We asked four AI engines to recommend a business 1,120 times. ChatGPT named it least — and named its rivals most.

Finding 1: The four engines split into two tiers — and ChatGPT sits alone at the bottom

Across the three audits with a real brand name (814 completed answers), here is how often each engine actually named the business when asked the questions its own customers ask:

ChatGPT recommended the business in roughly one answer in six — less than half the rate of every other engine. If you've ever checked your AI visibility by asking ChatGPT a few questions and feeling invisible, this is why: ChatGPT is the stingiest of the four, and it's also the one most people check first.

Finding 2: ChatGPT isn't staying quiet — it's recommending someone else

The comfortable explanation would be that ChatGPT just declines to name specific businesses. The data says the opposite. ChatGPT had the lowest “named nobody” rate of any engine (22%) — it answers with real businesses more readily than the others. It simply wasn't naming the one we audited.

In 63% of its answers, ChatGPT recommended a competing business but not the one we audited — far more than any other engine — and it packed the most competitor names into each answer. The biggest engine isn't ignoring your market. It's confidently handing it to your rivals, by name, more than anyone.

Finding 3: The gap held for every business — across wildly different categories

This wasn't one weird vertical. ChatGPT was the lowest-recommending engine for all three real-brand subjects individually — a SaaS product, a local dental practice, and a B2B agency. The absolute numbers swung with how established each business was (one well-known product cleared 44% on ChatGPT; a local practice barely cleared 4%), but the order never changed: ChatGPT last, every time. The pattern is about the engines, not the business.

Finding 4: When AI leaves you out, it leaves a name in your place — 1,198 of them

Across just these four audits, the engines named 1,198 distinct competing businesses. About three of every four answers named at least one specific business — usually around four of them. The engines almost never shrug; they substitute. Every answer that doesn't include you is an answer actively pointing a ready-to-buy customer at a competitor you can name.

Finding 5: The instinct to “optimize for ChatGPT” is backwards

The natural move is to chase ChatGPT because it has the most users. The data says ChatGPT is simultaneously the hardest room to get into and the one already sending the most buyers to your competitors — so a single ChatGPT check is the least representative read of where you actually stand. Your real position is four different numbers, one per engine, and the only way to fix the worst of them is to know which sources each engine is pulling from when it builds its shortlist without you.

Methodology

Four AskedAbout production full audits, run 2026-06-12 and 2026-06-13. Each: 25 buyer-intent questions written from real search and forum phrasing, submitted to ChatGPT (OpenAI's search-grounded API), Claude (Anthropic), Gemini (Google), and Perplexity, three independent times each — 300 scheduled answers per audit, 1,120 completed across all four. Engine rate limits (chiefly OpenAI's search API) dropped some scheduled runs, so per-engine sample sizes differ; every rate is computed over completed answers, not scheduled ones. Each answer was parsed for the subject business and competitor mentions by a judge model. Per-engine recommendation rates use the three audits whose subject is a real brand (814 answers); competitor-density and “named a rival” figures span all four (1,120 answers). We query official APIs, which approximate but do not equal the consumer ChatGPT/Gemini apps — we say so. Subjects anonymized; this is benchmark data, not a callout, and we'll republish as the corpus grows.

See your four numbers

Your AI visibility isn't one score — it's four, and the gap between them is where you're losing customers without seeing it. A free AskedAbout check shows whether the engines name you or your rivals; the full audit reports your mention rate per engine across 300 answers, the competitors taking your slot, and the sources each engine cites about your market.

Companion data: why AI answers change every time you ask — the run-to-run consistency study behind these per-engine gaps. Deciding which tool to use to fix this? See the AI visibility tools compared.