Does Claude name more businesses per answer than ChatGPT?

There is a widespread impression among people who watch AI answers closely that Claude tends to list more businesses per answer than ChatGPT does. This study, which classified 926,792 individual answer units drawn from 193 query groups over the window 2026-04-27 to 2026-09-18, finds the opposite.

Every answer unit in the study was sorted into exactly one of four classes based on how many distinct businesses it named: names none, names 1 to 5, names 6 to 9, or names 10 or more. No rows were excluded from this classification (0 excluded), so the comparison below covers the full set of answers collected for each engine.

Looking specifically at the top class, names 10 or more, ChatGPT placed 49.7% of its 329,786 answers there. Claude placed 43.4% of its 298,968 answers in that same class. That is a real gap in the same direction the popular impression gets backwards: ChatGPT, not Claude, is the engine more likely to produce a long list of ten or more named businesses in a single answer.

The gap does not close if you look at the other end of the distribution either. Claude concentrated more of its answers in the names 1 to 5 class, at 5.8%, compared with ChatGPT's 1.6% in that same class. So Claude answers cluster toward shorter lists and ChatGPT answers cluster toward longer ones, consistently across both ends of the scale.

Why might the opposite impression have taken hold? One plausible mechanism is that people remember the answers that stood out, and a Claude answer that runs long with a conversational list of options may simply be more memorable or more often screenshotted than a ChatGPT answer that produces a similarly long list formatted as a numbered directory. Memorability and frequency are different things, and this study measures frequency across a large, mechanically classified set of answers rather than relying on anyone's recollection of a handful of examples. A second plausible mechanism is that Claude's longer individual list items, when it does name businesses, may read as more expansive even when the count of distinct businesses is lower. This study only counts distinct named businesses, so it cannot speak to how elaborate the surrounding prose is for each one.

What would show this finding to be wrong, or at least incomplete? If the query groups sampled here happen to skew toward question types where ChatGPT is unusually generous with list length, and Claude is fed a different mix of prompts elsewhere, the gap could shrink or reverse for a different sampling window. That is a genuine limitation: this study covers one window, 2026-04-27 to 2026-09-18, and one set of 193 query groups, not the full space of everything either engine might be asked. A reader who wants to argue with this finding should ask what query mix they have in mind when they say Claude names more businesses, and whether that mix looks like the one measured here.

What should a company do differently on Monday because of this? If you are tracking how often your business shows up in AI answers, and you are using list length as a proxy for how much competition you face in a given engine's answer, this finding says to weight that proxy by engine. A ChatGPT answer is measurably more likely to be a long list; a Claude answer is measurably more likely to be a short one. Treating the two engines as equivalent in this respect will misstate how crowded the field looks in each.