Download the Perfect Prompt Worksheet
There’s no way around it, AI is transforming the way we all work, whether we like it or not. From content creation, automating customer comms, to researching and market analysis, AI tools are becoming indispensable for businesses of all shapes and sizes.
According to the UK AI Council, businesses that effectively leverage AI communication can see productivity improvements of up to 40% in content-related tasks.
However, the key to getting the best results from whatever AI tool, is the ability to craft the right prompts. A good prompt makes all the difference – giving the AI enough of the right information so it can focus in on your topic, context, role and tone of voice.
This article will guide you through the Six-Pillar Framework we recommend, including a template that you can implement immediately. And to make implementation even easier, we’ve created an accompanying workbook that you can download at the end of this article.
So, here goes.
Pillar 1: Role Assignment
The first pillar in crafting effective AI prompts is clearly defining the role you want the AI to assume. Think of this as casting an actor in a film—you wouldn’t ask Alan Carr to play James Bond (though that would be interesting).
When I instruct an AI to “write a social media post about our new product,” I get vastly different results than when I ask it to “as a social media manager with 10 years of experience, craft an engaging LinkedIn announcement for our new software release.
Why does this matter? AI systems are trained on vast datasets that include different writing styles, knowledge bases, and perspectives. By assigning a specific role, you’re effectively filtering those possibilities and focusing the AI on the most relevant patterns.
Here’s some things to avoid:
- Assigning overly general roles (“as an expert”)
- Choosing roles that contradict the task (“as a poet” for technical documentation)
- Neglecting to regionalise roles for UK-specific content
For best results, select roles that align with your desired outcome and add specificity where regional knowledge matters. Consider the difference between “as a property expert” versus “as a RICS-accredited surveyor with experience in the Liverpool housing market.”
Pillar 2: Task Description
The second pillar focuses on how clearly you describe what you want the AI to do. In our experience, this is where most prompts fall short.
Vague instructions lead to vague outputs. When I ask an AI to “write something good about sustainable fashion,” I shouldn’t be surprised when the result misses the mark. However, when I specify “analyse the growth of sustainable fashion retailers in British high streets between 2020-2023, highlighting the three most successful business models,” I receive precisely what I need.
To eliminate ambiguity in your prompts, we recommend:
- Using specific verbs (analyse, summarise, compare, rather than “do” or “write”)
- Quantifying where possible (word count, number of points, time periods)
- Breaking complex tasks into clear components
- Considering British English conventions and terminology
A practical technique we use is to “explain it to a new intern” test. If your prompt wouldn’t give a new team member enough information to complete the task, it likely won’t give an AI system enough direction either.
Pillar 3: Output Formatting
The third pillar addresses how your AI-generated content should be structured. This becomes particularly important when integrating AI outputs into existing workflows or systems.
I often find that clients focus so much on content that they forget to specify format, leading to unnecessary reformatting work. Whether you need bullet points for a presentation, a formal report with sections, or JSON data for your website, explicitly stating your format requirements saves considerable time.
As an example, rather than simply asking for “product descriptions,” specify:
Format the output as:
- Product Name: [name]
- USP: [one sentence unique selling point]
- Description: [75-100 words in two paragraphs]
- Key Features: [3-5 bullet points]
- Technical Specs: [in table format with two columns]
This approach reduces content integration time by 78% and ensures consistency across thousands of products.
When specifying output format, consider:
- Document structure (headings, paragraphs, lists)
- Data formats for technical applications
- Tone and style parameters (formal/casual, British English conventions)
- Length requirements for each section
- Visual formatting elements like tables or numbered lists
For businesses adhering to specific reporting standards or industry regulations, format specifications are doubly important. We’ve found that being explicit about compliance requirements in your format instructions helps ensure AI outputs meet necessary standards.
Pillar 4: Audience Definition
The fourth pillar focuses on clearly defining your target audience – perhaps the most overlooked yet critical element that determines whether your AI-generated content resonates or falls flat.
Without clear guidance on who you’re communicating with, AI systems default to generic content that attempts to speak to everyone but connects with no one.
Think about the stark difference between explaining investment strategies to seasoned financial professionals versus first-time investors.
Provide your AI with these audience guidelines:
- Demographic details: Age range, professional background, income level, and geographic location (particularly important for UK regional nuances)
- Knowledge level: Specify whether your audience are novices, intermediates, or experts in the subject matter
- Pain points and motivations: What specific problems are they trying to solve?
