AI in marketing: separating hype from useful

The pressure to "do something with AI" has become a constant background hum for marketers. And if you're feeling slightly exhausted or overwhelmed by the whole conversation, you're not alone. The phenomenon has a name now too - AI imposter syndrome. That nagging sense that everyone else has figured this out, while you're still trying to work out which tool does what. If it helps, research suggests most of your peers feel exactly the same way. Marketing Week's 2025 State of B2B Marketing survey found that over half of B2B marketers (51.7%) recognise an AI skills gap within their teams. One respondent summed it up bluntly - they feared "being left behind" as AI adoption accelerates around them.

Here's where we stand at KISS: we're not calling ourselves AI evangelists or sceptics. We're practitioners. We've been testing, learning and integrating AI tools into our work for a couple of years now, and we have some thoughts on what's genuinely useful versus what's just noise so you can protect your energy and use it where it’s worth it.

What AI is genuinely good at right now

Let's start with the good news. AI has become genuinely helpful for many in the marketing sector.

Research and summarisation is where we've seen the biggest wins. Pulling together competitive analysis, synthesising long documents, or getting a quick overview of an unfamiliar topic used to eat days. The quality isn't perfect, but as a starting point it's valuable, and can often offer a new perspective.

Drafting and iteration has improved significantly too. We're not talking about letting AI write your final copy. We're talking about using it to generate rough outlines, explore different angles on a message, or quickly produce multiple variants of a headline for testing. It's a thinking, collaborative partner more than a wholesale ghostwriter.

Data analysis and pattern spotting is another strength. If you've ever stared at a spreadsheet wondering what's actually interesting in there, AI can help you find the story faster. It won't replace your strategic thinking, but it can surface patterns you might have missed.

Speed and scale are what we hear everyone talking about too. But this is where we need to be careful, because speed without direction just gets you to the wrong place faster.

Where AI still falls short

Strategy and prioritisation remain stubbornly human - thankfully! AI can generate options, but it can't tell you which option is right for your business, your market, your moment. It doesn't understand your commercial pressures, your board's appetite for risk, or why that particular stakeholder always pushes back on certain approaches.

Brand nuance and tone are harder than they look. Most AI-generated content has telltale sameness to it. It tends toward the generic, the safe, the slightly over-polished. We’ve also seen it create work that didn’t need to exist, simply because it was easy to generate. In practise, getting AI to capture and relay what makes a brand distinctive requires significant human input and editing. And if your brand voice isn't clearly defined in the first place, AI will only amplify that problem.

Stakeholder politics don't compute. AI has no idea that the CEO hates that particular phrase, or that the sales team will never use materials that don't mention the product first. Context that seems obvious to humans is invisible to AI.

Originality and taste are perhaps the hardest gaps to bridge. AI is trained on what already exists. It can recombine and remix, but genuine creative leaps – the ideas that make people sit up and pay attention – still come from human minds. Brian Macreadie of Addleshaw Goddard puts it well: ‘AI alone won't increase your market share. Strategy still drives growth. The technology can accelerate execution, but it can't tell you what's worth executing in the first place.’

How we're using AI at KISS

Rather than speaking in generalities, here's what we've been doing:

• Our developers are using AI to accelerate code review, generate boilerplates and troubleshoot bugs. The time saved on routine tasks means more time for the complex problem-solving that requires human judgement. We've found AI particularly helpful for explaining unfamiliar codebases and suggesting optimisation approaches.
• Our content team treats AI as a research assistant and sometimes a first-draft or structure generator. We use it to explore topic angles, pull together background information and create rough frameworks for longer pieces. But the strategic thinking, the point of view, and the final polish remain human work. Anything that touches brand voice gets carefully edited, reviewed and yes – read aloud (our golden rule for content!).
• Our project managers are experimenting with AI for meeting summaries and task organisation. The wins here are incremental but very real - less time on admin, more time on the conversations, thinking and decisions that move work forward.

What runs through all of this is our defining red-thread: AI doesn't replace thinking. It accelerates certain types of execution, which frees up time for the thinking that matters.

Common traps to avoid

We've seen enough AI adoption attempts to spot the patterns that lead nowhere.

Tool-first adoption is the most common mistake. Buying a shiny new AI platform before you've identified a problem to solve usually results in expensive software that nobody uses. Start with the workflow you want to improve, then find the tool that fits.

Expecting AI to fix unclear thinking is equally problematic. If your strategy is muddled, AI will generate muddled outputs at scale. If your brand positioning is vague, AI will produce vague content faster. The garbage in, garbage out principle applies with particular force here.

Automating before understanding creates its own problems. If you don't understand why a process works (or doesn't), automating it just locks in your existing inefficiencies. Map your workflows before you try to optimise them.

Treating AI output as neutral is a trap that still tricks even the most experiences professionals these days. AI tools have biases baked in from their training data. They can also generate confident-sounding nonsense. Human review isn't just quality control in this era - it's necessary fact and bias checking.

A practical way to prioritise

If you're trying to figure out where AI might help your marketing function, here's a simple filter we've found useful.

First, ask whether the task primarily saves time or creates clarity. AI is excellent at both, but knowing which you're optimising for helps set expectations. Time-saving tasks can often be automated more completely. Clarity tasks usually require more human involvement in the loop.

Second, consider the risk of error. For internal documents and early-stage thinking, errors can be cheap to fix. For client-facing work, published content, or anything touching compliance, the bar needs to be significantly higher. AI should always assist but never lead on high-stakes outputs.

Third, think about whether the task touches brand or reputation. The more directly something affects how your organisation is perceived, the more carefully AI outputs need to be reviewed, refined and edited.

Not every task passes this filter. That's fine. AI doesn't need to be everywhere to be useful. The most important thing to remember is this: you don’t need to have all the answers right now. AI isn’t a race, and it isn’t a badge of credibility. Start small, stay curious and stay critical. Focus on problems worth solving, not tools worth testing. Used thoughtfully, AI can make good marketing teams better by giving them more time to think, question and create. That’s where the real advantage lies.