Prompt Generator for Data Analysts – PromptByJob
Prompt Builder

Prompt Generator for Data Analysts

Describe your analysis task and this page builds structured prompts you can paste into an AI assistant of your choice. It does not run an AI model itself.

Build your prompt

Fill in the task, your tools, and your question. The page assembles three different prompts in your browser: a task prompt, a data-quality and assumptions review to run first, and a stakeholder write-up prompt for when you have results. Paste the one you need into ChatGPT, Claude, Gemini, Copilot or another assistant. Fields marked * are required.

Privacy: Do not enter real customer, patient, or employee records, personal data, credentials, access tokens, or confidential financial figures. Describe table structures and use made-up or aggregated values.

Changes how much explanation the prompt asks for.

Each task requests a different deliverable and different checks.

Used to request the right language, dialect, or tool steps.

Example: why did monthly churn rise in a fictional subscription product, and which segments drive it?

Do not paste real personal or confidential data here.

Table and column names, units, date range, known quirks. Names only; no real rows.

Use anonymised or synthetic schema details. No credentials or connection strings.

What will this analysis inform? Example: decide whether to change pricing for one plan.

Aggregated, non-identifying numbers only. If blank, the stakeholder prompt tells the AI to ask you for results rather than invent them.

Constraints (optional, select all that apply)

Worked example (illustrative)

This example uses fictional inputs and the same prompt-building function as the live tool. It shows how prompt construction works. It does not show an AI response or any verified result.

Example inputs

    Prompt generated from those inputs (task prompt)

    
  
    How the inputs shaped the prompt
    
  
      
    
    • Task: choosing SQL query writing adds query-specific deliverables (dialect, comments, edge cases, a check against a query plan) that other tasks do not include.
      • 
    
    • Data description: the listed tables and columns are marked as the only names the AI may use, which discourages invented columns.
      • 
    
    • Experience level: a mid-level setting asks for brief reasoning rather than beginner tutorials.
      • 
    
    • Constraints: each ticked constraint appears as an explicit instruction, and the privacy constraint asks the AI not to claim legal compliance.
      • 
  
    

    


    
  
    How it works
    
  
      
    
    • Your selections are combined into text by JavaScript on this page. No AI service is called by the tool.
      • 
    
    • The three prompts differ in purpose: the task prompt asks for the main deliverable, the review prompt asks for data-quality checks and open questions before you analyse, and the write-up prompt turns your results into audience-appropriate wording.
      • 
    
    • Every prompt tells the AI to use only the facts you supplied, to label assumptions, and to say how you can verify its output.
      • 
    
    • Prompts are templates written for this tool. They have not been individually tested for every combination of inputs.
      • 
  
    

    


    
  
    Limitations and privacy guidance
    
  
      
    
    • Generated prompts are starting points. AI responses can contain errors, including wrong SQL, wrong statistics, or invented figures. Run code on sample data and review results before relying on them.
      • 
    
    • Output is not professional advice. For healthcare, finance, HR, or risk work, a qualified person must review anything that affects decisions about people or money, and nothing here confirms legal or regulatory compliance.
      • 
    
    • Before using any AI tool with work data, check your organisation’s data policy. Prefer anonymised schemas and aggregated or synthetic values.
      • 
    
    • The prompt text you type is processed in your browser by this page’s script. This page does not store it in browser storage. This statement covers only this tool’s code, not other scripts, analytics, or cookies on the wider website.
      • 
  
    

    


    
  
    Frequently asked questions
    
  
    Does this tool analyse my data or answer my question?
    No. It only builds a prompt. You paste it into an external AI assistant, which produces the answer.
    
  
    What should I not paste into a prompt?
    Real personal data, customer or employee records, credentials, API keys, and confidential financial figures. Use table and column names, aggregated numbers, or fictional values.
    
  
    Why does the tool ask for table and column names?
    When names are supplied, the prompt tells the AI to use only those. Without them, assistants tend to guess, and the guesses can look plausible but be wrong.
    
  
    Can I use the prompts for SQL optimisation?
    Yes: choose SQL Query Writing & Optimisation and describe the engine, table sizes, and indexes in the data description. Always check suggested rewrites with your database’s query plan (for example EXPLAIN) before production use.
    
  
    When should I use each of the three prompts?
    Run the data-quality review before analysis, the task prompt to produce the main deliverable, and the stakeholder prompt once you have verified results you can share.
    
  
    Which AI assistant should I use?
    The prompts are written to be assistant-neutral. Try more than one and compare outputs against your own checks.
    
  Back to the form ↑

    

    

    



	
	

    

	
	

    

			
    
			
		
	
    


	
     
	
    

	
    



    
	
                
                	Scroll to Top