The assistants knew about a launch before the first store opened
Before a single store opened, GapWatch measured what four AI assistants told people about a restaurant chain re-entering a national market in July 2026. All thirty-five answers correctly stated it was not yet operating. Ninety-one per cent volunteered the return unprompted.
Two findings, pointing opposite ways
The first is reassuring, and slightly surprising: the assistants were factually current on a market entry that had not happened. No answer claimed the chain was already open. Nearly all of them knew it was coming and said so without being asked.
The second is the one that matters commercially. Knowing a brand exists and is arriving is not the same as recommending it. Asked the questions a customer actually asks — the best coffee chain in the country, where to get coffee and donuts in each of two major cities — the brand was named zero times out of thirty-five, on every assistant and every run.
Why the timing was the point
This measurement is only interesting because of when it was taken. Once stores open, advertising runs and coverage accumulates, the picture changes — and there is no way to reconstruct what it looked like beforehand. The before-picture exists only if somebody freezes it.
That is true of every launch, every rebrand, every acquisition and every campaign. The measurement is cheap and takes a day. The window closes permanently.
What we would tell anyone about to launch
Freeze the picture first. Not because the number will be flattering — this one was not — but because in six months the only question anyone will ask is whether it moved, and that question is unanswerable without a starting line.
Measured 22 July 2026 across ChatGPT, Gemini, Claude and Perplexity, three runs per question. Brand, category and market not named. Read the methodology →