This week brought several reliability improvements to the analytics dashboard, including a critical fix for event charts that were silently truncating data when tracking many event types. We also improved device categorization, solidified error handling for API key scopes, and refined the insights engine context for more useful recommendations. If you're tracking multiple custom events or debugging device tracking, these changes will make a difference.
What's new in Databuddy
All the latest features, improvements, and fixes shipped to Databuddy, straight from the team.
Databuddy changelog: March 6-13, 2026
Highlights
Fixed events trend chart cutting off with many event types
The custom events trends query was silently limiting results to roughly 11 days when tracking 14+ event types. The query default limit was 100 rows; with ~14 event types over 30 days, this truncated the entire dataset. Bumped to 5000 and fixed org-level filtering to show accurate trending data.
Improved error handling for API key scopes
Cleaned up MCP protocol integration issues and clarified error messaging when API keys lack required permissions. Better defaults on scope requests mean fewer cryptic auth failures.
Better loading state feedback
Improved visual feedback during data fetches so users know the dashboard is working when requests take time.
Fixed device categorization for tablets
Tablet detection was miscategorizing some devices. Now correctly identifies tablets separately from phones and desktops.
Sharper insights engine context
Refactored event page context and consolidated tooltip utilities to give the insights engine better information for generating relevant recommendations.
More Updates
Bug Fixes
- Unstack chart visualization - Fixed stacked chart layout issue affecting readability.
- Org-level query filtering - Corrected WHERE clause for org-specific dashboard queries to respect organization boundaries.
Features & Enhancements
- Consolidated chart utilities - Unified tooltip and context handling across event dashboards to reduce duplication.
- Removed experimental LLM analytics tracking - Cleaned up temporary analytics module pending future redesign.
Internal Changes
- Dead code cleanup - Removed obsolete event dashboard code paths.
- UI refinements - Visual tweaks to gradient fills and background patterns across dashboard views.