How to talk to AI: beginner guide to effective prompting

Have you ever asked ChatGPT a question only to receive a vague, unhelpful response? The problem likely wasn't with the AI—it was with how you asked the question.

Effective AI prompting is becoming an essential business skill. Much like learning how to craft effective Google searches transformed our ability to find information, mastering how to talk to AI systems can dramatically improve your productivity.

This guide will show you exactly how to communicate with AI systems to get the results you need—no technical background required. And while we're keeping this beginner-friendly, a quick note for the tech-savvy readers: we won't be diving into advanced techniques like few-shot prompting or retrieval-augmented generation. This is deliberately a foundations-first approach.

What are LLMs and why should business owners care?

LLM stands for Large Language Model—essentially, an AI system trained on vast amounts of text data. These systems learn patterns in language that allow them to predict what words should come next in almost any context.

The most well-known LLM is the OpenAI GPT (Generative Pre-trained Transformer), which powers ChatGPT. When you're "talking" to ChatGPT, you're actually giving prompts to an LLM that then generates text responses based on patterns it learned during training.

Other powerful LLMs include:

• Claude (Anthropic): Known for nuanced responses and longer context windows

• Gemini (Google): Excels at knowledge-intensive tasks

• Llama (Meta): Open-source models for customization

• Mistral (Mistral AI): Emerging open-source models with impressive performance

Why should small business owners care? Because even basic usage of these tools can save hours on writing tasks, generate creative ideas, analyze information, automate communication, and provide knowledgeable assistance outside your expertise.

But the key difference between mediocre and exceptional results lies in how you communicate with these systems.

The basics of effective prompting

Be specific and clear

LLMs can't read your mind. While they might seem intelligent, they're fundamentally prediction machines responding to the text you provide.

Poor prompt: "Give me marketing ideas."

Better prompt: "Give me 5 email marketing campaign ideas for a small gardening supply store trying to increase summer sales to existing customers. Each idea should include a subject line example and main offering."

The second prompt provides specific details about the number of ideas needed, business type, target audience, marketing channel, and what elements each idea should include.

Provide context for better answers

LLMs have no knowledge of your specific business unless you tell them.

Poor prompt: "Write me a customer email."

Better prompt: "I run a boutique accounting firm serving small restaurants. Write an email announcing our new quarterly tax preparation package at £599 per quarter. Emphasise our restaurant industry specialisation and include a call-to-action."

The context helps the AI understand your business, audience, offering, and desired outcome.

Use the right tone and format

Tell the AI how you want the response structured and what tone to use.

Poor prompt: "Write social media content."

Better prompt: "Create 3 LinkedIn posts for my IT consulting business. Each should be around 150 words, use a professional but conversational tone, include 2-3 hashtags, and end with a question. Format each post as a numbered section."

This specifies the platform, number of outputs, length, tone, structural elements, and format.

Learn from AI responses and iterate

If you don't get what you need, refine your prompt based on the response.

Initial prompt: "Write a customer service script for handling complaints."

Refined prompt: "That's too generic. Please rewrite specifically for an online clothing retailer handling size exchange requests. Include specific questions about the garment and clear exchange steps."

Prompt structures that get results

The task-context-example framework

This three-part structure works wonderfully for most business scenarios:

Example:

• Task: Create a professional email response to a customer asking about our return policy.

• Context: We're a small online jewellery store. Our policy allows returns within 30 days with receipt. Custom pieces are non-returnable.

• Example style: Keep the tone warm but professional, no more than 100 words, and mention our spring collection at the end.

Role-playing prompts for specialised knowledge

Assigning a role to the AI can dramatically improve responses for specialised content.

Example:

"Act as an experienced HR consultant specialising in small businesses. Review this draft employee handbook section on remote work policies and suggest 3-5 improvements that would clarify expectations and protect the business legally."

Chain-of-thought prompting for complex problems

For analysis or problem-solving, guide the AI to show its reasoning:

Example:

"I'm trying to decide whether to invest in a customer loyalty app for my cafe."

Please analyse this decision step by step, considering:

• Initial costs and ongoing fees

• Typical adoption rates for small cafes

• Potential impact on repeat business

• Administrative overhead

• Alternatives to a full loyalty app

These prompting techniques work across various AI tools designed for non-technical users, making them valuable skills regardless of which platforms you choose.

Finding it challenging to implement these strategies in your business? Book a free 15-min consultation to find out how to apply these techniques to your specific business needs.

Common prompting mistakes small businesses make

Being too vague

The most common mistake is providing insufficient detail, resulting in generic responses.

Vague prompt: "Give me content ideas."

Specific prompt: "Suggest 5 blog topic ideas about sustainable packaging that would interest small food product manufacturers concerned about environmental impact while managing costs."

Not providing enough context

Without background information, the AI can't tailor responses to your situation.

Missing context: "Write an email to clients about our new service."

With context: "Write an email introducing our virtual bookkeeping service to existing clients who currently use our tax preparation services. These are mainly local service businesses with 5-15 employees."

Expecting the impossible

LLMs have limitations. They can't access real-time data, browse the web independently, access your private business data, or perform actions outside the conversation.

Not iterating and refining

Effective prompting is a conversation. If the first response isn't quite right, provide feedback instead of starting over completely.

Real-world prompting examples

Customer service responses

Effective prompt:

"My customer wrote: 'I've been waiting 7 days for my order and still don't have a shipping confirmation. This is ridiculous!'

Write a response that acknowledges their frustration, apologises sincerely, explains shipping delays due to supplier issues, offers a 15% discount, and provides a resolution timeline (48 hours)."

Marketing content creation

Effective prompt:

"Create an Instagram caption for eco-friendly cleaning products that are 100% biodegradable with plastic-free packaging. Price is 20% higher than conventional cleaners. Target audience is environmentally-conscious millennials. We're donating 5% to ocean cleanup. Include 3 hashtags and invite comments."

Effective prompting is essential when implementing broader business automation strategies, as clear instructions are the foundation of any successful workflow.

Advanced techniques to level up your AI interactions

Using system prompts

Some AI platforms allow you to set "system" prompts that define how the AI should behave throughout the conversation.

Example system prompt:

"You are a knowledgeable but accessible business advisor specialising in small retail businesses. You provide practical, actionable advice without jargon. When suggesting solutions, you consider implementation costs and complexity."

Building prompt libraries

Create a document of your most effective prompts for tasks you perform regularly, such as email templates, social media content, meeting preparation, and competitor research.

Combining AI tools

For more complex workflows, learn to combine AI tools—use one for brainstorming, another for refinement, and yet another for visuals. This approach can significantly improve your business scaling efforts by leveraging the strengths of different systems.

Start your AI prompting journey

Effective communication with AI isn't just a technical skill—it's becoming a fundamental business advantage. The difference between basic and advanced prompting can mean hours of productivity gained or lost each week.

Start by applying the principles in this guide to one specific area of your business. Pay attention to what works, refine your approach, and gradually expand to other areas. As you gain experience, you can incorporate these skills into a comprehensive approach to AI adoption for your business.

Remember that even AI experts constantly refine their techniques. The goal isn't perfection but improvement—each interaction is an opportunity to learn and get better results.

Need more help mastering AI prompting for your specific business challenges? I'm available for a free 15-minute consultation to help you develop custom prompt strategies that deliver immediate results.