How many rows of data can I include?

GPT-4o and Claude handle roughly 50-200 rows of CSV comfortably. For larger datasets, aggregate the data first (e.g., weekly totals instead of daily rows) or use the SQL generation prompt to query the data before analysis.

Can the AI run calculations on the data?

Yes, but verify all computed numbers. LLMs can make arithmetic errors on large datasets. For critical calculations, ask the model to show its working or use Code Interpreter in ChatGPT to have Python run the calculations.

Can I use this for A/B test results?

Yes. Include the sample sizes and conversion rates for each variant and add to the question: "Is the difference statistically significant? Assume 95% confidence threshold." The AI will walk through the statistical test.

How many rows of data can I include?

GPT-4o and Claude handle roughly 50-200 rows of CSV comfortably. For larger datasets, aggregate the data first (e.g., weekly totals instead of daily rows) or use the SQL generation prompt to query the data before analysis.

Can the AI run calculations on the data?

Yes, but verify all computed numbers. LLMs can make arithmetic errors on large datasets. For critical calculations, ask the model to show its working or use Code Interpreter in ChatGPT to have Python run the calculations.

Can I use this for A/B test results?

Yes. Include the sample sizes and conversion rates for each variant and add to the question: "Is the difference statistically significant? Assume 95% confidence threshold." The AI will walk through the statistical test.

dataOpenai

Data Analysis Prompt (ChatGPT)

Most AI data analysis prompts produce vague observations like "sales increased in Q2". This prompt forces the model to answer a specific business question, cite supporting numbers, and distinguish findings from recommendations. The Limitations section is critical — it prevents overconfident conclusions from small or incomplete datasets. This variant is formatted for ChatGPT: Optimised for GPT-4o and GPT-4 Turbo. Uses markdown formatting and system/user message separation.

Prompt Template

## System
You are an expert AI assistant. Respond using clear markdown formatting.

## User
You are a senior data analyst. Analyse the following dataset and produce a structured report.

Dataset description: {{description}}
Business question to answer: {{question}}
Time period: {{time_period}}
Key metrics to focus on: {{metrics}}

Dataset ({{format}}):
{{data}}

Report structure:
1. **Executive Summary** (3 bullet points: key finding, trend, recommendation)
2. **Key Findings** — numbered list of the 3-5 most important insights with supporting numbers
3. **Anomalies** — any unexpected patterns, outliers, or data quality issues
4. **Recommendations** — 2-3 actionable next steps based on the data
5. **Limitations** — what the data cannot answer and what additional data would help

Variables

{{description}}What the dataset contains (source, columns, granularity)

{{question}}The specific business question you need to answer

{{time_period}}Date range of the data, e.g., "January to March 2024"

{{metrics}}The KPIs or columns that matter most for your analysis

{{format}}Data format: CSV, JSON, table, SQL result set

{{data}}The dataset itself (pasted as text)

Example

Input

description: Monthly SaaS subscription revenue by plan type (monthly data)
question: Which plan has the highest growth rate and is at risk of churning?
time_period: January to June 2024
metrics: MRR, churn rate, new subscriptions
format: CSV
data:
month,plan,mrr,new_subs,churned
2024-01,starter,12000,45,8
2024-01,pro,38000,20,3
2024-02,starter,13200,52,7
2024-02,pro,39400,22,4
...

Output

**Executive Summary**
- Starter plan MRR grew 38% in 6 months vs. 12% for Pro
- Starter churn rate increased from 6.7% to 9.2% in May-June, a warning sign
- Recommendation: investigate Starter-to-Pro upgrade friction to convert high-growth users before they churn

**Key Findings**
1. Starter plan added the most absolute MRR ($4,600) but its churn rate rose 37% in the last 2 months
2. Pro plan churn is stable at 2-3 churns/month with consistent 20-22 new subscriptions...

FAQ

How many rows of data can I include?: GPT-4o and Claude handle roughly 50-200 rows of CSV comfortably. For larger datasets, aggregate the data first (e.g., weekly totals instead of daily rows) or use the SQL generation prompt to query the data before analysis.
Can the AI run calculations on the data?: Yes, but verify all computed numbers. LLMs can make arithmetic errors on large datasets. For critical calculations, ask the model to show its working or use Code Interpreter in ChatGPT to have Python run the calculations.
Can I use this for A/B test results?: Yes. Include the sample sizes and conversion rates for each variant and add to the question: "Is the difference statistically significant? Assume 95% confidence threshold." The AI will walk through the statistical test.

Related Prompts

Data Analysis Prompt

Most AI data analysis prompts produce vague observations like "sales increased in Q2". Thi...

Data Analysis Prompt (Claude)

Most AI data analysis prompts produce vague observations like "sales increased in Q2". Thi...

Data Analysis Prompt (Gemini)

Most AI data analysis prompts produce vague observations like "sales increased in Q2". Thi...

Data Analysis Prompt (LLaMA / Ollama)

Most AI data analysis prompts produce vague observations like "sales increased in Q2". Thi...

SQL Query Generation Prompt

Generating SQL without schema context produces queries that reference non-existent columns...

CSV Data Processing Prompt

CSV processing tasks involve numerous small decisions about null handling, column types, a...

Data Analysis Prompt (ChatGPT)

Variables

Example

Related Tools

FAQ

Related Prompts