The AI Hype in Marketing: What Actually Helps a Small Business
We use these tools every day. They sit in our drafting, our repurposing, our alt text and our translation between Dutch, French and English, and we would not hand them back. We also would not let one of them publish anything with a client’s name on it without a person reading it first.
Both are true at once. Most of the hype insists on only one of them, depending on what is being sold that week. Here is the split as we experience it: what the tools do well, what they do not, and what changes once production stops being the hard part.
What the tools genuinely do well
Every item below is the same category of work — the part that comes after somebody has already decided what to say.
- Beating the blank page. A mediocre draft in seconds beats a perfect one in an hour, because you can argue with a draft. Starting is where most SME content quietly dies.
- Turning one idea into several formats. A customer story becomes a long post, three short ones, an email and a set of captions. The thinking was done once; the reformatting used to cost an afternoon.
- Captions and alt text at volume. Tedious, necessary on every image, and the first thing dropped when the week gets busy.
- Summarising comments and messages. A week of activity sorted into what needs a reply, what is a complaint, and who is trying to buy something.
- Translation between Dutch, French and English. Fast, close, and a real saving in a two-language market — as a starting point, never as a finished post.
Notice the shape of that. Every one of those tasks follows a decision. Somebody chose the customer story. Somebody knew the objection about lead times needed answering. The tool executes; it does not decide.
What they do not do
Shorter list, more important.
Strategy
Strategy is mostly decisions about what you will not do: the customer you are not chasing, the platform you are ignoring, the service you stopped promoting because it never converts. Ask a tool for one and you get something plausible, structured and completely generic — the average of everything ever written about your industry. It has the shape of a strategy and commits to nothing, because it has nothing at stake.
Knowing your customers
The useful things about your customers were never written down anywhere a model could have read them. That the decision in your sector is made by the operations manager and merely signed by the owner. That three deals last year died over lead times and none over price. That is not on the internet. It is in your head and in sales calls nobody transcribed.
Judgement about what to say
Whether to publish the post admitting you got something wrong. Whether to stay quiet while your industry is having an argument. Whether a joke lands in Flanders and not in France. These calls make a feed read like a company rather than a pipeline, and each is made by somebody with something to lose.
Producing publishable work unsupervised
This is the one that costs money. Left to run alone, a tool will state a delivery time you do not offer, describe a service you discontinued, or answer a customer in your comments with something confidently wrong — in the same fluent register it uses when it is right, which is exactly why it gets through. No amount of prompt refinement fixes this. It needs somebody who knows the business reading everything before it goes out.
When production gets cheap, taste gets expensive
For as long as small businesses have done marketing, the bottleneck was making the stuff. Writing the posts, sizing the images, drafting the newsletter. That is why so much SME marketing stops after six weeks — not a failure of ambition, a failure of hours.
That constraint has largely gone, and it went for your competitors on the same afternoon. So output stopped being evidence of anything. If everyone can produce thirty posts a month, thirty posts says nothing about you.
What stays scarce is what did not get cheaper: knowing which thirty are worth making, having a point of view a stranger would read twice, and still being there in month nine when the novelty has worn off. Cheap production helps with none of those. It removes the excuse for not having them.
The tools took away the excuse, not the work. What is left is the part software was never going to do for you: deciding what is worth saying.
— Ben Van Aken, Co-Founder & CTO
The obvious trap: every feed now reads the same
Same tools, similar prompts, same defaults. Anyone who spends time on a professional network has learnt to spot it — the opening line that restates the headline as a provocation, the tidy rule of three, the closing question nobody ever answers.
That is not a flaw. It is what the tools are built to do: converge on the most probable next sentence — which is, by definition, the one everybody else is getting. Reach for the default and you publish a feed that could belong to any of a hundred thousand businesses.
The fix is not to avoid the tools. It is to put in things only you have. Your numbers. The objection you heard three times last month. The way your production manager explains the process to a new hire. That cannot be averaged, because nobody else has it.
What to actually do with this
- Decide who you are talking to and what you want them to believe. An hour with a notebook, before any tool is opened.
- Use the tools for drafting, repurposing, captions, alt text and translation — the work that eats the hours.
- Make one named person accountable for every published word. You, an employee or a partner, but one person who reads all of it.
- Put your own material in: real numbers, real objections, real language from real conversations.
- Measure what you measured before — replies, enquiries, calls booked. Volume was never the metric and is even less of one now.
Nothing there is exciting, which is the point.
Tools multiply direction. They do not supply it.
This is where the hype gets it backwards. AI does not reduce the need for somebody to own your marketing — it raises the value of having one, because the tools amplify whatever direction they are pointed in, including none. A business with a point of view and somebody consistent to execute it gets meaningfully faster. A business with neither now produces more of nothing, sooner.
So the question was never which tool. It is who is steering, and whether they can keep it up for a year. That is a budget question. A full-service European agency runs roughly €2,500–5,000 a month. An in-house marketer lands near €55,000–80,000 a year all-in, and you still brief them every week. A dedicated Filipino team sits between the two: our packages start at €595 a month.
They use the same tools everybody else has — we would be suspicious of any team that did not. What you pay for is the person deciding what goes in, and the person reading what comes out before your customers do. If you would rather see the rhythm first, €1 a day buys one post a day on one platform.
See what each package includes
Be sceptical of anyone selling AI as a substitute for a marketing decision, and equally sceptical of anyone refusing to touch it — they are doing the tedious half by hand and charging you for it. The honest position is unglamorous: good tools, a clear brief, somebody who reads everything before it goes out.