Writing better prompts: How to get more from AI

If you’re running a small or medium sized business, it’s likely that you’re using artificial intelligence in some capacity. A recent report from the Small Firms Association (SFA) in Ireland found that 9 out of 10 small firms are using AI for one or more business processes, including the automation of simple tasks, data analytics reporting.

However, despite the growing number of SMEs using AI, many struggle to unlock the technology’s full potential. That’s because the quality of AI input is directly tied to the quality of our input. That means if you learn how to write good prompts for AI, you’re more likely to get results that are more useful – meaning you need to spend less time editing or re-prompting and can spend more time focused on what really matters to your business.

So, let’s look at the art of writing AI prompts, helping you to turn vague requests into specific instructions that deliver genuinely useful results.

What does a good AI prompt look like?

A prompt is the request, question or task you give to your AI tool. Whether you use ChatGPT, Microsoft Copilot, or any other AI assistant or agent, you need to give it directions, so it knows what content to create or what information to give you.

When it comes to good AI prompt writing, you need to be clear, specific, and leave little room for misinterpretation.

When you give the AI tool enough information to understand not just what you want, but also how you want it to be delivered, you’re more likely to get good results. Consider these two AI prompt examples:

• Bad example: “Write marketing copy for B2B project management software”

• Good example: “Write a 150-word product description for B2B project management software, targeting small business owners. The main USPs are that it’s easy for SMEs to implement and saves them time compared to their current manual processes like spreadsheets and status update meetings. Write using a tone of voice that’s professional but approachable.”

The first prompt is unlikely to generate the results that you’re looking for, as it’s too vague. The ‘marketing copy’ produced could be anything from a social media caption to a full sales landing page.

The second prompt, on the other hand, provides clear parameters. It specified the word count, target audience and key benefits, as well as what type of marketing copy, and the desired tone. This helps guide the AI tool to produce content that more closely matches your needs, giving you a much better starting point.

It’s important to carefully review any content created by AI tools, as it can sometimes be misleading, inaccurate or even offensive. It’s a good idea to treat any content created by AI as a first draft. That means you should take the time to review it for accuracy, tone, and brand alignment before using it in your business communications.

How to write good prompts for AI

Now that you understand more about what separates good prompts from bad ones, let's dive into the specific techniques you can use to improve your own prompt writing.

Be specific about format and length

Always specify what you want the AI tool to produce, in what format, and an approximate length. Instead of asking for “an article”, it would be better to request “a 1000-word blog post that uses H2 headings and bullet points for key information”. This means that the AI will produce content that’s in line with what you’re looking for, rather than something that’s too long or too short.

Show what ‘good’ looks like

Providing examples of things that have worked well for you in the past, or other industry examples that you like, can be a great way of letting your AI tool know the style, tone and format you prefer. It helps to give the tool a concrete reference point to work with, rather than relying on abstract descriptions. If you’re using examples from competitors, it’s important to note that it’s for inspiration only, and it shouldn’t be copied.

Role play

Asking your AI tool to assume a specific role or perspective can help it to craft the right narrative, based on your business needs or target audience. Rather than just asking for feedback on wireframes, you could ask it to “act as a senior UX designer reviewing our mobile app wireframes and give feedback on conversion bottlenecks and user experience pain points”. This targeted approach means you’ll get more specific, specialised insights rather than broad, general suggestions.

Be explicit about what you do and don’t want

Specifying what you want the answer to include – and importantly, what you don’t want to be included – can save you time. If you’re writing an article and know that you want to include supporting facts and figures but want to make sure that the language is easy to understand, even for non-technical readers, then you could say: “Include specific metrics and sources but avoid technical jargon”.

Define the audience

For the best results, you should specify who will be reading or using the output. A prompt that states “write for customers with limited technical knowledge” will produce very different results to “write for experienced software developers with deep technical understanding”.

Refine and reiterate

The key to getting the most out of AI tools is not to expect too much on the first attempt. If the output isn’t quite what you were looking for the first time, use follow-up prompts to refine it. You could say things like:

• “Make this more conversational”

• “Add more specific examples”

• “Reduce the length by 200 words while keeping the key points”

Find out how AI can help your business

It can take some trial and error to get it right, but if you can master the art of prompt writing, it will help to unlock the true potential of AI. By being specific about your needs, providing relevant context, and iterating on results, you can transform AI into a powerful business tool.

Find out more about how you can use AI to solve common business problems.