My husband is a bit freaked out that I’m going to lose my job to AI.
He’s not the only one worried. Headlines like Qantas axing 850 jobs in an offshoring deal with Accenture have plenty of people in marketing wondering what’s next. And business owners are asking a fair question: if AI can write your blog posts, build your reports and set up your Google Ads, why pay an agency at all?
(For the record, I’ve told him I’m not going anywhere. I did let AI help structure this post, though, so you can be the judge.)
My answer is that AI will replace a lot of what agencies used to charge for. It won’t replace what businesses actually need from a marketing partner, and the gap between those two things keeps growing.
What AI is already doing well
AI is now very good at a lot of marketing execution: first drafts, reporting, research, routine campaign admin and simple creative variations.
If an agency’s main value is selling hours for those tasks, that agency should be worried. Those hours are getting cheaper and faster every month, and clients can see it.
Where marketing still needs people
Here’s how I see the split:
Where AI helps
Fast, tireless, good at patterns
- First drafts and outlines
- Summarising data and reports
- Research and competitor scans
- Routine campaign admin
- Variations for testing
- Formatting and publishing
Where marketing needs people
Judgement, empathy and accountability
- Understanding your customers
- Deciding strategy and priorities
- Original ideas and creativity
- Judging what’s true, on-brand and good enough
- Building trust and relationships
- Owning the results
The parts of marketing that make the biggest commercial difference are the parts AI is weakest at. The first is understanding your audience. Marketing is still aimed at humans, and knowing what your customers worry about, how they talk and what makes them trust you comes from real conversations with them. A prompt can’t give you that.
Then there’s deciding what matters: which customers to chase, which offer to lead with and, just as importantly, what to leave alone. Those are strategy calls, and they depend on knowing the business.
Someone also has to join the dots so your website, ads, SEO, email and social are working towards the same goal instead of competing for budget. And someone has to check the work. AI produces confident-sounding errors and off-brand copy, and those need catching before they go out with your name on them. When the results come in, good or bad, a person should own them.
The bigger risk is sounding like everyone else
When every business in your category has the same AI tools, they tend to end up with the same output: the same blog topics and the same promise to be “your trusted local experts”. AI learns from what already exists, so left to its own devices it gives you the most average version of your marketing.
That’s a real problem when you’re trying to win customers. If your website, ads and social posts could belong to any of your competitors, buyers have nothing to choose you on except price. A strong brand gives them another reason. It’s a clear point of view, a look people recognise and a way of talking that’s yours. We’ve written more about this in AI can write your marketing, but it can’t tell you what you stand for.
Being different has to show up in your messaging as well as your visuals. The best campaigns usually start with a question nobody else in the category is asking, or an angle a competitor would be too cautious to try. AI is useful for testing and refining an unusual idea once you have one. It’s much less useful at coming up with it, because it’s built to predict the most likely answer, and the most likely answer is the one your competitors are already using.
That’s why branding and strategy come first in our work, before anyone writes a headline or builds a campaign. If your website has started to feel interchangeable with the rest of your industry, our piece on AI-built websites losing their personality is worth a read.
How a marketing partner keeps its value
Over the next few years, saying “we use AI” won’t set an agency apart, because every agency will be saying it. The ones worth working with will use AI to strip out low-value work, then spend the time they save on the things clients still need:
Audience understanding
Marketing is still aimed at humans. Knowing what your customers worry about, want and trust.
Strategy
Choosing where to focus, what to say — and what not to do.
Creativity
Ideas that stand out when AI makes generic content everywhere.
Judgement
Checking accuracy, brand fit and risk before anything goes out.
Accountability
Someone who owns the results and changes course when needed.
In practice, I’d expect a good marketing partner to offer:
- Reporting on outcomes such as qualified enquiries, sales and repeat customers, rather than on how many blog posts were delivered.
- Strategy before channel work starts, with positioning, customer insight and a clear plan agreed up front.
- Human review you can actually see, including fact-checking, brand voice checks and approvals.
- Measurement that connects activity to leads and revenue, then tells you what to do next.
- Visibility in AI search, so your business shows up and is trusted in Google’s AI Overviews, ChatGPT and other AI answers.
What this means for you
If you’re weighing up whether to keep paying for marketing support, ask three questions:
- Is my agency (or team) setting direction, or just completing tasks?
- Do my reports lead to decisions, or just describe what happened?
- Is anyone accountable for whether marketing is actually growing the business?
If the answers aren’t clear, it’s worth having a conversation, whether that’s with us or someone else. Book a 30-minute chat and we’ll give you a straight answer on where AI can help your marketing and where you still need people.
As for my husband, I’ll let him know my job’s safe. It just looks a bit different now.
AI transparency note
AI helped outline this article, edited my draft and — I’ll admit it — genuinely made my writing more structured. The perspective and client experience are mine, and I approved the final version. (Any questionable puns, sadly, are also all mine.)
Katie Cowling · Director of Digital Services How we use AI ›


