Why Does My AI Visibility Change Between Versus, Listicle and Category Prompts?
Your AI visibility changes between versus, listicle and category prompts because each type asks a different question and draws on different sources. A versus prompt, such as "[you] vs [competitor]", names you, so you appear by construction. The real measure there is how you are compared. A listicle-style prompt, such as "best project management tools", is answered largely from ranked lists and review sites, so you appear if you are in those lists. A category or problem prompt, such as "how should a small agency manage client projects?", is answered from the assistant's understanding of the category. It favours brands strongly associated with that need, and often names only a few. Report each type separately and measure the right thing for each. Blending them hides where you actually stand.
How do the prompt types differ?
| Prompt type | Example | What drives the answer | What to measure |
|---|---|---|---|
| Versus | "[You] vs [competitor] for agencies" | Comparison pages, reviews, your own and competitors' pages | Who is recommended, and on which criteria |
| Alternatives | "Alternatives to [competitor]" | Alternatives pages and review-site comparisons | Whether you are listed, and how high |
| Listicle-style | "Best project management tools for agencies" | "Best of" articles, review platform rankings | Appearance rate and position |
| Category | "What software do agencies use to manage projects?" | The assistant's category understanding and major sources | Appearance rate; usually lower |
| Problem | "How can a small agency stop missing client deadlines?" | Advice content; brands may not be named at all | Whether any brand is named, and whether it is you |
SparkToro's research shows the pattern within a single intent. In one test, 142 people wrote different prompts for the same need, and their wording was almost entirely different. Across 994 answers, the top brands still appeared in 55% to 77% of them. The brands with the strongest association with the need appeared regardless of phrasing. Other brands appeared less consistently, but lists vary even between runs of the same prompt. What changed the list sharply was a different need, such as gaming or podcasting headphones.
What does a typical breakdown look like?
Consider an illustrative project management company tracking 20 prompts of each type:
| Prompt type | Appearance | Recommended as first choice |
|---|---|---|
| Versus | 100% (named in the prompt) | 45% |
| Alternatives | 60% | 15% |
| Listicle-style | 38% | 8% |
| Category | 12% | 3% |
| Problem | 5% | 1% |
Averaged together, the company would report "43% visibility". That number is meaningless. It is inflated by versus prompts, where appearance is guaranteed, and it hides a weak position in category and problem prompts, where new buyers start. Brand-name prompts inflate scores in a similar way, as covered in whether brand-name prompts inflate AI visibility scores.
What does it mean when AI recommends our brand only in versus prompts but not best-tool prompts?
It means assistants know you and can compare you when asked, but do not think of you when a buyer asks for the best options without naming you. The usual causes:
- You are missing from the lists that answer "best" prompts: roundups, review platform rankings and comparison articles.
- Your category association is weak. Your name rarely appears near the category terms buyers use, so assistants do not connect you to the category unprompted. How that association builds is covered in how co-citation and co-occurrence affect AI recommendations.
- You are positioned in a niche. Assistants may associate you with a narrower use case than the prompt asks about.
- Versus pages carry you. Comparison pages, often your own, rank for your name next to competitors, but nothing puts you in the broader set.
The fixes follow the causes: get into the lists that are cited, make your category phrase consistent across your site and profiles, list your product in the right category on review platforms, ask every customer for an honest review in their own words, and earn coverage that names you as an example of the category. If you really serve a niche, own it. Aim to appear in the niche's "best" prompts rather than the broad category's.
Which sources matter for each prompt type?
Because each type draws on different sources, the work to improve each is different:
- Versus prompts: your own honest comparison pages, review platform comparisons and third-party "A vs B" articles. The assistant judges on the criteria those pages discuss, so make sure your strengths are stated with evidence.
- Alternatives prompts: alternatives pages on review sites and publisher articles listing substitutes for the market leader. Being listed there matters more than anything on your own site.
- Listicle-style prompts: the "best of" articles and review platform rankings that assistants cite most for your category. Check which ones they are, then work on inclusion.
- Category prompts: pages that define the category, such as reference entries, analyst and industry reports and widely read guides. Being named there as an example of the category builds the association.
- Problem prompts: practical guides that answer the problem. If your own guides are the best answer and mention your product where relevant, you have a chance to be the brand named.
How should each type be measured and reported?
- Report prompt types separately, each with its own appearance and recommendation rates.
- For versus prompts, measure the win rate and the criteria given, such as price, ease of use or integrations.
- For listicle and category prompts, measure appearance, position and list length. Longer lists make appearance easier, so compare with competitors on the same prompts.
- For problem prompts, measure whether any brand is named, then whether you are.
- Keep the mix of types fixed from month to month, or changes in the mix will move your totals.
Whether to include competitor names in tracked prompts at all is covered in whether competitor names should be allowed in prompts for fair category visibility reporting.
Which type matters most for growth?
Usually listicle and category prompts, because that is where buyers who do not yet know you meet their options. Versus prompts matter later, when a buyer is already comparing you with someone. A healthy profile shows reasonable appearance in "best" prompts and strong win rates in versus prompts. A profile that is strong only in versus prompts depends on buyers already knowing you exist.
What might an illustrative case look like?
Consider an illustrative HR software company that celebrated winning 60% of "[us] vs [competitor]" answers. Splitting its prompts by type showed it appeared in only 9% of "best HR software for small businesses" answers. Those answers cited three roundup articles and one review platform category where the company was not listed. Its homepage described it as an "employee experience platform", a phrase buyers rarely used. The company changed its category phrase to "HR software for small businesses", completed its review platform profile in that category, and pitched two of the roundup publishers. Six months later, it appeared in a quarter of the "best" answers.
What do people ask about prompt types and AI visibility?
Should versus prompts count towards visibility at all?
Count them separately, as win rate. Mixing them into appearance rates inflates the headline.
Why do category prompts name so few brands?
Because the assistant is explaining the category rather than listing options, and names only the brands it most strongly associates with it.
How many prompts of each type do I need?
Enough per type to report on its own: at least 20 to 30 distinct prompts, each run several times. Repeated runs of one prompt are correlated, so precision depends mainly on the number of distinct prompts. Even at 100 answers, a share near 40% has a margin of about ±10 points, and the change between two months needs to be about 14 points or more before it is more than noise. Confirm smaller moves with a rerun before reporting them.
Do different assistants treat prompt types differently?
Yes. List lengths, how often they search and which sources they favour differ by assistant, so break results down by assistant as well as by prompt type.
Published by The Kovmere Team · updated Sep 29, 2026 · More resources