Prompting - how to go about it

Artificial intelligence (AI) and tools such as ChatGPT, Midjourney, Copilot, Gemini and Claude have become an ordinary part of more than just marketing teams. This technological boom has brought not only new possibilities but of course pressure on speed, performance and constant creativity too. AI can do almost anything. Write an article, create a slogan, analyse data. But only when you know exactly how to ask the question properly, or rather how to give it a good prompt. A badly framed prompt leads to a poor output, pointless rewriting and lost time. The quality of the brief becomes the decisive factor between an average and an exceptional result.

"Even the best AI tool is only as good as your prompt."

AI today is not just a tool but a creative "sparring partner"

In digital marketing we long grew used to seeing tools as auxiliary means. Applications that carry out routine tasks, save time and raise efficiency. But with artificial intelligence's arrival the very nature of the work is changing. AI is no longer just a "tool" in the true sense of the word. It is becoming our creative sparring partner. Imagine, for instance, AI as a colleague available 24/7 who does not forget, does not know what a mental block is and can react immediately to any challenge. And that from proposing a campaign, through analysing data, to creating copy, an idea or a concept. It has no ego, does not worry about deadlines and never says "I don't know". And it is exactly at this moment of collaboration that a new quality arises. It is not that AI will replace people. Quite the opposite, AI complements people where time, energy or inspiration falls short. And a marketer or copywriter who can work with AI effectively becomes more effective and more creative. Collaboration between people and AI is not the future, it is the present.

Why is prompting the foundation?

Only a few years ago, writing an article, creating a campaign or an analysis took several hours of focused work and a team of people. Today? A few seconds are enough… and a well-framed prompt. But although AI brings almost unlimited possibilities, using it has one key condition: you have to tell it exactly what you want from it. And it is exactly here that prompting comes onto the scene. That is, the ability to frame a quality brief for artificial intelligence so that you get a relevant, accurate output. A bad prompt? You get vague answers, generic content and spend more time on revisions to reach the result you want. A good prompt? You save hours of work, eliminate misunderstandings, speed up brainstorming and production and, above all, gain an output that makes sense.

How to prompt properly?

The ability to prompt already belongs among basic digital skills. And the sooner you master it, the greater the head start you gain. Effective prompting is not about a single sentence along the lines of "create a slogan" but about setting the task, the context, the aim, the expected format and the role in which AI is to answer properly. A good prompt structure markedly raises the output's quality. A properly defined prompt contains five basic elements. The basic structure of an effective prompt therefore consists of five elements: the task, the aim, the context, the role and the format of the answer.

The task represents the specific action we require from AI. That can be, for instance, creating text, analysing data, rephrasing content, proposing ideas, summarising information or another type of output. The task set must be clear and comprehensible. Ideally it is framed as a command in the active form, for instance "create a blog article", "analyse the campaign's performance" or "propose 10 ideas for a slogan".

The aim explains why we are giving the prompt and what the final result is to be. If AI knows the purpose, it can adapt the answer's tone, depth and structure. The aim can be, for instance, raising sales, educating an audience, evoking emotion or preparing material for a presentation. Setting out the aim helps AI grasp the wider context and focus on the output.

The context supplies the details and connections AI needs to know. Context can cover information about the brand, the target group, the product, the current situation, the tone of voice or particular requirements matching the given environment. It can also be a reference to existing content, competitors' examples or a platform. Sufficient, accurate context removes assumptions and minimises inaccuracies in the output.

The role determines the perspective from which AI is to answer, that is, who it is to become when generating the output. This is a key aspect for setting a suitable style, vocabulary, depth and expertise in the output. AI can appear as a marketer, copywriter, UX designer, lawyer, consultant, psychologist or another expert depending on the user's need. A precisely defined role helps AI take the right stance and tone, in line with the brief and the target group.

The format of the answer defines the form in which the output is to be delivered. AI can generate outputs in various forms, such as paragraphs, bullet-point lists, tables, structured articles, infographic descriptions or brief summaries. If the result is meant for a particular channel (for instance a LinkedIn post, a newsletter, a blog article), the format of the answer should reflect that. It equally matters to take the language level, length and structure of the output into account.

