Prompting Guide

This page gives guidance on how to best prompt the Linkup API and get optimal results.

​

Core Elements of an Effective Prompt

1. Instruction Component

   • Start with a description of what you desire
   • Specify the task or outcome as precisely as possible
   • Example: “Analyze the market position of [Company]” vs “Company market info”

2. Context Framework

   • Provide relevant background information
   • Include time frames if applicable
   • Specify industry or domain context
   • Example: “As a fintech startup founded in 2021 with 50 employees…”

3. Input Parameters

   • List all the required data points if you know them already
   • Specify format requirements if needed
   • Include any constraints you might have
   • Example: “[Full Name] (required), [Company] (required), [Time Period] (optional)”

4. Output Specifications

   • Define desired format
   • Specify length requirements
   • Include any structural preferences
   • Example: “Format as a 2-paragraph biography with bullet points for key achievements”

​

Prompt Types & Examples

1. Research & Analysis

Bad: "Tell me about AI companies"
Good: "Describe the #1 AI company in France by revenue for 2024, focusing on their main products and recent partnerships"

2. Content Generation

Bad: "Write a bio"
Good: "Write a professional biography for [Name] highlighting their experience in [Industry], current role at [Company], and key achievements from the past 3 years"

3. Sales

Bad: "Help with sales meeting"
Good: "Create a discovery call agenda for a meeting with [Prospect Name] from [Company], focusing on their AI integration needs and current tech stack challenges"

​

Best Practices

1. Clarity & Precision

   • Use specific, actionable terms
   • Names can often point to different companies and persons. Specify with industry, or share a website url to avoid any ambiguity.
   • Include measurable parameters and define scope clearly
   • Example: “Analyze last 6 months of data” vs “Look at recent data”

2. Structure & Organization

   • Break complex requests into sections
   • Use numbered lists for sequential tasks
   • Include headers for different components
   • Example: “1. Company Analysis, 2. Contact Research, 3. Action Items”

3. Context Enhancement

   • The system does not have your unspoken context
   • Include relevant constraints
   • Provide industry context and specify target audience
   • Don’t hesitate to prepend a context prompt to set the stage
   • Example: “For a technical audience familiar with AI deployment in enterprise settings…”

​

Common Pitfalls to Avoid

1. Vagueness

Avoid: "Research the company"
Better: "Analyze [Company]'s market position, focusing on their AI products, recent partnerships, and competitive advantages in the French market"

2. Information Overload

Avoid: "Research the company, write a bio, prepare sales materials, and analyze competitors all at once"
Better: Break into separate, focused prompts for each task

3. Missing Context

Avoid: "What are their pain points?"
Better: "Based on [Company]'s size (50 employees), industry (AI/SaaS), and recent Series A funding, identify likely operational challenges in their customer support scaling"

4. Name Colision

Avoid: "What is the birth date of Roosevelt?"
Better: "What is the birth date of Franklin Delano Roosevelt?"

​

Iterative Refinement

1. Testing Strategy

   • Start with a basic prompt
   • Analyze response quality
   • Add specific parameters
   • Refine based on results

2. Improvement Cycle

   • Note missing information
   • Add relevant constraints
   • Clarify ambiguous points
   • Test revised version