- Decision-making stage: Are they researching, comparing options, or ready to purchase?
- Existing relationship: Are they new prospects, existing customers, or long-term clients?
How much audience detail is enough? We recommend following the “stranger test” – provide enough information that someone who has never met your audience could still create tailored content for them.
For B2B contexts, include industry-specific challenges, regulatory considerations, and typical approval processes. For consumer audiences, lifestyle factors and value priorities often prove crucial for content that connects.
By mastering audience definition in your AI prompts, you’ll transform generic outputs into precision-targeted content that speaks directly to your intended readers’ needs, concerns, and aspirations.
Pillar 5: External Context
The fifth pillar acknowledges that AI systems, despite their impressive capabilities, lack specific knowledge about your business, industry trends, or previous interactions unless you provide it.
Without context, even the most sophisticated prompt will yield generic results.
As an example, if you’re a heritage property management company on The Wirral, you need AI content that reflects your specific restoration philosophy, past projects, and the historical significance of their properties.
Simply asking for “content about historic building restoration”, will produce generic advice. However, by providing context about your unique approach, previous successful restorations, and the Georgian architectural features you specialise in, you will receive content that genuinely reflects your expertise.
For effective context integration:
- Prioritise information the AI wouldn’t know from general training
- Include relevant data, statistics, or facts specific to your situation
- Reference past conversations or previously established parameters
- Provide British market specifics when they differ from global patterns
One approach we’ve found particularly effective is creating “context banks”—reusable blocks of information about your business, products, or industry that you can paste into prompts when needed. This ensures consistency across different AI interactions and saves considerable time.
Providing context about regional differences is particularly important. The property landscape in Liverpool city centre for example, will differ from North Wales, which differs from Runcorn – AI systems need this context to produce truly relevant outputs.
Pillar 6: Task Specifics
The final pillar focuses on precisely defining what success looks like for your AI interaction. This goes beyond basic task description to pinpoint exactly what you need.
I often compare this to the difference between asking someone to “get me something from the shop” versus providing a detailed shopping list with brands and quantities. The more specific you are about your end goal, the more likely you are to receive what you need.
Task specifics might include:
- The exact problem you’re trying to solve
- Success criteria for the output
- Audience details and their level of expertise
- Specific examples of what you consider good (or poor) outputs
- Step-by-step processes to follow
Rather than starting from scratch each time, you can build on previous instructions: “Using the same approach as the previous email campaign, but this time focusing on our sustainability initiatives and targeting our B2B customers in the Midlands.”
Practise makes perfect
Okay, now that we have our 6 pillars, let’s discuss implementation. Integrating all six pillars might seem daunting at first, but with practice, it becomes easier. To help, here’s a simple template:
- Role: [Specify who/what the AI should be]
- Task: [Clearly describe what needs to be done]
- Format: [Detail how the output should be structured]
- Audience: [Define specifically who you are addressing]
- Context: [Provide relevant background information]
- Specifics: [Outline exact requirements for success]
For different business purposes, you might emphasise different pillars. Content creation might lean heavily on role and format, while data analysis might prioritise task specifics and context.
Common implementation challenges include:
- Overloading prompts with too much information
- Contradicting instructions between pillars
- Not adapting the framework for different AI tools
- Failing to refine prompts based on results
To measure prompt effectiveness, I suggest tracking:
- Always, always, always, include enough editing time to personalise – and humanise – the AI output
- Consistency of outputs across multiple generations
- Stakeholder satisfaction with results
- Time saved compared to non-AI methods
Prompt engineering is iterative. Even with the Six-Pillar Framework, you’ll refine your approach over time as you learn what works best for your specific needs.
By implementing the Six-Pillar Framework – Role Assignment, Task Description, Output Format, Target Audience, External Context, and Task Specifics – you’ll gain a significant competitive advantage in how efficiently and effectively you utilise AI tools.
Prompt engineering will become a core business skill, much like email or presentation abilities became essential in previous decades. The businesses that master this skill early will be best positioned to leverage the next generation of AI advancements.
To help you implement these techniques immediately, we’ve created a comprehensive workbook that includes templates and exercises based on the Six-Pillar Framework.
Remember, effective AI interaction isn’t about technical expertise—it’s about clear communication. With the right prompting framework, you’ll unlock AI’s full potential and save yourself tons of time.
