We Sell AI. We Still Think Most AI-Written Marketing Is a Mistake.
This might sound strange coming from a consultancy that builds AI systems for a living. But here's our honest view: most businesses using AI to write their marketing are making a mistake, and the data suggests their customers have noticed.
The industry is reading the room wrong
In early 2026, the Interactive Advertising Bureau (IAB) and Sonata Insights published research on AI in advertising (opens in a new tab). They surveyed US consumers aged 16 to 43, and the advertising executives trying to reach them.
- 83% of ad executives now use AI in their creative work, up from 60% in 2024.
- 82% of those executives believe younger consumers feel positive about AI-generated ads.
- Only 45% of those consumers actually do.
- The gap between what the industry thinks and what customers feel grew from 32 points to 37 in two years.
- More than half of consumers want brands to say when an ad is fully AI-generated.
That's a US survey of younger audiences, so it won't map perfectly onto every UK business. But the direction is hard to ignore. The people making marketing are getting more excited about AI while the people receiving it are getting less so.
Some big brands have already picked a side. Digiday reported (opens in a new tab) that Aerie and Dove have publicly rejected AI-generated imagery in their advertising.
"Nobody will know" is no longer true
The other assumption behind cheap AI content is that nobody can tell. That's getting weaker by the month. In August 2026, Anthropic said it would watermark text generated by its Claude models (opens in a new tab), largely to meet the EU AI Act. The mark is invisible to readers, but it means AI involvement can increasingly be detected.
And people were already spotting it without any tools. The same stock phrases, the same three-point structure, the same polished emptiness. Once a reader notices it, they stop trusting the rest of the page.
Where we think AI belongs
We're not anti-AI. We're against using it in the one place where it does the most damage: your voice.
AI is excellent at the work customers never see. Sorting enquiries, drafting replies for someone to check, pulling together reports, chasing invoices, researching a market. That's where it saves real hours, and it's most of what we build.
It's much weaker at the things customers do see and judge you on. Your homepage, your case studies, your about page and your posts are where people decide whether you're worth trusting. Those should sound like you, because that's what customers are buying.
Our rule of thumb: if AI wrote it and nobody who knows the business rewrote it, it isn't marketing. It's filler.
What to do instead
Use AI for first drafts and research if it helps, then have someone who actually knows your customers rewrite it in your own words. And put the real effort into automating the back-office work that eats your week, which is where AI earns its keep.
If you'd like to know where that line sits for your business, our AI opportunity audit gives you a ranked list of the jobs AI should be doing, and is just as clear about the ones it shouldn't. For a quick sense check before talking to anyone, read the questions to ask before paying for custom AI.
Keep reading
Visa Tested 22 AI Shopping Agents. Nearly Half the Runs Failed.
AI agents are starting to shop on people's behalf. Visa's new benchmark shows how often they get it wrong, and what that means for your website.
AI That Works in the Background: October 2026's Big Shift
Anthropic, OpenAI and Meta have all launched AI agents that work on their own and check in with you. Here's what that means for a UK small business.