Cracker Barrel dropped its old logo for a plain wordmark this year. The stock fell 10% and the original mark was back within days. Jaguar went further, swapping decades of British heritage for abstract fashion imagery, and watched its sentiment score fall from 23% positive to 8%. Both are textbook cases of a brand losing the room it was speaking to.
What's new in 2026 is that there's a second room to lose. Alongside the customers, journalists and diehard fans reacting to a rebrand, there's a growing audience of AI systems, ChatGPT, Google's AI Overviews, Perplexity, forming their own opinion of who you are in the background. Get the rebrand wrong and you can lose both audiences in the same afternoon.
A rebrand fails with people when it removes the thing they'd actually attached themselves to. Titans fans didn't care about a flaming T logo because it was well drawn, they cared because it referenced Greek mythology their team had built decades of identity around. Strip that out for something cleaner and more abstract, and you haven't modernised the brand, you've evicted the people living in it.
It's a bit like moving house and, in the process of decluttering, throwing out the one chair everyone actually sits in. The new place might photograph better. Nobody wants to be in it.
Jaguar's mistake wasn't boldness for its own sake. It was removing every visual anchor a Jaguar owner had used to recognise the brand at a glance, in one move, with no bridge between the old identity and the new one. Cracker Barrel's was simpler still: the "Old Timer" logo wasn't decoration, it was the entire promise of the brand, nostalgia and homeliness, and a wordmark can't carry that on its own.
A language model doesn't learn who you are from your launch day, it learns from the accumulated weight of everything written about you over time, across sources you don't control: press coverage, directory listings, Wikipedia, forum threads, old reviews. You can execute a flawless rebrand on your own website and still have ChatGPT introduce you using your old name and your old positioning for months afterwards, simply because the rest of the internet hasn't caught up yet.
This is where an incoherent rebrand does double damage. Brands that blur their positioning or drop the visual and verbal threads that once tied everything together aren't just confusing to a customer scrolling past. They're confusing to a model trying to work out what's consistent enough to be worth repeating. Some 2026 tracking has put the effect at a 40 to 60% monthly drop in AI mentions for brands whose rebrands read as inconsistent or unclear. The model isn't being unfair. It's doing what it always does, which is favour the pattern it can find evidence for again and again, and a rebrand that severs its own history gives it less evidence, not more.
Think of it kind of like a rumour spreading through a school. The version that sticks isn't the most accurate one, it's the one that got repeated most consistently by the most people. A rebrand that confuses even your own audience about who you are now is a rebrand that gives the rumour mill nothing solid to repeat.
Not better. Different, and it's worth being honest about where each one earns its keep. AI is genuinely strong at speed and range, turning out a dozen logo directions before lunch or colour and type variants at a scale no studio could match, which makes it a good way to react to and rule out quick concepts. What it struggles with is originality in the sense that matters for a brand, because a model trained on the sum of existing branding tends to hand you back a fluent average of what's already out there, useful for exploring the edges of a problem but less useful for deciding what your brand should actually mean to the people who rely on it.
That decision, what a business is really for and how much can change before it stops being recognisable, is still a human judgement call. It's the kind of nuanced, strategic thinking that comes from sitting with a business, understanding its history and its market, and weighing up what to keep against what to evolve. This is exactly what the Brand Cycle, HRZN's model for how brands should change over time, is built to support: moving through Define, Adapt, Evolve and Refine rather than lurching from one static identity to a completely different one every five years. A brand built as a modular system, with the flexibility to adapt at the edges while keeping its core intact, gives both people and AI models a consistent thread to follow. That's the practical side of Evolution-Readiness, one of the five Brand Principles behind this approach: it keeps a brand legible to an algorithm that only trusts what it's seen repeated, as much as it keeps a brand recognisable to the people who've followed it for years.
The Jaguar and Cracker Barrel stories read like cautionary tales about customers and stock prices, and they are. But they're also early examples of a wider shift, where the audience judging your next rebrand includes systems that never forget and never forgive an inconsistency, because inconsistency is precisely what they're built to notice. The brands that will handle this well aren't the ones avoiding AI or the ones handing branding over to it. They're the ones who let people make the calls that need judgement, and let the system do the exploring, while keeping enough continuity in the brand itself that both audiences can still tell who they're looking at.
A rebrand that strips out the details people had emotionally attached to, such as a symbol, a name or a colour, risks losing the audience even when the new design is objectively cleaner.
AI systems build their understanding of a brand from years of accumulated, consistent signals across sources a company doesn't control, so a rebrand can leave a model describing an outdated version of the business for months.
AI is strongest at generating fast, wide-ranging variants, while people remain better placed to make the nuanced strategic calls about what a brand should mean, which is why the two work best paired rather than one replacing the other.
