datapeeker

Summarize interview transcripts or field notes, extract key quotes, and identify initial observations to prevent context pollution in the main analysis session.

Analyze free text column values and propose categorical groupings with complete value-to-category mappings

Identify exact duplicate rows in a table, return groups with counts and examples

Identify foreign key relationships using heuristics, value overlap, cardinality analysis, and referential integrity validation

Find near-duplicate records using fuzzy string matching on key text columns

Detect statistical outliers in numeric columns using MAD (Median Absolute Deviation) with 3 MAD threshold

Find the most illustrative verbatim quotes for each theme to support findings in Phase 6 reporting.

Analyze data segments and suggest initial codes with definitions and examples to accelerate codebook development in Phase 4.

Group codes into potential themes and extract supporting data extracts to accelerate theme development in Phase 5.

Independently code a sample of data (10-20%) to verify coding consistency and catch systematic bias in code application. This is MANDATORY in Phase 4 before proceeding to Phase 5.

Validate marketing concepts via internet research - market demand signals, similar solutions, audience needs, and validation evidence

Analyze raw_* table for quality issues - NULL percentages, duplicates, outliers (3 MAD), free text candidates

**CRITICAL BIAS PREVENTION:** Systematically search for data that contradicts proposed themes to prevent confirmation bias. This is MANDATORY in Phase 5 for EVERY theme.