Prompt of the Day: AI Data Detective -- Spot Patterns, Trends, and Hidden Insights in Your Data
You open an Excel spreadsheet with sales figures from the last six months. 2,000 rows, 15 columns. You scroll down, scroll back up, sort a column, undo it. After 20 minutes, you know exactly as much as before: nothing.
The problem: Most people can collect data but cannot read it. You have access to more data than ever -- sales figures, website statistics, survey results, project time logs, expenses, customer feedback. But a spreadsheet full of numbers does not tell a story on its own. You need someone who asks the right questions and makes the answers understandable.
The real problem: Data analysis feels like something for specialists. For people who studied statistics and write Python scripts. But 80% of everyday data analysis does not require regression or machine learning. It needs someone who says: 'Here is the pattern. Here is the outlier. And here is what you should do about it.'
The solution: AI can be your data analyst. Give it your data (as a table, CSV, or simply typed out), describe the context -- and it finds patterns you overlooked, explains relationships in plain language, and gives you concrete recommendations.
What you can use this for:
- Business: Analyze sales figures, understand customer behavior, spot revenue trends
- Marketing: Compare campaign performance, identify best channels, evaluate conversion rates
- Finance: Recognize spending patterns, find budget deviations, uncover savings potential
- HR: Evaluate employee satisfaction, identify turnover risks, assess recruiting channels
- Personal: Analyze household budget, understand energy consumption, evaluate fitness data
- Project management: Analyze time spent, identify bottlenecks, spot planning deviations
How it works:
1. Export your data as CSV or copy the table directly into the chat
2. Fill in the prompt -- the more context you provide, the better the insights
3. Read the analysis and ask follow-up questions about the points that surprise you most
Important: Do not upload confidential data when using a cloud AI service. Anonymize sensitive information (names, addresses, IDs) beforehand. For confidential business data, check whether your AI tool uses data for training -- and use the API or a local solution if needed.
Pro tips:
- Test hypotheses: 'I suspect that [hypothesis]. Check this against my data. How strong is the correlation?'
- Segmentation: 'Split my data into meaningful groups (e.g., by region, time period, customer type). Where are the biggest differences between groups?'
- Forecasting: 'Based on current trends -- how will the numbers develop over the next 3 months? What assumptions does this forecast rely on?'
- Anomaly deep-dive: 'You found an outlier. What possible explanations are there? What should I check first?'
- Visualization recommendation: 'Which 3 charts would best display the key findings? Describe them so I can recreate them in Excel or Google Sheets.'
- Comparison: 'Compare the data from [period A] with [period B]. What changed significantly? What stayed the same?'
- Root cause analysis: 'The value in [month/category] is unusually high/low. What could be the cause? What external factors should I check?'
You are an experienced data analyst who makes complex datasets understandable -- for people who are not statistics experts. You find patterns others overlook, explain relationships in plain language, and always provide concrete recommendations for action.
**My data:**
[Paste your data here: as a table, CSV, or describe the data and its structure. The more context, the better the analysis.]
**Context:**
- What the data shows: [e.g., 'Monthly sales figures by product and region', 'Website traffic for the last 12 months', 'Results of a customer satisfaction survey with 200 respondents', 'My household budget for the last 6 months']
- My goal: [e.g., 'Understand why sales dropped in March', 'Find out which marketing channels work best', 'Find savings potential in my household budget']
- What I am most interested in: [e.g., 'Seasonal patterns', 'Differences between regions', 'Which customers are most profitable', 'Where most money is being wasted']
- Time period: [e.g., 'January to June 2026', 'The last 3 years', 'Calendar weeks 20 to 25']
---
Analyze my data in 5 steps:
**Step 1: Data Overview**
- How many data points do I have? What variables are there?
- Are there missing values or obvious data errors?
- Summary of key metrics: mean, median, minimum, maximum -- explained in plain language ('The typical value is X, but there are spikes up to Y')
- Data quality check: Can I trust the data, or are there red flags?
**Step 2: The 3 Most Important Patterns**
Find the three most notable patterns in my data. For each pattern:
- What is the pattern? (In one clear sentence)
- How strong is it? (In percentages or concrete numbers from my data)
- What could be the cause? (At least 2 possible explanations)
- Why is this relevant to my goal?
**Step 3: Outliers and Anomalies**
- Which data points deviate significantly from the rest?
- Are these errors, special cases, or important signals?
- For each outlier: What should I check first?
- Are there suspicious patterns (e.g., suspiciously round numbers, identical values, implausible jumps)?
**Step 4: Relationships and Correlations**
- Which variables are connected? (Does A increase when B increases?)
- How strong is the relationship? (Strong, moderate, weak)
- Which relationships are surprising or unexpected?
- **Important:** Clearly separate correlation from causation. Explain the difference using a concrete example from my data.
**Step 5: Recommendations for Action**
Based on the entire analysis:
- **Do now:** 1-2 concrete actions I can take based on this data
- **Investigate further:** 1-2 areas where I need to collect more data or dig deeper
- **Monitor long-term:** 1-2 trends I should continue tracking over the coming months
- **Visualization:** Which 2-3 charts should I create to present the findings? (Name the chart type, what goes on which axis, and what the chart shows)
- **The one number:** If you could only tell me ONE number from the entire analysis that I should remember -- which would it be and why?
**Rules:**
- Explain everything so someone without a statistics background can understand it
- If you use technical terms, explain them immediately in parentheses
- Clearly distinguish between what the data shows (facts) and what it might mean (interpretation)
- If the data volume is insufficient for a reliable conclusion, say so honestly
- Always cite concrete numbers from my data as evidence
- If I have not provided important information, ask -- do not guess
- Avoid false precision: 'about 30%' is more honest than '29.7%' when the data basis is small