Which Kinds of Buying Questions Get AI Engines to Name Companies Most Often?

Not all buying questions are the same shape. Someone asking "what is the best project management software for a ten-person team" is doing something different from someone asking "where is the nearest branch of this company," even though both might be fielded by the same AI assistant. This study sorted its 8,188 distinct prompts into four types, commercial-investigation, informational, navigational, and transactional, and tracked how often a company from the 58,803-company cohort got named in response to each type, across 536,301 usable answers collected over a run window from September 4 to September 25, 2026 on ChatGPT, Gemini, and Claude.

The appearance rate, meaning the share of answers in which at least one cohort company was named, came out as follows. Commercial-investigation prompts, the kind where someone is actively comparing options before a purchase, produced 21,920 appearances across 56,085 answers, an appearance rate of 39.1%. Informational prompts produced 15,576 appearances across 40,245 answers, a rate of 38.7%. Navigational prompts, 20,531 appearances across 54,573 answers, a rate of 37.6%. And transactional prompts, by far the largest category by volume at 385,398 answers, produced 147,465 appearances, a rate of 38.3%.

Lined up, the four rates are close: 39.1%, 38.7%, 37.6%, and 38.3%. Commercial-investigation prompts have the highest rate of the four, and navigational prompts the lowest, but the full spread between the top and bottom rate is modest. This is a useful corrective to an intuitive assumption, that an AI assistant would be much more eager to name specific companies when a user is clearly in comparison-shopping mode (commercial-investigation) than when they're asking a general informational question. The data shows that intuition is only weakly true. Engines name companies at broadly similar rates regardless of what kind of question is being asked.

Why might the gap be this narrow? One explanation is that prompt type, as a category, is a weaker predictor of company-naming behavior than the specific content of the prompt. An informational prompt that happens to ask "what are some well-regarded providers of X" is going to pull in company names almost as readily as an explicit commercial-investigation prompt asking for a comparison, because the underlying information need is similar even though the prompt is classified differently. The appearance rate is likely driven more by whether a prompt invites a list-style answer at all than by the formal type label attached to it.

The raw volume differences matter for a different reason: they tell you where most of the measured data actually comes from. Transactional prompts, at 385,398 answers, dwarf the other three categories combined, meaning the overall appearance statistics in this study are weighted heavily toward transactional behavior even though its appearance rate, 38.3%, sits in the middle of the pack rather than at either extreme. A company deciding where to focus its attention should weight prompt types by both their appearance rate and their volume. A prompt type with a slightly lower appearance rate but much higher volume, like transactional here, represents more total opportunities to be named than a smaller category with a marginally higher rate.

What this finding can't tell you is why a specific company would or wouldn't appear for a specific prompt type. The appearance rate is an aggregate across the entire 58,803-company cohort, not a company-level diagnostic. A company that specializes in exactly the kind of query that falls under navigational prompts, the lowest-rate category overall at 37.6%, might still do very well on that category specifically if it's a strong match for the particular navigational questions relevant to its business. The prompt-type breakdown describes the overall field, not any one player's position on it.