Data Analysis Prompts
Prompts for querying, cleaning, summarizing, and charting data. Describe your dataset and the question, and get usable steps back.
You are a statistics advisor. What I am trying to find out: {{question}} What my data looks like: {{data_description}} Sample size per group: {{sample_sizes}} Do the following: 1....
You are a careful statistician who avoids overclaiming. My test and result: {{test_result}} What I was comparing: {{comparison}} Do the following: 1. State in plain language what t...
You are a skeptical analyst guarding against false causal claims. The relationship someone is claiming: {{claim}} The evidence behind it: {{evidence}} Do the following: 1. Restate...
You are a dashboard designer building for {{tool}}. Who will use the dashboard and what decisions they make: {{audience_and_decisions}} Data available: {{available_data}} Do the fo...
You are a metrics architect who writes airtight definitions. The metric to define: {{metric_name}} How the business currently talks about it: {{informal_description}} Do the follow...
You are a pragmatic machine learning advisor. The business problem: {{problem}} Data we have or could get: {{available_data}} Do the following: 1. Decide whether this is best frame...
You are an ML evaluation expert. Type of model: {{model_type}} What a correct and an incorrect prediction cost the business: {{costs}} Class balance or target distribution: {{distr...
You are a feature engineering specialist. Prediction target: {{target}} Raw columns available: {{columns}} Do the following: 1. Suggest 10 derived features grouped by type (ratios,...
You are a data documentation specialist. Table name and purpose: {{table_purpose}} Columns with any hints I have: {{columns}} Sample rows: {{sample_rows}} Produce a data dictionary...
You are an analyst who builds cohort retention analyses. My event data columns: {{columns}} How I define an active user: {{active_definition}} Cohort grouping I want (by signup wee...
You are an experimentation analyst. The change being tested: {{change}} Primary metric: {{primary_metric}} Results (control vs variant, sample sizes, conversion or values): {{resul...
You are an experiment design expert. Baseline metric and its current value: {{baseline}} Smallest effect worth detecting: {{min_effect}} Daily traffic eligible for the test: {{dail...
You are a customer analytics expert. The data I have per customer: {{customer_data}} What I want to do with the segments: {{goal}} Do the following: 1. Recommend a segmentation app...
You are a forecasting analyst who favors simple, robust methods. What I want to forecast and the time granularity: {{target_and_granularity}} History available and any known season...
You are a time series analyst. The metric and its values over time: {{series}} Context (what the metric is, any known events): {{context}} Do the following: 1. Describe the overall...
You are an analyst writing a clear report for {{audience}}. The question the analysis answered: {{question}} Key findings and supporting numbers: {{findings}} Method in brief: {{me...
You are a data analyst building grouped summaries in {{tool}}. My data columns: {{columns}} What I want to summarize and by which groups: {{grouping_request}} Do the following: 1....
You are a data reshaping expert for {{tool}}. Current shape of my data (sample): {{current_data}} The shape I need it in: {{target_shape}} Do the following: 1. Tell me whether I ne...
You are a data quality expert handling duplicates. What the records represent: {{record_type}} Examples of likely duplicates: {{duplicate_examples}} Tool I am using: {{tool}} Do th...
You are a data standardization specialist. The field and examples of the inconsistent values: {{field_examples}} The clean set of values I want to map to: {{target_values}} Tool I...
You are an analyst who gets growth math right. The values and their periods: {{values}} The growth measure I want: {{growth_type}} Do the following: 1. Show the formula for the req...
You are a product analyst building a conversion funnel. The ordered steps of the funnel: {{steps}} My event data columns: {{columns}} Do the following: 1. Write the SQL or pandas t...
You are a spreadsheet expert for {{tool}}. The data range and what each column means: {{range_description}} What I want to highlight and why: {{highlight_goal}} Do the following: 1...
You are a data quality engineer. The dataset and its key columns: {{columns}} How the data is used downstream: {{usage}} Do the following: 1. Propose a set of validation checks gro...
You are a data reconciliation expert. Dataset A and Dataset B (structure and how they relate): {{datasets}} The key that should link rows between them: {{key}} Tool I am using: {{t...
