It’s crazy to think we’re over three years since ChatGPT was released to the world. In that time, if you work in marketing, chances are you’ve had the thought: could this make my job a lot easier? Could I just get the robot to write my blogs for me?
It’s an appealing idea. Blogs that would normally take days to pull together, by gathering information, compiling notes, writing and editing. Not to mention the keyword research, trend analysis and optimising so that get your article actually gets found in Google is now suddenly feeling like a five-minute job. Content is often the fuel for many other channels – SEO, Social, Digital PR, even ads. Like I said, super tempting.
The honest answer is yes, you can use AI to produce content. Nice, well-structured, grammar-and-spelling-error-free content. But whether that content will actually perform, whether it will rank, resonate, and do the job you created it to do… (remember, unless you are a publisher, content is a means, not an end), well, that depends on a few things.
And that’s exactly what we’re going to dig into in this blog.
The AI content factory fallacy
Here’s the thing: AI doesn’t necessarily reduce those stages. Our experience is, it adds to them or at least substitutes them for new ones. It naturally plagiarises. It states things confidently that are factually wrong. It hallucinates entire statistics and sources. It has a habit of drifting tonally. A piece that starts sounding like your brand can end up reading like something entirely different by the third paragraph, even with the appropriate TOV uploads and prompts.
Everything that goes into getting a blog live on your site is a labour-intensive process. The reality is that using AI content can actually make that process more laborious, and risk-laden, not less.
Before anything AI-generated is website-ready, there are several hurdles it needs to jump through. And this is just the production part; we haven’t even considered the purpose of much of that content yet. AI content has been shown to create real problems for both traditional SEO and AI-driven search rankings, making it most useful, arguably, for pages you’d never index anyway. Think UX copy, terms and conditions (again, checked by legal).
So, before you hit publish, ask yourself, does your blog pass each of these hurdles?
Hurdle 1: Have you used a detailed prompt structure?
A good prompt is the foundation of anything useful that comes out of an AI tool. And by good, we don’t mean a single sentence lobbed into a chat box.
We can’t just type “write me a blog about the risks of using finfluencers” and expect something publish-ready to come back. It won’t. What you’ll get is something generic, probably surface-level, and almost certainly not written with your audience in mind. Recent studies have found that this content is not just homogeneous with anything else on the topic; it’s also slightly weird. It subconsciously gives your brain the ick.
What you actually need to do is give the AI proper direction. Who is the content for? What’s the key message? What tone should it take? How long should it be? What do you want the reader to do at the end of it? The more context and structure you feed in, the more useful what comes out will be. Again, this isn’t a substitute for a writer, and is only recommended for content that doesn’t need to be referenced by search engines, or ironically AI search results.
Think of it less like a search engine and more like briefing a junior writer, one who is very fast but needs clear instructions and has absolutely no knowledge of your brand, your audience, or your industry. If you wouldn’t hand a vague brief to a new team member and expect great work back, don’t expect it from AI either.
Hurdle 2: Have you uploaded your tone of voice document?
First things first: do you actually have a tone of voice document? If the answer is no, get in touch with our content team. We can help you identify your style based on the audiences you want to attract. Sales pitch over – for now.
But in all seriousness, this matters. Again, an LLM doesn’t know what you haven’t told it. Left to its own devices, AI will write in its own voice, and that voice is a kind of bland, middle-of-the-road average of everything it’s ever been trained on. It’s not your brand.
Uploading a tone of voice document — or at the very least, including clear guidance on how your brand communicates — gives the AI the guardrails it needs to produce something that actually sounds like you. Consistent content across your website isn’t just a nice-to-have; it builds trust with your audience and reinforces your brand at every touchpoint.
In financial services, this is even more critical. Your tone isn’t just a brand choice. It signals credibility, professionalism, and reliability to an audience that is making significant financial decisions. Getting that wrong or letting it drift because an AI was left to fill in the gaps, can undermine everything else you’re doing and leave you uncompliant.
Hurdle 3: Have you checked for plagiarism?
Okay, so you nailed your prompt structure, you uploaded your tone of voice document, and you’ve given the AI everything it needs. Go ahead — start clicking buttons and let the magic happen.
And there it is. A fully formed blog, right in front of you. Not bad, right? Not so fast, chief.
AI pulls its answers from a vast range of sources across the web, and sometimes it doesn’t just take inspiration; it lifts. Duplicate content is heavily penalised by Google, and if sections of your blog already exist word-for-word on a competitor’s site, you can kiss those rankings goodbye.
And if you’re not ranking, what was the point of any of this?
Passing hurdle 3 isn’t glamorous, but it’s non-negotiable. Before your AI-generated content goes anywhere near your website, you need to manually work through it, copy sections and paste them into Google, and see what comes back. If lines have been lifted from somewhere else, rewrite them. It’s that simple, and that important.
Hurdle 4: Have you checked it for accuracy?
Okay, this one feels like a no-brainer. But it’s easy to become complacent when you suddenly have the ability to produce twenty blogs an hour. The speed is seductive. But it doesn’t matter how fast the content is produced if it’s wrong.
