Head of Content
Imagine the hours saved by a chatbot that can write anything. That can answer anything!
Sounds like an AI dream for marketers looking to cut costs. And a nightmare for content professionals who’ve spent their careers honing and adapting their craft.
But if you’re here for the drama, I’m going to disappoint. As a content marketer, a lack of writers and ideas isn’t what keeps me up at night. Focusing on these is more of a red herring than silver bullet – distracting from the bigger, more structural dramas that deserve our energy.
That’s impressive. But it’s never been an issue finding writers. The difficulty is finding good ones.
“It’s like when your child does a drawing,” an SEO Director playfully described ChatGPT’s output to me recently, while discussing its pros and cons. “You’re proud of them, but you’re not going to frame it and put it up on your bathroom wall.”
As content guardians, we can’t risk brand reputation (and our own) on immaturity. We look for writers who:
Know their subject area, audiences and nuances within them. Who get the differences between local markets, and stages of a customer’s journey.
Understand how to write for different digital channels – the language, structure and tone considerations for each.
In editorial, can research topics thoroughly and bring original thought. In technical copywriting, can make dry subjects clear and engaging.
Are the type of human who can deal with difficult topics sensitively and engage through expertise and vulnerability.
One of the reasons clients struggle to bring content operations completely in house is that it’s unlikely one person will be able to tick all of these boxes.
To create a fully-formed content programme, you need a network of writers, editors, subject specialists and real people, willing to share their stories. Trying to create this in a permanent internal team can become a false economy that sees senior stakeholders become disillusioned with the return on content investment.
To test its chops, I prompted ChatGPT to write me a handful of guides on medical topics. Then I noted edits I’d raise if working with a writer who could make a cuppa in the office round.
The main red marks came from the generality of the guidance it churned out, which won’t cut it for search algorithms that (based on user needs) emphasise the importance of experience, expertise, authority and trust for sensitive topics.
While AI models could be used as a starting point to summarise source material in a digestible way, there will always need to be a next-step layer of unique experiences and expertise to build trust.
Then there’s the structural issues. While there’s nothing wrong with opening a sentence with “it’s important to”, it’s a problem to start every other sentence this way. Good writers consider the context of their subject and impact of their words on the audience and use repetition sparingly and strategically.
Imagine being newly diagnosed and overwhelmed. Emphasising every sentence as important could exacerbate anxiety, lessen the power of key points and make it difficult to keep track of where you are when scanning.
Starting with a declarative rather than imperative also means it takes longer to get to the point. It’s less direct than simply giving instructions, “find a support group”, and less compelling than leading with the benefit of doing so, “it could help you feel less alone to find a support group”.
Writers will always need experienced editors to refine the content, structure and tone of their work, as well as to fact-check and proof. AI-written content even more so – there’s no cost saving there. Editors also guard a brief and translate vague client feedback into actionable amends. It’s a human-to-human job.
Granted, ChatGPT can be used as a starting point for source material. But first you need to know what you’re looking for, as it can only draw from what’s already out there.
You need people for original ideas and AI can’t replace a strategist who shapes why and how these are generated. It can’t build the support structures to activate them: defining content and channel purpose, testing strategies, filters to prioritise the best ideas, controlled vocab and content design systems. And it definitely can’t play diplomat between individuals who have subjective opinions on which ideas are best and most pressing.
Of course, AI is only as intelligent as you can teach it to be. And there’s a fair bit of lesson prep before you can ask it to consistently create relevant, on-brand prose.
You can cover a topic by multiple angles, for multiple audiences. How you identify and choose which makes the difference between helpful and irrelevant content.
Then you need to get AI writing in your brand voice. To do this you could, in theory, feed it your ToV docs or let it learn patterns from your existing content. But this relies on you having so thoroughly defined and implemented the character, construction and style of your voice – for different contexts – that a language model could pick it up.
The fact is that these content foundations are usually shaky at best – often non-existent. It’s here where the real work lies.
Absolutely – metadata. A grey area between SEO and Content, it’s no one’s favourite task at scale. If we book-end with clear inputs and edits, here we could put ChatGPT to work.
As content professionals, we can’t afford to assume. Yet that’s all AI can do. Assuming we just need more, cheaper, easier words is a distracting oversimplification when dealing with the intricacies of human connection.
We believe that moving too slowly in digital is the biggest risk your business faces. If you are ready to move faster in digital, we are here to help.