Practical tips on creating better prompts

Effective prompting is a skill that can be improved systematically. Even a small change in the way a brief is framed can have a fundamental effect on the quality of the output artificial intelligence generates. In this part we will introduce a few tips for more relevant, more accurate answers when working with AI.

1. A good prompt - do not beat about the bush

One of the most frequent mistakes when writing prompts is excessive caution or pointless politeness, which in communicating with people does come across as courteous but with AI causes vagueness. Artificial intelligence needs no polite phrases such as "please", "thank you", nor will it appreciate lengthy specification of a task without a specific brief.

Why does that not work?

AI does not work with emotions, intentions or non-verbal hints. Instead it relies on clear, structured information and explicit instructions. Too cautious or general a brief leads the model to:

• not know what to do first,
• not know the task's scope,
• not know what output format you expect,
• or head in the wrong direction, because it is guessing what you meant.

A tip: always start straight with a command: use strong, specific verbs such as:
"Analyse", "Compare", "Summarise", "Propose", "Create", "Generate", "Categorise".

2. Use positive formulations

When working with AI the rule is that clear, directly expressed requirements lead to better results than prohibitions or restrictions. In many cases users frame instructions in the negative. What AI is not to do and what they do not actually want from it.

Why do negative instructions not work?

Artificial intelligence has no intuition, cannot "work out" what exactly you intend, and a negative formulation can leave too much room for interpretation. The result can be an answer that does meet the prohibition but will not be usable.

A tip: turn prohibitions into precise instructions. Instead of "do not do", write "do it this way".
The average AI model cannot always evaluate the consequences of a negation, above all if the prompt is ambiguous.

3. Divide a complex prompt into smaller parts

Many requirements aimed at AI tend to be too general or too complex. That leads the model to "get lost" in the brief's wide scope, be unable to determine priorities, or answer superficially and inaccurately. The solution is dividing the task into smaller, manageable steps that are easy to carry out and follow on logically from one another.

Why does dividing work?

• It helps structure the output – AI knows what to devote itself to and in what order.
• It lowers the risk that AI will leave out an important step.
• It allows checking the quality of individual parts.
• It raises the answer's accuracy and depth, because every part has its own specific aim.

A tip: use ordinal numbering or headings. Clear numbering or parts marked with headings (e.g. Step 1: Analysis, Step 2: Proposed improvements) make it easier for the model to find its way and often lead to outputs with a clear structure and logical flow.

4. Be as comprehensible as possible when prompting

Comprehensibility is a basic pillar of quality prompting. Artificial intelligence does not read between the lines and does not understand unclear or incomplete requirements the way a person might. The more unambiguously, specifically and precisely you frame your requirement, the greater the likelihood that the result will be usable without further revision.

Why does that work?

• AI processes language by probability and patterns – not by intuition.
• With an unclear brief it "fills in" what you might have meant – and often gets it wrong.
• If the aim, scope, tone, format or context is missing, the output can be generic, shallow or entirely beside the point.

A tip: use 5W + 1H. If you are not sure whether your prompt is comprehensible, check whether it answers these basic questions:

• Who? – Who does it concern? (e.g. the target group)
• What? – What exactly is AI to do?
• Why? – Why do you need it? What is the aim?
• Where? – Which channel/platform does it relate to?
• When? – At what time? Or over what scope?
• How? – In what format is the output to be?

Just as we once learnt to write an essay, put together a presentation or frame a business email, today we are learning how to ask questions and set tasks for artificial intelligence so that it understands us and is helpful. It matters to realise that prompting is an interactive process. Sometimes a small change in one sentence, adding context or defining the aim more precisely is enough and the answer's quality improves dramatically. In practice prompting is not only about gaining text outputs. It is a tool for testing ideas quickly, creating prototypes, analysing data, planning campaigns, researching information and even developing creativity. In the right hands artificial intelligence becomes a partner in solving problems, a creative adviser, an analyst and a copywriter.

Learning to prompt does not mean learning to "control AI". It means learning to collaborate with AI. And collaboration starts with being able to say what we actually need.