In June 2026, Design Week ran an opinion piece by Mark Nichols of WMH&I with a blunt argument: stop designing brands for machines, and design them for people. He called the problem 'blanding', the quiet slide towards clean, efficient, algorithm-friendly identities that all end up looking like one another.
The piece landed because everyone in branding has watched it happen. Whole categories now share the same rounded sans-serif, the same soft palette, the same carefully neutral tone. The mattress sector is the running joke: Eve, Emma, Casper and Purple are close to interchangeable on a phone screen. But the pattern runs through fintech, automotive, luxury and utilities too.
There is a second force pulling in the same direction, and it is newer. Brands are starting to worry about being legible to AI, because AI is increasingly the thing that decides whether a brand gets recommended at all. The instinct is to sand off the edges so the machine can read you cleanly. This post argues that instinct is backwards. The brands that win AI-era discovery are the distinctive ones, not the safe ones.
Blanding is the flattening of a brand into something clean, generic and inoffensive in the pursuit of scale and system-friendliness. The term describes identities built to work seamlessly across every digital platform and to avoid any feature that might be misread in any market, with the result that they stop signalling anything at all.
It usually comes from good intentions. A brand wants to be consistent everywhere, fast to deploy, safe across cultures and easy to template. Each of those is reasonable on its own. Stacked together, they produce what Nichols calls polite nothingness: a brand that tries to appeal to everyone and ends up meaning nothing to anyone.
The tell is simple. Cover the logo. If you cannot name the brand from anything else on the page, it has blanded.
Brands design for machines because machines increasingly stand between them and their customers, and because generative tools quietly reward sameness. Every touchpoint is now mediated by a feed, a search box or a model, so the temptation is to build for legibility to the system rather than resonance with a person.
Generative design makes this worse. AI image and identity tools are trained on what already exists, so their output gravitates towards the statistically average: the familiar, the proven, the safe. Without strong creative direction, using AI to make a brand is a fast route to looking like everything else that has already been made. As Nichols put it, generative systems risk accelerating a culture of visual sameness that was already growing.
There is also the discovery worry, and it is real. Kantar's 2026 BrandZ analysis found that brands consumers see as meaningful are not only growing faster but becoming more visible inside large language models, the AI systems now shaping what gets suggested when someone asks for a recommendation. A large language model, or LLM, is the technology behind tools like ChatGPT and Claude that answer questions in natural language. Once you know an AI might be the one naming three brands in your category, you start wondering how to be one of the three. The wrong answer is to make yourself blander so the machine can file you neatly.
Blanding costs recognition, pricing power and mental availability, and all three are measurable. This is not an argument about taste. It is an argument about money.
Start with recognition. Research from Ipsos and Jones Knowles Ritchie, analysing over 5,000 brand assets, found only around 15% were 'truly distinctive', meaning they immediately bring the brand to mind on their own. Distinctive brand assets are the non-name elements, such as logos, colours, characters and sounds, that trigger a brand in memory. If 85% of assets barely signal anything, most brands are spending money building recognition they never actually bank.
Then pricing. Kantar's BrandZ work across 40,000 brands found that those seen as meaningfully different command stronger pricing power, in some cases justifying close to double the willingness to pay of undifferentiated rivals. Sameness is not neutral. It is a discount you apply to yourself.
Underneath both sits Professor Byron Sharp's work at the Ehrenberg-Bass Institute on mental availability: brands grow by being easily recognised and recalled at the moment someone is ready to buy, which requires distinctive assets, emotional associations and a recognisable voice. A brand optimised for algorithmic neutrality has quietly given up all three. Burberry learned this the hard way, stripping back to a minimalist wordmark in 2018 and reinstating its heritage font and Equestrian Knight less than five years later once it became clear the flattened version had thrown away what made it Burberry.
AI discovery rewards distinctive, meaningful brands, not safe ones. The same Kantar 2026 BrandZ data that worries marketers into blanding actually points the other way: the brands most visible inside LLMs are the ones with the clearest, most specific associations, not the ones with the smoothest edges.
The skincare example is instructive. Kantar found La Roche-Posay and CeraVe held the highest share of LLM responses in skincare-related queries, at roughly 11.29% and 10.94%. Neither got there by blending in. La Roche-Posay is tightly associated with specific concerns like acne, scarring and sun damage. CeraVe built relevance through internet culture and conversation. In both cases, an AI surfaces them because they mean something precise, and precision is the opposite of blandness.
This makes sense once you think about how a model answers. When someone asks an AI to name a good brand for a particular need, the model reaches for brands with strong, specific, repeated associations to that need. Distinctiveness is not a barrier to machine legibility. Distinctiveness is the thing the machine reads. A brand that has flattened itself into a category average has made itself statistically indistinguishable from its competitors, which is the worst possible position when the machine is choosing just a few names to say out loud.
There is a short-term trap worth naming. A study from NYU and Emory, covered by Design Week in March 2026, found fully AI-generated ads can beat human-made ones on click-through rates by up to 19%. It is tempting to read that as proof the machine should run the brand. But click-through rate measures a single moment of attention, not preference, loyalty or the willingness to choose you again unprompted. Optimising everything for what performs this week is how a brand ends up bland and, eventually, forgettable.
You build it distinctive at the core and adaptive at the edges, rather than flattening the whole thing to fit every surface. This is the premise of Future-Focused Branding (FFB), HRZN's methodology for brands that need to hold their meaning while everything around them changes.
Blanding treats consistency and distinctiveness as opposites, so it trades one away for the other. FFB treats them as separate jobs handled by separate parts of the system. The Brand Sphere holds what is fixed: the distinctive core that must never dissolve into the category average, the assets and associations a machine and a person both use to know it is you. Around it, the four states of the Brand Cycle (Define, Adapt, Evolve and Refine) let the expression move with context, market and channel without touching that core.
Five Brand Principles keep the system honest as it scales: Scalability, Modularity, Adaptability, Evolution-Readiness and Platform-Agnosticism. Platform-Agnosticism is the one that matters most here. It means a brand should carry its distinctiveness onto any surface, including an AI-generated answer, rather than deforming to suit each one. That is the precise opposite of blanding, which lets the platform set the terms and quietly erodes the brand to match.
Technology is not the enemy in this. The same creative code and live systems that let a brand flex by audience, location or moment can be used to express personality more richly, not less. The danger, as Design Week noted, is that most brands will use these tools to optimise for sameness rather than expression. The tools are neutral. The brief is not.
Blanding is what happens when a brand mistakes being legible to the machine for being chosen by it. Those are not the same thing. Machines and people both reach for what is distinctive and what means something specific, which is why the flattened, category-average brand loses on both counts at once.
The brands that will be found in the AI era are the ones that still look like something, sound like something and stand for something, and that hold that core steady while the expression around it adapts. If you are not sure whether your brand has been quietly sanding off its own edges, that is worth an honest look before the next redesign, not after. Getting in touch is a good place to start.
Blanding is the flattening of a brand into a clean, generic, algorithm-friendly version of itself, and it costs recognition, pricing power and memorability all at once.
AI discovery rewards distinctive, meaningful brands rather than safe ones, because a large language model surfaces the names with the clearest, most specific associations, not the smoothest edges.
Only around 15% of brand assets are truly distinctive, so most brands are already closer to the category average than they think, which is the worst place to stand when a machine is naming just a few brands.
