I’m going to say something that might seem odd coming from someone who uses AI tools regularly: you should not use AI to write your marketing copy.
Not your website copy. Not your service page headlines. Not your about page. Not your email campaigns. Not your social posts. None of it.
And before you close this tab, hear me out – because this isn’t a Luddite rant. AI is genuinely useful for plenty of things. Writing the words that represent your business to the world is not one of them.
The copy problem nobody wants to talk about.
Here is what I see constantly. A client will spend weeks – sometimes months – going back and forth on the design of their website. Tweaking button colours, debating font weights, and requesting a third round of amends on the hero section. Then, when it comes to the copy, they knock it out in ten minutes on ChatGPT.
That copy then goes live and represents their business to every single person who lands on the site.
Think about that. Hours of design obsession, ten minutes of copy. And copy is the thing that actually converts visitors into customers.
I’ve been working in branding and web for over 25 years. I’ve built sites for clients including Broadbean and The Victorian Society. The thing that makes the difference between a site that generates business and one that just sits there looking decent is always the copy. Always.
AI does not know your business.
Good marketing copy comes from a deep understanding of three things: what you do, who you do it for, and why someone should choose you over the alternative.
AI does not know any of that. It knows what you type into a prompt box.
If you type “write a homepage for a solicitors firm in Manchester”, you will get homepage copy for a generic solicitors firm in Manchester. It will be grammatically fine. It will hit the expected beats. It will sound like every other solicitor’s firm in Manchester.
That is the precise opposite of what marketing copy is supposed to do.
Your copy needs to reflect your positioning, your values, your specific client base, and your genuine point of difference. AI cannot invent that for you, and feeding it a rough brief and hoping it figures it out is wishful thinking.
AI writes to the average.
This is probably the biggest issue and the one people think about least.
Large language models are trained on vast amounts of existing text. When you ask one to write marketing copy, it produces something that statistically resembles good marketing copy based on everything it has ever processed. That means it tends toward the middle. Competent, inoffensive, and utterly forgettable.
Marketing copy that converts is not average. It has a point of view. It challenges something. It speaks to a specific person with a specific problem in a way that makes them feel understood. That takes genuine thinking about your customer, not pattern matching against millions of web pages.
There is a reason experienced copywriters charge what they charge. They are not just arranging words – they are doing the strategic thinking that turns a brief into something that actually works.
What better copy?
We get that AI is really tempting for writing marketing copy, but if you use generic prompts, you’ll get generic copy. We’ve developed different ways of working with AI to generate first-draft content that sounds like you – it’s a time-saver.
Google already knows.
Some people use AI copy with the idea that it will help their SEO. It will not. Google’s Helpful Content Guidelines are explicit on this point.
Content created primarily to rank rather than to help people is not what Google’s systems are built to reward. The guidelines ask a direct question: is this content primarily made to attract search engine visits, or is it genuinely useful to the person reading it?
AI-generated copy that has been produced quickly from a shallow prompt almost always fails that test. It tends to be surface-level, repetitive, and lacking any first-hand experience or genuine expertise – the things Google’s E-E-A-T framework specifically looks for.
The “Experience” element of E-E-A-T is particularly important here. Google wants to see evidence that the person writing the content has actually done the thing they are writing about. AI has not done anything. It has processed text about doing things. That is a significant difference, and Google is getting better at detecting it.
What happens to your brand?
Your brand is not just your logo and your colour palette. It is every interaction someone has with your business – and that includes every word you publish.
When your copy sounds like everyone else’s copy, your brand sounds like everyone else’s brand. You become a commodity. And when you are a commodity, the only thing left to compete on is price.
I’ve been doing branding work for 25+ years. The businesses that build strong brands are the ones that have a clear voice and stick to it. That voice does not come from a prompt. It comes from understanding what you stand for and having the discipline to express that consistently.
AI copy is inconsistent by nature. Run the same prompt twice, and you will get two different results. Use it across your site without careful editing, and your brand voice becomes a patchwork of vaguely related tones. Visitors notice that, even if they cannot articulate why.
The ten-minute problem.
There is a broader issue at play here: an attitude problem.
When a client spends weeks on design and ten minutes on copy, they are telling you something about how seriously they are taking the words on their site. They are treating copy as a checkbox rather than as the most important commercial tool on the page.
Words are the most powerful thing in any marketer’s toolkit. Cleverly crafted copy engages visitors and converts them. Design gets judged in a fraction of a second, and then people start reading. If what they read does not land, the design was irrelevant.
Using AI to bash out copy quickly is usually a symptom of this attitude, not a solution to it. It gets words on the page without any of the thinking that makes those words worth reading.
Where AI actually helps with copy.
I am not saying AI has no place in the copywriting process. It does – just not in the role most people give it.
AI is genuinely useful for:
- Brainstorming angles and structures before you write
- Checking grammar and readability on a draft you have already written
- Generating first-pass meta descriptions that you then edit properly
- Repurposing existing copy you have written into a different format
- Helping you break writer’s block on a section you are stuck on
Notice the pattern there. In every useful case, you have already done the thinking. You are using AI as a tool to assist with work you have defined, not to replace the work entirely.
The moment you hand over the strategic thinking – the positioning, the voice, the specific claims about your business – you have handed over the thing that makes the copy worth having.
The bottom line.
If you are a small or medium-sized business, your website is probably your most important marketing asset. The copy on that site needs to clearly explain what you do, who you do it for, why they should trust you, and what they should do next.
That requires someone to actually think about your business. AI is a text prediction engine. It is very good at producing text that looks like the text you asked for. It is not thinking about your business.
Hire a copywriter. Brief them properly. Give them time to understand your customers and your positioning. Then let them write something that actually sounds like you, speaks to the right people, and gives Google something worth indexing.
Ten minutes and a prompt is not a content strategy. It is a placeholder.
David Foreman
Dave is the MD at Toast and has been working on branding, creative and web development projects for over 25 years. He's a founding member of Toast and enjoys a good rant.
