Ungrounded vs. grounded: why a global QSR brand lost to a regional challenger in AI search
An AI-visibility audit across three QSR brands in the same category. Ungrounded, the global brand won on decades of training recall. Grounded, the regional chain dominated live search and the global brand was nearly invisible. Training recall and live discoverability are two separate competitions.
A global QSR brand with decades of recognition. Nearly invisible when AI search actually goes looking. A regional challenger capturing that demand instead. Same category, same customers, different outcome depending on which layer is doing the deciding.
I ran an AI-visibility audit across three QSR brands competing in the same category: a hyperlocal independent, a fast-growing regional chain and a global brand with decades of recognition. The global brand was expected to win. It didn't — and the reason has direct implications for how brand investment gets measured and defended.
I tested two conditions. Ungrounded: the model answering from training data alone, no live search. Grounded: the model searching and citing live sources, the way AI search increasingly works in practice.
Ungrounded, the global brand won convincingly. Decades of brand-building are baked into the model's training data. That's brand equity performing exactly as intended.
Grounded, the result inverted. The regional chain dominated live search for this category while the global brand's presence was close to zero. Not because the model forgot who they are, but because grounded answers draw on delivery platform listings, Google Business Profile entries, reviews and local press — and the regional chain has a stronger, more current footprint across those sources. The global brand's equity was built around different category language to the one this audit tested.
The hyperlocal independent scored low on both conditions. Too small a footprint to be memorised or grounded. That's the baseline risk of no digital presence at all.
The finding that matters at leadership level: training recall and live discoverability are two separate competitions, and winning one buys no ground in the other. Training recall is built over years through brand investment. Live discoverability is contested continuously, through footprint in the sources AI models cite today. A brand can lead one and be nearly invisible in the other, in the same category, at the same time. Most brand tracking doesn't currently distinguish between them.
This isn't a verdict on the brand. It's a signal that existing brand equity doesn't automatically transfer to the language customers use to search — and that gap needs to be measured directly. Nor are the two mutually exclusive; a brand with a strong platform and citation hygiene could plausibly lead both at once.
- —Measure ungrounded and grounded AI visibility as two separate scores, never one blended number.
- —Audit the live sources models cite in your category — delivery platforms, GBP, reviews, local press — not just your owned channels.
- —Defend training recall through brand investment; win live discoverability through citation and footprint hygiene in the category language customers actually use.