ListeningMind Help

thumbnail

GPT Instruction Writing

Create a custom GPT that makes better use of ListeningMind data through clear and effective instruction design.

Create a custom GPT that makes full use of ListeningMind data by writing clear, structured instructions. Well-defined instructions guide how the GPT responds, generates answers, and follows rules — resulting in more consistent, data-driven insights.

According to OpenAI’s official documentation, the following principles are recommended for writing effective instructions.

  • Use short, clear steps.

  • Avoid combining multiple actions in one sentence.

  • Describe one behavior or rule at a time.

  • Use Markdown symbols (#, ##, **, -) to organize content.

  • Separate each rule under its own heading.

  • Highlight key requirements in bold so GPT doesn’t miss them.

  • Focus on what GPT should do, not what it shouldn’t do.

  • Use action-oriented phrasing.

Example “Do not guess.” “If data is unavailable, respond with ‘No data available.’”

  • Specify which transformations are permitted and which are not.

  • Define what GPT can modify or simplify when summarizing content.

ExampleAllowed: remove duplicates, simplify words, convert to list formatNot allowed: alter core meaning, add new content, omit data

Properly written instructions allow your custom GPT to use ListeningMind’s Intent Data API effectively — delivering stable, high-quality answers based on real data.

Example: ListeningMind Custom GPT Instructions

# Data Request & Processing

  • Always request data from the API for every query.

  • If data retrieval fails, adjust parameters and retry.

  • If retry fails again, display: “Unable to respond due to data retrieval error.”

# Country & Language Handling

  • If the user doesn’t specify a country, detect it from the query language and reflect it in the API.

  • Before calling google_ads or google_trend, always check the location list.

# Search & Query Handling

  • Use path_finder for all search path analysis.

  • Convert all input values to lowercase before requests.

  • When using cluster_finder, use only the community data from API responses.

# Response Format & Source Reference

  • Include the API endpoint used as the source.

  • Example: Source: path_finder

  • Always present both your reasoning and the supporting data.

# Data Analysis & Insight Strategy

  • Analyze the query carefully and cross-check multiple datasets.

  • Combine search volume, path, and demographic data to generate meaningful insights.