For a locksmith, AI cites availability. For an accountant, never

Short answer: when ChatGPT or Gemini recommend a business, they cite its availability (24/7, same-day, emergency call-outs, quick response times) in 2.6% of passages: 410 out of 16,022 across four countries. But the figure is the trade, not the average. For British locksmiths it is 35.6%, for American plumbers 23.0%. For American hotels, accountants, lawyers and estate agents, it is zero. And unlike most of what we have measured, the two engines barely differ: 2.5% for ChatGPT, 2.7% for Gemini.

When the door is locked, the reason to call is speed. Nobody needs an accountant at 2 a.m.

What was measured

Our four studies, re-read on October 4, 2026: the same buyer questions ("a good locksmith in Manchester", "the best accountant in Denver") put to ChatGPT (gpt-4o-mini) and Gemini through their APIs, without web search, in the United States (collected September 13, 2026), the United Kingdom (July 31), France (July 30) and Brazil (August 6). 3,600 answers, cut into passages; we kept the 16,022 that name a business, with the same cleaning as our recent editions. In each, we looked for availability given as a reason, in English, French and Portuguese: "24/7", "same-day", "emergency service", "open on Sundays", "quick response times", "responds within an hour", "dépannage rapide", "service d'urgence", "atendimento 24 horas". The detector was tested on 14 phrasings it must catch and 6 it must not ("in business for 24 years", "a 24-room hotel"). On reading, the bare words "urgence" and "en urgence" picked up French advice and introductions ("finding a plumber is hard, especially in an emergency"), so only phrasings where the business itself acts were kept.

Locksmiths, plumbers, heating: the emergency trades

Every one of the twelve trades where availability comes up most is a trade people call when something breaks. British locksmiths 35.6%, French 28.1%, American 25.3%; American plumbers 23.0%, French 20.3%; American heating and cooling contractors 16.1%. Electricians follow, around 12%. At the other end, 0 of 296 passages about American hotels, 0 of 245 about accountants, 0 of 220 about lawyers and 0 of 155 about estate agents mention hours or speed. The assistant recommends an emergency trade for being reachable, and every other trade for something else: what it does, who it suits, what it has done before.

The one reason both engines share

In our recent editions, Gemini gave reasons far more often than ChatGPT: who a business is for (48% against 5%), how long it has existed (11% against 3%). Availability is the exception. ChatGPT cites it in 2.5% of its passages, Gemini in 2.7%; in the United States ChatGPT is even ahead, 4.6% against 3.8%. In an emergency trade, both engines reach for the same argument, often in the same words: "known for their quick response times", "offers 24-hour emergency service". Where they differ is in who gets named at all (Gemini tells customers who you are for, ChatGPT almost never does).

What it means if you are the business

If you are a locksmith, a plumber, a heating engineer or an electrician, availability is one of the few reasons the assistant gives for naming you, in a third of locksmith passages in the United Kingdom. It can only repeat what is written: your hours, whether you take emergency calls, at night, at weekends, and how fast you usually arrive, stated plainly and the same way on your site, your map listing and the directories that list you. A claim the assistant cannot read, it cannot repeat. If you are a hotel, an accountant or a lawyer, hours are not what the answer is made of: the assistant described those businesses by what they do and who they suit (AI cites how long you have existed in 6% of its recommendations). Nothing here guarantees being named; it tells you which fact the answer uses in your trade.

What this does not prove

• These are the models' answers through their APIs, without web search: the way our audit asks. Consumer apps with search switched on may show live opening hours from map listings, which this measurement does not see.
• Four studies collected between July 30 and September 13, 2026, one pass per question. The models have moved since, and nothing here measures that drift.
• A passage can name several businesses; a mention is then counted once for the passage.
• A vocabulary misses reasons phrased otherwise ("they came within the hour" in a form we did not list). The trade ranking is clear; the exact shares are a floor more than a precise figure.
• Some trade figures rest on fewer than 100 passages (French locksmiths: 64; French plumbers: 74): read them as an order of magnitude.