Return to 'Spotlight on AI'
  8 mins

Prompting Best Practice: It’s All About Prompts

It's All About Prompts

If you’ve read The Hitchhiker’s Guide to the Galaxy, you’ll remember Deep Thought, Douglas Adams’ fictional supercomputer, an Artificial General Intelligence of unimaginable power.

Its creators, a race of hyper-intelligent pan-dimensional beings, ask it to answer the Ultimate Question of Life, the Universe, and Everything. Deep Thought thinks about this… for 7.5 million years.

While its circuits hum and millennia pass, philosophers debate what the answer might be. Finally, the big day arrives. The crowds gather, the computer clears its metaphorical throat, and in a voice of serene authority, it announces:

“Forty-two.”
After a stunned silence, Deep Thought explains: the problem wasn’t the answer – it was the question. Their prompt was so poorly framed that they’d need to build another computer entirely to discover what the question should have been.

And that’s the point.

AI – even ours, thankfully quicker than Deep Thought – is only as good as the prompts we give it.

So read on. This guide will help you craft prompts that turn your AI into a genuinely powerful thinking and creative marketing partner. Who knows, you might just find your own Ultimate Question.

This guide covers the key principles of effective prompting, common mistakes to avoid, and good → better → best examples you can try right away.

Be clear and specific

AI tools don’t read between the lines – they take your words literally. The more precisely you describe what you want, the more tailored and accurate the response will be. Vague prompts produce generic outputs. Being specific about task, goal, audience, channel, tone, format, length and context will help give stronger, more useable results. Clear prompts save time and make results easier to use, whether you’re requesting a content outline, summarizing research, or comparing competitors. For example:

Provide context and role

AI performs better when it understands the perspective it’s meant to take. Giving it a “role” – such as copywriter, market analyst, or category manager – helps shape its thinking and makes the outputs more relevant and actionable. This technique works for everything from internal summaries and content creation to brainstorming support and deep research. For example:

Use step-by-step instructions

Complex marketing tasks often involve multiple stages, which could include – research, synthesize, summarize, analyze, create, format – and AI handles these best when you spell them out in sequence. Breaking your prompt into smaller, numbered steps prevents the AI from guessing what you want first. Structured requests lead to structured outputs that are easier to review, edit and use. 

Ask for style, tone, and format

AI needs guidance on how the content should sound and look – it can adapt its writing and output style if you tell it what you want. This applies to all output including narrative documents, strategic plans, competitor summaries and PowerPoint presentations not just marketing communications copy. Specify if you want the output to be conversational, formal, friendly, professional, playful, or informative to guide tone. If you need a specific structure – such as “three short paragraphs” or “two bullet points and a CTA / takeaway”, include that too. You can also ask for particular output formats – tables, bullet lists, executive summaries or slide-ready outlines – to save editing time later. For example:

Refine and iterate

AI prompting is a process, not a one-shot task. Think of it as a conversation: you give direction, review the output and fine-tune it. Use iteration to clarify calculations, reformat insights, or adjust tone; small adjustments, such as “make this sound more casual”, “add three bullet points to summarize key benefits” or “explain this in simpler language”, can dramatically improve the end result. Iteration isn’t a sign of failure; each time you refine your prompt, the AI learns more about what you want, and your results will improve.

Avoid common mistakes

Even skilled marketers can overlook details when prompting. The most common issues are being too brief, forgetting to define the audience or purpose, and failing to specify the desired output. These gaps leave the AI to guess your intent, often producing bland, vague or inconsistent results. Another frequent mistake is using AI output without proper review – especially important in regulated industries like ours. Always edit, fact-check, and validate the final work (keeping in mind that it must be acceptable to HEMA etc.) before sharing it externally. For example:

Experiment and learn

Prompting is a much art as science. There is no single formula that works every time because each AI model interprets language differently. The best way to improve is to test variations – try different wording, formats, or roles, and compare the results. Over time, you’ll develop an intuition for how to phrase prompts that fit your objectives. For example: 

Looking for inspiration? Check out our article on marketing prompt examples or download our checklist to make sure every AI prompt you write delivers clear and useful results. Or for more information on using AI within Boston Scientific, explore the dedicated section on BSCMarketingU.

Your AI Prompting Checklist for Marketers

Use this checklist before, during, and after working with AI tools to make sure your outputs are clear, compliant, and useful. Download your checklist