Writing Effective Prompts

Writing Effective Prompts

In Bourd, a prompt is not an instruction; it is a question you ask an AI model to simulate a real user query. The quality of your competitive intelligence is directly tied to the quality of these questions. Well-crafted prompts reveal how AI models perceive, rank, and recommend brands in your market.

This guide covers the principles and frameworks for writing effective prompts for Answer Engine Optimization (AEO).

The Core Principle: Simulate Real User Questions

The fundamental goal is to ask questions the way your potential customers would. You are not instructing a monitoring agent; you are querying a large language model to see how it responds to a user seeking advice or recommendations.

Think like a customer, not a marketer.

• Bad Prompt (Instructional): Monitor Asana and Monday.com for new AI features.
  • This is an instruction for a web scraper, not a question for an LLM.
• Good Prompt (User Query): Which project management tools have the best AI features?
  • This simulates a real user query and will reveal which brands the AI associates with advanced AI capabilities.

Strategic Prompt Framework

To gather comprehensive intelligence, structure your prompts around these four strategic categories.

1. “Best of” Prompts

These are broad, category-defining questions designed to identify the market leaders and your brand’s overall ranking within a specific category.

Goal: Understand your position in the consideration set.

Examples:

• “What are the best CRM platforms for a small business?”
• “List the top 5 project management tools for software teams.”
• “What is the most popular accounting software for freelancers?”
• “Which companies make the most reliable running shoes?“

2. Comparative Prompts

These prompts directly compare a few specific brands. They are useful for understanding how AI models differentiate you from your closest competitors on key features or attributes.

Goal: Analyze your positioning against specific competitors.

Examples:

• “Compare HubSpot and Salesforce for enterprise marketing teams.”
• “What are the differences between Asana, Trello, and Monday.com?”
• “Stripe vs. Adyen: which is better for international payments?”
• “Oura Ring vs. Whoop: which is more accurate for sleep tracking?“

3. Problem/Solution Prompts

These prompts describe a user’s problem and ask for a solution. They are highly effective at revealing which brands are associated with solving specific pain points.

Goal: Discover if your brand is recommended as a solution to customer problems.

Examples:

• “What’s the best way to automate marketing emails for an e-commerce store?”
• “How can I track my team’s tasks if we all work remotely?”
• “I need software to help me manage my business finances. What should I use?”
• “My team finds Jira too complex. What’s a simpler tool for tracking software bugs?“

4. “Alternatives to” Prompts

These prompts target users who are looking to switch from a competitor. They are crucial for capturing high-intent users who are actively looking for a new solution.

Goal: See if your brand is positioned as a viable alternative to a market leader or competitor.

Examples:

• “What are the best alternatives to Salesforce?”
• “What are some cheaper alternatives to Adobe Photoshop?”
• “I’m looking for a project management tool that’s more visual than Jira.”
• “Which CRMs are easier to use than HubSpot?”

Common Mistakes to Avoid

1. Giving Instructions Instead of Asking Questions.

   • Mistake: Analyze the top 5 CRMs.
   • Correction: What are the top 5 CRM platforms?

2. Including Your Own Brand Name.

   • Mistake: Is Acme CRM a good tool for small businesses?
   • Why it’s a mistake: This biases the AI and doesn’t tell you if your brand would be recommended organically. The goal is to see if the model mentions you without being prompted.
   • Correction: What are some good CRM tools for small businesses?

3. Being Too Broad.

   • Mistake: What is the best software?
   • Correction: What is the best accounting software for a freelance graphic designer?

4. Asking Yes/No Questions.

   • Mistake: Is Asana good for remote teams?
   • Correction: What are the strengths and weaknesses of Asana for remote teams?

Testing and Iteration

The key to a powerful prompt library is continuous refinement.

1. Run Manually First: Before adding a new prompt to a schedule, run it manually across a few different models.
2. Analyze the Response: Did the AI understand the question? Did it mention relevant brands? Was the response high-quality?
3. Iterate: If the results are poor, refine the prompt. Try making it more specific or rephrasing it to better match a real user’s language.
4. Schedule for Tracking: Once a prompt consistently delivers valuable intelligence, add it to a schedule to track trends over time.

Next Steps

• Run Your First Analysis
• Set Up Automated Schedules
• View Results and Analytics