AI hallucinates. If you’ve spent any time using these tools, you’ll already know exactly what that means. You’ll have watched it confidently cite a statistic that doesn’t exist, reference a study that was never published, or go so far down a rabbit hole that when you pull it up on the inaccuracy, it just rolls over with a cheerful “oh sorry, you’re absolutely right!” As if it knew all along.
In financial services, the stakes here are higher than in most industries. Inaccurate content isn’t just embarrassing, it can be misleading, it can erode trust, and depending on what’s been written, it can create compliance headaches, fines and even worse.
Beyond the regulatory angle, Google’s E-E-A-T framework — Experience, Expertise, Authoritativeness, and Trustworthiness — is still very much a ranking factor, and accuracy sits right at the heart of it. If your content can’t demonstrate genuine expertise and trustworthiness, it will struggle to perform. Full stop.
It’s always worth reminding ourselves that just as content is a “means”, not an “end”, so is online visibility. The goal is actually to provide trusted value for your customers so they prefer you over alternatives. Google devised the EEAT principles because they found those were the things customers valued.
So if you do use AI to write, check everything. Every claim, every figure, every source. The speed AI saves you in writing will need to be reinvested in derisking.
Hurdle 5: Do you actually want people to find it?
The steps we’ve covered above aren’t insurmountable. Writing a solid prompt, uploading your tone of voice document, checking for plagiarism, and verifying accuracy none of it is beyond a competent marketing team. It just takes discipline and time.
But here’s the question that ties it all together: do you actually want this content to generate leads? Do you want people to find it?
Because if the answer is yes, then AI-generated content alone isn’t going to get you there. Google has been pretty clear on what it rewards. In their own words, “those seeking success in Google Search should be producing original, high-quality, people-first content that demonstrates E-E-A-T.” not content that ticks a box. Not content that was generated, lightly checked, and pushed live. Content that genuinely serves the reader.
And it’s not just Google. We’re seeing LLMs increasingly pull their information from sources like YouTube, because human-generated, experience-led content is seen as more authentic and trustworthy. The platforms are actively trying to surface the real stuff.
Which means the bar isn’t getting any lower. It’s getting higher. As AI slop drowns the internet, mediocrity will wash away the low-end (there was plenty of rubbish content out there before AI came along, after all), but it’ll take even better content to keep your head above the waves.
AI can absolutely be part of your content process. But if you want your content to rank, to be surfaced by AI search tools, and to actually convert, then it needs a human layer. Your expertise, your insight, your experience. That’s what makes it people-first. And that’s what makes it perform.
So, am I allowed to use AI?
Let’s be really clear about something. This article is not telling you to ditch AI. We use it ourselves, to aid research, summarise briefs, review data sets and so on. But that’s to take away the drudgery so we can focus on the valuable, and frankly, fun work. Why would anyone let the machines do the rewarding work just to be left with the low-value drudgery?
Because when you add up everything involved in properly using AI to produce quality content — the prompt crafting, the tone of voice guidance, the plagiarism checks, the accuracy review, the SEO optimisation — the time saving isn’t quite as dramatic as it looks on the surface. It’s a shortcut that requires a surprising amount of work to actually make it happen. It’s like spending 20 minutes rearranging the dishwasher to fit that last dish rather than spending 5 washing the dish by hand (Ok, admittedly, I’ve done that).
We’re already seeing the results of teams that skip the work. And the results can be damaging. Take what happened to VideoGamer.com when they replaced their team of 20 writers with AI-generated content. Their organic traffic collapsed.
But here’s the important nuance – that wasn’t solely because they used AI. It was because they replaced qualified, expert-led content with AI and didn’t do the work to maintain quality. They didn’t jump the hurdles.
Google has been transparent about this. Their Search Quality Rater Guidelines state that the use of AI tools alone don’t determine page quality. AI can be used to create both high-quality and low-quality content. The tool isn’t the problem. The approach is.
So use AI if it works for your process. But use it carefully, use it with human oversight, and never let it replace the expertise and experience that make your content worth reading in the first place.
Final words
This is a strange article for us to write. We’re a digital marketing agency, and we’ve essentially just walked you through how to use free tools to do a job that our incredibly talented financial services content team (another shameless internal link, can you tell I want you to check them out) gets paid to do.
But Balance was created to help financial services marketers, and we know you’ll be getting pressure on this stuff, so all we can do is help arm you for this new world.
Nothing we’ve described here replicates the quality of genuinely expert-led content. Researched properly, written with real industry knowledge, crafted for a specific audience. That’s what consistently outranks competitors, because it’s exactly what Google and AI search tools are designed to reward.
In a world where everyone is chasing the quick win and AI content is flooding the internet at scale, doing it properly is actually your biggest differentiator right now. So clear the five hurdles, or hand it to someone who does this for a living. You might actually find that AI has made it’s easier than ever for good human content to win the visibility and customers you’re seeking.