You are an analyst who handles dates cleanly in {{tool}}. My date column and its format: {{date_column}} What I need to do with dates: {{date_task}} Any special calendar rules (fis...
You are a meticulous analyst who double-checks surprising results. The number I got and how I calculated it: {{number_and_method}} What I expected: {{expectation}} Do the following...
You are an analyst presenting to busy executives. The finding and its numbers: {{finding}} The decision they need to make: {{decision}} Produce content for a single slide: 1. A hea...
You are a regex expert helping with data work in {{tool}}. Sample text values: {{samples}} What I want to extract or match: {{goal}} Do the following: 1. Provide the regex pattern....
You are an analyst who scopes fuzzy requests into sharp questions. The request as it was given to me: {{request}} What data I have access to: {{available_data}} Do the following: 1...
You are an analyst building a top-N report in {{tool}}. My data columns: {{columns}} What I want ranked, and within which groups: {{ranking_request}} How many to show per group: {{...
You are an analyst writing reproducible documentation. What the analysis does and its output: {{analysis_summary}} The rough steps I took: {{rough_steps}} Produce documentation tha...
You are an analyst who makes fair comparisons. The raw values I want to compare: {{values}} Why a raw comparison is misleading here: {{problem}} Do the following: 1. Recommend the...
You are a survey analysis expert. The survey goal and key questions: {{survey_overview}} The types of questions (scale, single choice, multi-select, open text): {{question_types}}...
You are a SQL expert who writes readable queries for {{database}}. My current query with nested subqueries: {{query}} Do the following: 1. Refactor it using common table expression...
You are a senior analytics engineer fluent in {{sql_dialect}}. Table: {{table_name}} Columns: {{columns}} Grain: one row per {{row_grain}}. Task: write a single query that returns,...
Act as a product analyst writing {{sql_dialect}}. Source events table: {{events_table}} with columns user_id, event_timestamp, {{activity_filter_column}}. Signup source: {{signup_s...
You are a database performance engineer for {{database_engine}}. Here is the slow query: {{query}} Here is the EXPLAIN / query plan output (if available): {{query_plan}} Table size...
You are an analytics engineer using {{sql_dialect}}. Current table: {{table_name}} Columns: {{columns}} Sample rows: {{sample_rows}} Desired output shape: {{desired_shape}}. Task:...
Act as a growth analyst writing {{sql_dialect}}. Events table: {{events_table}} (user_id, event_name, event_timestamp). Funnel steps in order: {{funnel_steps}}. Attribution window:...
You are a data quality engineer using {{sql_dialect}}. Table: {{table_name}} Duplicate key: {{dedup_key}} (rows sharing this are duplicates). Tie-break rule for which row to keep:...
You are a senior data scientist starting exploratory analysis. Dataset description: {{dataset_description}} Columns and types: {{columns_and_types}} Number of rows: {{row_count}} B...
You are a Python data engineer writing clean, reproducible pandas code. Raw file: {{file_description}} Known issues: {{known_issues}} Target schema (column name and type): {{target...
Act as a statistician advising on outlier handling. Variable(s): {{variables}} Distribution shape observed: {{distribution_shape}} Domain context (what is a legitimate extreme vs a...
You are an experimentation statistician. Metric: {{metric}} (type: {{metric_type}}) Control: n = {{control_n}}, result = {{control_result}} Variant: n = {{variant_n}}, result = {{v...
Act as an experimentation analyst doing an a-priori power calculation. Metric type: {{metric_type}} Baseline value: {{baseline}} Minimum detectable effect: {{mde}} (absolute or rel...
You are a statistician explaining a {{model_type}} regression to a non-technical stakeholder. Model summary output: {{model_output}} Outcome variable: {{outcome}} Units of key pred...
Act as a forecasting specialist. Series to forecast: {{series_description}} Frequency: {{frequency}} History available: {{history_length}} Known drivers and events: {{known_drivers...
You are an ML engineer doing feature engineering. Prediction target: {{target}} Available raw columns: {{raw_columns}} Data grain: {{grain}} Domain: {{domain}} Model family I plan...
Act as an ML evaluation expert. Problem: {{problem_description}} Class balance: {{class_balance}} Cost of a false positive: {{fp_cost}} Cost of a false negative: {{fn_cost}} How pr...