Should Competitor Names Be Allowed in Prompts for Fair Category Visibility Reporting?

Why do competitor names distort category results?

• The named brand is always mentioned. An answer to "alternatives to [Competitor A]" will mention Competitor A, usually first. Counting that as visibility inflates A's share.
• The retrieval changes. Named prompts retrieve comparison articles, alternatives lists and switching guides. Those pages favour challengers, so brands positioned as alternatives look stronger than they are in neutral questions.
• The choice of names is a lever. Whoever writes the panel decides which competitors to name. Naming a weak competitor, whose alternatives lists always include you, lifts your score.
• Composition changes break trends. Adding more competitor-named prompts one month changes the average without anything changing in the market.

The same problem arises with your own brand name, as covered in whether brand-name prompts inflate AI visibility scores.

What does the distortion look like in numbers?

Take an illustrative category with five brands: you and competitors A to D. On 10 neutral category prompts, the mention rates are: you 30%, A 50%, B 40%, C 20% and D 10%. Share of voice, each brand's mentions divided by all mentions, is 20% for you, 33.3% for A, 26.7% for B, 13.3% for C and 6.7% for D.

Now add 10 prompts of the form "alternatives to A". In those answers, A is mentioned every time as the subject. You are mentioned in 60%, B in 50%, C in 30% and D in 10%. Blend all 20 prompts:

Your mention rate jumps from 30% to 45%, and A's share rises, though nothing about the market has changed. B, which was well ahead of you on neutral prompts, now appears level with you. Only the prompt mix changed. If the "alternatives to A" prompts were added in the second month of tracking, the report would show a large improvement that never happened.

How should a fair panel be designed?

A category panel for the headline. Unbranded questions buyers ask before they know the vendors: "[category] for [use case]", "how to [job the product does]", "best [category] for [segment]". No brand names, including yours. This is the panel behind your share of voice and trend line. How to build one across several products is covered in how to create an AI search prompt taxonomy for a multi-product company.

A competitive panel, reported separately. Questions that name competitors:

• "Alternatives to [competitor]"
• "[Competitor] vs [competitor]"
• "Is [competitor] good for [use case]?"
• "Switching from [competitor] to what?"

Balance it: the same questions for each main competitor, including your own brand as the named one. "Alternatives to [you]" shows who is recommended to your customers when they look around. That is often the most useful result in the panel.

A brand panel, for questions about you specifically: accuracy, reputation and pricing. Report it on its own.

How do you calculate share within competitor-named prompts?

Exclude the named brand from each prompt's count, and calculate share among the others. In the example above, the "alternatives to A" prompts give you 40% of the mentions among the remaining four brands. B has 33.3%, C 20% and D 6.7%. That is a fair measure of who is recommended as an alternative to A. It does not reward A for being named, and it does not leak into the category measure.

Report, for each competitor named:

• Alternative share: your share among the brands recommended as alternatives to that competitor.
• Head-to-head outcome: in "[you] vs [competitor]" answers, which brand the answer favours, or whether it stays neutral.
• Defensive exposure: in "alternatives to [you]" answers, which competitors appear most often.

How visibility moves between comparison, list and category prompts is covered in why your AI visibility changes between versus, listicle and category prompts.

What should a report disclose?

Whether you produce the report or receive it from an agency or tool, it should state:

• The number of prompts in each panel, and the share containing any brand name.
• Which competitors are named, and how they were chosen.
• Whether the named brand is excluded from share calculations in competitor-named prompts.
• Changes to the prompt set since the last report, with their effect on the numbers.

A report that cannot answer these questions cannot be compared from month to month. The same checks help spot selective reporting, as described in how to detect cherry-picked prompts in an AI search case study. Your own messaging can bias a panel in a similar way, as covered in how to find prompt bias caused by your own marketing language.

Are there cases where blending is acceptable?

Only when the blend matches how buyers actually ask, and the weights are fixed and disclosed. If you know from search and sales data that a fixed share of buyer questions name a competitor, a weighted overall index can reflect that. Still report the category and competitive panels separately alongside it, and never change the weights without restating earlier periods.

What about publishing results that name competitors?

Internal tracking can name anyone. If you publish results naming competitors, in a report, a case study or sales material, treat them as comparative claims. Disclose the method, keep the panels separate, and have the claims reviewed before they go out. A share figure that depends on how prompts were chosen is easy to challenge.

What might an illustrative case look like?

Consider an illustrative project management software company whose agency reported a jump in AI visibility from 28% to 41% in one quarter. The prompt set had grown from 30 to 50 prompts. Of the 20 new prompts, 15 were "alternatives to" questions naming one large competitor, where the company appeared often. Recalculated on the original 30 neutral prompts, visibility had moved from 28% to 30%, within run-to-run noise. The company split the panels and added "alternatives to" prompts for each of four competitors and itself. It now excludes the named brand from share in the competitive panel. Its next report showed flat category visibility but strong alternative share against two competitors. That redirected the next quarter's work.

What do people ask about competitor names in prompts?

Should we name competitors at all?

Yes, in a separate competitive panel. Buyers ask those questions, and the answers show where you win and lose.

How many competitors should the panel cover?

The three to five you meet most often in deals, plus your own brand as the named subject.

Does this apply to tool-generated prompt sets?

Yes. Check what share of generated prompts name brands, and separate them before reporting.