
Step 1: Pick one topic, and make it narrow. Choose a single technology-level or application-level topic where you have real product fit and expertise. Because niche topics are easier to own and more relevant to your buyers, resist the urge to go broad. "Dry powder inhalation" beats "inhalation." "Raman spectroscopy" beats "spectroscopy."
Step 2: Build the five core prompts. Every topic cluster is anchored by five buyer questions. Write your version of each for your topic:
| Prompt Type | Focus | Example (Dry Powder Inhalation) |
| 1. Definition / Fundamentals | What is the technology/concept? | "What is dry powder inhalation and how does it work?" |
| 2. Use Cases & Applications | Where and how is it applied? | "What are the primary formulation applications for dry powder inhalers?" |
| 3. Comparison | How does it compare to alternatives? | "Dry powder inhalers vs pressurized metered-dose inhalers" |
| 4. Alternatives / Solutions | What options exist in this space? | "What are the top technologies for dry powder inhalation formulation?" |
| 5. Buying / Selection | How do buyers choose a provider/system? | "What key factors should pharma companies consider when selecting a dry powder inhalation partner?" |
Step 3: Expand each of the five into the real questions buyers ask. Each core prompt has follow-ups. Under "comparison," a buyer might ask about efficiency, particle size, patient adherence, or cost. Add every follow-up you can source from your sales calls, your support inbox, and the questions your technical team actually gets. The fuller the map, the more of the cluster you can cover.
Step 4: Check your current share of voice. Run your mapped prompts through ChatGPT (and an AI visibility tool, such as SEMrush's AI Visibility Toolkit or HubSpot's new AEO tool) to see where your brand is named and where competitors are. This gives you an honest baseline: which prompts you already appear in, which you are missing, and how far you are from a 5-point lead over the runner-up.
Step 5: Fill the gaps and hold the lead. Publish content that directly answers the prompts where you are absent or weak, then keep going. Because a 1.3-point lead flips easily, you do not stop once you appear. You keep publishing until your margin is wide enough to hold.
Every tip below ties back to how LLMs surface brands.
1. Pick two or three niche topics, not twenty. Spreading across many subjects dilutes all of them. Choose two or three narrow topics where you have existing content and a real product angle. Because niche topics are more ownable, concentration is what gets you to the 5-point lead.
2. Cover the whole prompt cluster, not the headline question. ChatGPT scores a brand across all the related prompts. Thin overviews will not do it. Publish detailed, technical content that answers every prompt angle in your map, so the model keeps finding you across the topic rather than in one answer.
3. Aim to be named in the answer, not just cited below it. The metric that matters is being mentioned in ChatGPT's written answer. Earlier Growth Memo research found 74% of users chose the top-mentioned brand as their final pick. And in the SEMrush study, only 21% of the most-cited domains in a topic were also the most-mentioned brand. Being in the footnotes is not the same as being in the answer.
4. Earn third-party mentions in the sources ChatGPT draws from. LLMs lean on external consensus. Get your brand named in the places the model pulls from: industry publications, review sites, expert commentary, and technical communities. Pair that with strong entity signals: consistent brand naming, structured data, and third-party citations that tie your brand to the topic.
5. Protect your lead with continuous coverage. A narrow lead flips easily; the median flipped lead was just 1.3 points. Keep publishing and refreshing content on your chosen topic to push past a 5-point margin and lock in long-term visibility. Growth is not the same as safety.
What does it mean to "own a topic" in ChatGPT? Owning a topic means being named in at least 4 of the 5 related prompts in that topic, with at least a 5-percentage-point lead in share of mentions over the next competitor. It is a topic-level standard, not a single-prompt one.
Is it enough to rank for one prompt in AI search? No. Brand visibility in ChatGPT changes prompt by prompt, so appearing in one answer does not mean you own the topic. You need consistent presence across all the related prompts.
What is share of voice in AI search? Share of voice here is share of mentions: how often your brand is named in ChatGPT's written answer across a topic's prompts, compared with competitors. It is measured on answer mentions, not source citations.
Why should scientific brands target niche topics? Niche topics are easier to own because fewer brands compete in them, and they are more relevant to a specialised audience. The SEMrush study found less-asked-about topics had a clear owner 19% of the time, versus 11.3% for the most popular topics.
Does strong SEO guarantee AI search visibility? No. In the SEMrush study, the topic owner beat the runner-up on domain-level SEO metrics only about half the time. Strong SEO helps you show up, but deep, topic-specific coverage is what wins ownership.
How do I start owning a topic in AI search? Pick one narrow topic, map its five core prompts and their follow-ups, check where your brand is named versus competitors, then publish content that fills the gaps until your lead is wide enough to hold.
Source: AI visibility is a topic-level game: A study of 50,000 brands in ChatGPT, SEMrush with Kevin Indig (Growth Memo), July 2026.