Every marketing team now has access to the same writing assistant and the same image generator. Gartner’s 2025 survey of 402 marketing leaders found that only 1% said generative AI investment was not currently a priority. The tools are universal. That has a consequence few teams have thought through.
When everyone uses the same engine, everyone drifts toward the same output.
The sameness problem has been measured
In 2023, Boston Consulting Group ran one of the largest controlled experiments on AI at work, with more than 750 of its own consultants. On creative product innovation tasks, people using GPT-4 performed about 40% better than those without it. Around 90% of participants improved.
Then the researchers looked at the group as a whole. The diversity of ideas among AI users dropped by 41% compared with the group working without it. BCG’s summary was precise: the output was “individually better but collectively repetitive.”
Apply that to a market. Ten competitors each use AI to write their website copy and plan their content. Each one’s work gets more polished. Together they sound like one company with ten logos, and the buyer can no longer tell them apart.
Difference is what actually grows a brand
Polish without difference does not move a market. Kantar analyzed more than 6.5 billion consumer data points from a decade of brand research and shopper data. Brands that are meaningfully different to more people have five times the market penetration of others, and a real advantage in growing it over the following two years.
Kantar also found that when advertising and brand experience build that difference well, it is associated with nine times the volume share, double the average selling price and four times the likelihood of growing share. Differentiation, the report says, is what defends pricing power.
Put BCG and Kantar side by side and the risk is clear. AI makes it easy to be good. It makes it harder to be different. And different is what gets paid.
Where difference comes from now
If generated content is cheap, scarcity moves to whatever a model cannot produce on its own. For most brands that means four sources.
Your own evidence. Client results, internal data, field notes, the patterns you see across hundreds of projects. A model can summarize public research. It cannot know what happened inside your work last quarter.
Real people on camera. Founders and clients speaking in their own voices, with their own faces. Synthetic presenters are getting better, which is exactly why real ones are becoming more valuable.
A stated belief. Most brands avoid saying anything a competitor might disagree with. A clear position, argued with evidence, is instantly distinct because the safe middle is where AI output collects.
Craft. Taste is still a human edge, in editing and in story. AI can produce a rough cut. Deciding what to leave out remains a craft.
The buyer’s view of sameness
Picture the buyer’s side of this. They shortlist four firms and open four websites. Each one promises tailored solutions and a client-first approach, with a confident line about innovation. The copy is clean and grammatical in all four, and all four say nothing in particular.
When everything reads well, nothing reads as different, and buyers fall back on the only differences left: price and whoever answered fastest. That is a race most good firms do not want to run.
The firms that escape it give the buyer something to remember. A specific claim about how they work. A film of a real project. A founder saying something slightly uncomfortable and backing it up. Memorability has become a competitive advantage, and AI has made it rarer.
Use AI for speed, keep the thinking
None of this argues for avoiding the tools. BCG’s own results show real gains, and Gartner found marketing leaders reporting returns from generative AI in time efficiency and cost. The question is where you let it lead.
Let it draft summaries and transcribe interviews. Let it resize assets. Keep the point of view with people, along with the final edit. BCG noted that participants did better on creative tasks when they did not edit the model’s output, which sounds like an argument for trusting it fully. Read it with the 41% figure and it becomes a warning: what is easy for one person is the same for everyone.
One practical rule we use with clients: if a piece of content could have been produced by a competitor with the same prompt, it is not finished.
What to do next
Write down the three beliefs your company holds that a competitor would argue with. If you cannot find any, that is the first problem to solve.
Audit your last ten pieces of content. Mark each one that contains something only your company could supply, such as a client story or an inside fact. Aim for every piece to carry at least one.
Start capturing proprietary material on purpose. Film client conversations with permission, keep a running log of project insights and photograph real days.
Set a house rule for AI: it can draft and assist, but every published piece needs a named human editor who adds something the tool could not.
A limit to keep in mind: the BCG study involved consultants on specific tasks in 2023, and the tools have improved since. The direction of the finding is what matters. Shared tools pull shared output toward the middle, and the middle is where brands disappear.
We believe story beats noise, and noise is now automated. If you want help defining what makes your brand impossible to copy, start a consultation with us.