Why AI content makes brands average, and what escapes it

In 2025, the dictionary publisher Merriam Webster chose slop as its word of the year: the flood of low effort, machine generated content filling feeds, inboxes, and search results. A dictionary naming the phenomenon is a useful marker. The sameness was no longer a designer's complaint. It had become the general public's experience of the internet.

For brands the timing is uncomfortable. The tools that promise infinite content arrived at the exact moment audiences learned to recognise, and discount, the texture of that content. Readers now pattern match for machine tells the way they once pattern matched for stock photography, and they withdraw attention accordingly.

The commercial question is plain: when every business in a category holds the same content machine, what decides who gets remembered? The answer sits outside the machine entirely.

A generative model predicts the most probable continuation of whatever it is given. That is its training objective, and it is superb at it. Ask for a tagline for a wellness brand and it returns the centre of gravity of every wellness tagline it has read. Ask for a logo and the composition drifts toward the category's most common shapes.

This means the tool never makes a brand bad. It makes a brand typical, which is quieter and more damaging. Typical output carries no error a review meeting can catch. It reads as professional, looks finished, and resembles what leadership expected, because it is assembled from what everyone already published.

Understood this way, the sameness epidemic needs no conspiracy. Thousands of teams prompting similar tools with similar briefs converge on the modal answer, and the modal answer is by definition shared. The tool did exactly what was asked. The brief was the problem.

Memory research has held one finding steady for ninety years: items that deviate from their context get remembered, items that resemble their context get absorbed. Psychologists call it the isolation effect, from Hedwig von Restorff's 1933 experiments. A brand that matches its category's texture is choosing the absorbed pile.

This is why averageness costs more than it appears to. The modal post still gets impressions, still fills the calendar, still satisfies the dashboard. What it never does is leave a trace. Buyers scrolling past forty near identical claims retain none of them, and retention is where purchases begin.

Distinctiveness, in the strategic sense, was never about looking unusual for its own sake. It is about owning cues that deviate from category context and repeat until they attribute. The machine age changed none of that arithmetic. It only raised the volume of context to deviate from.

Sameness predates the tools. A decade of design system worship, conversion pattern libraries, and best practice roundups had already taught most categories one look and one voice. The startup aesthetic became a genre. The friendly, lightly witty brand voice became a default. Blanding had a name years before slop did.

What generative tools changed is the cost of producing that sameness: from cheap to free. A weak brand code that once took a junior designer a day to imitate now takes a prompt and a minute. The moat that mediocre consistency provided, simply being finished, evaporated.

That makes this a clarifying moment rather than a catastrophe. Brands with real codes lose nothing when imitation gets cheap, because their assets live in accumulated attribution the imitator has no way to retrieve. Brands whose entire identity was competent genre membership have discovered they owned nothing at all.

There is a payout on the other side of this. As feeds fill with modal content, anything genuinely coded gains contrast. The same held palette, recurring format, or owned phrase that read as mild eccentricity in 2020 now reads as relief, because it deviates from an ocean of statistical centre.

Marketers have started calling the reaction anti AI marketing: deliberately human texture, visible authorship, work signed by a person. The label will pass. The mechanism is durable: scarcity makes signals valuable, and unmistakable authorship is becoming scarce.

For small brands this is the rare shift that favours them. Holding a code steady requires conviction rather than budget. A solo practice with one voice, one palette, and one point of view can now be more recognisable than a funded competitor publishing modal content at volume.

None of this argues for abandoning the tools. It argues for instructing them. A model follows constraints as readily as it follows defaults; the difference is whether the brand has constraints to give. That is what a real voice document and asset system turn out to be in 2026: prompt material.

The working method is direct. Give every tool the vocabulary the brand owns and the words it bans. Provide reference imagery from the brand's own history rather than taste words like modern and clean, which route straight back to the mean. Set the rhythm: sentence lengths, punctuation habits, how a paragraph breathes. Then edit against the codes, since drift returns with every fresh session.

Teams that work this way get the productivity without the dissolution. The tool accelerates execution inside a system a human decided. The failure mode is running the same engine with no system, which produces the category's content faster than the category can forget it.

Every piece of work should pass one question before it ships: could a competitor's tool have produced this? The question is severe, and it needs to be. If the honest answer is yes, the piece adds volume to the category and nothing to the brand.

Passing the filter rarely requires more production effort. It requires a position. A specific claim, a named enemy idea, a worked example from real practice, a sentence only this brand would risk. These are the ingredients the modal answer will never supply, because the modal answer is an average and averages hold no opinions.

This is the honest summary of branding in the generative era. The machines write, draw, and publish. Deciding what a brand believes, which cues it will hold for a decade, and what it refuses to say stays beyond them. That remains the work, and it has quietly become the whole difference.