What problem does it solve?
Building box plot visualizations in PostHog requires distribution data with correct percentile statistics, and mistakes like using pre-aggregated averages or misconfigured chart settings produce invalid or misleading insights. This Skill guides the creation of box plot insights end to end, from choosing the right query type to validating and verifying the saved result.
Core Features & Use Cases
- Query type selection: Chooses between a standard Trends box plot (event or action with a numeric property) and a SQL-backed box plot for custom grouping, joins, or derived values.
- SQL validation rules: Enforces required statistics (min, p25, median, mean, p75, max), ordering checks, and limits of 200 series and 10,000 cells before saving.
- Save and verify workflow: Saves the insight with the correct DataVisualizationNode chart settings, then reads it back and re-runs it to confirm the expected columns and one row per box.
- Use Case: A user asks to compare request latency distributions across pricing plans over the last 8 weeks. The Skill builds a HogQL query grouping by week and plan, validates the percentile output, saves the box plot insight, and confirms it renders correctly.
Quick Start
Create a box plot insight in PostHog showing the distribution of the latency_ms property for the 'request completed' event grouped by week.
