AI in marketing: what's worth worrying about

Most of the worry about AI in marketing is pointed at the wrong thing. People ask whether it'll replace the people who do the marketing. For a business selling equipment or technical services, I think there are two better questions: is what AI says about your business accurate, and is anyone in your office pasting customer details into a chatbot?

Your buyers are already using it

Gartner surveyed 645 B2B buyers in late 2025 and published the results in May 2026. Forty-five per cent had used generative AI during a recent purchase, mostly to look into vendors and products. So when a buyer asks an AI tool to compare suppliers of what you sell, an answer about your business gets written whether you took part or not.

The same survey found 69 per cent of buyers prefer to check what AI tells them with a sales rep. And 51 per cent said they're more likely to get misleading information from AI, which would be more reassuring if 49 per cent hadn't said the same about sales reps.

AI tools build their answers from what's published, and your website is one of the first places they'll look. If it says plainly what you sell, where you deliver, which standards you meet and how long things take, there's something accurate to work from. If it's three paragraphs about quality and service since 1994, the tool fills the gaps from somewhere else.

Where it's genuinely useful

With someone who knows the subject checking the output, AI is good at the slow bits. It'll turn a dense spec sheet into a plain first draft, boil a forty-minute call down to the five things the customer actually asked for, or rough out a follow-up email that a salesperson then makes their own.

The person still decides what's true and what goes out. AI makes the first draft cheaper. Nobody gets to blame it for the final one.

Where it goes wrong

It mostly goes wrong through confident mistakes. AI can write a lovely paragraph with the wrong load rating, a standard that doesn't apply or a statistic nobody ever published. In a technical business, a wrong number on a product page can lose a sale or create a liability, so anything with a figure in it gets checked against the source.

It also makes everyone sound the same. If every supplier in your category asks the same tool to "write a homepage for an industrial equipment supplier", they'll all get roughly the same page. The things that actually set you apart, like your real capacities, the brands you won't carry and the job you finished last month, are exactly what a generic prompt leaves out.

And it tempts people to publish at volume. Google's spam policies call this "scaled content abuse", meaning lots of pages made mainly to game search rankings, and they say it applies no matter how the content was created. Fifty thin articles knocked out in an afternoon fit that description pretty well.

The privacy bit people miss

In October 2024 the Office of the Australian Information Commissioner published guidance on using AI products. One of its key recommendations is that organisations don't enter personal information, and particularly sensitive information, into publicly available generative AI tools.

Sales and marketing teams are full of personal information: customer lists, call notes with names in them, email threads, the quotes spreadsheet. Pasting any of it into a free chatbot to tidy it up is the kind of thing that happens on a busy Tuesday without anyone thinking of it as a privacy decision. I'd write a one-page rule on what can go into which tools, and make sure everyone's read it.

Where I land

Use AI where someone who knows the subject checks what it produces. Don't publish anything nobody's read. Keep customer information out of public tools. And worry less about being replaced than about being described wrongly, because that can happen to any business whose website doesn't say much.