Custom Reports
Editor Framework
Designing a scalable reporting framework for fleet safety managers who needed Excel-like power without the database complexity.
- Role
- Lead Product Designer
- Team
- 2 PM · 1 EM · 3 Eng · 1 Principal Eng · 1 DS Designer
- Timeline
- Jan – Jul 2025
- Company
- Samsara
Background
Problem
Samsara's existing reports were fixed templates — Safety Managers couldn't filter by what mattered to their operation, couldn't group by asset tag, and had no way to visualize trends. Support tickets for "reporting limitations" were piling up.
Goal
Build a flexible, scalable reporting framework that lets fleet operators build meaningful reports from scratch — without requiring SQL or BI tool expertise. A simple tool that could grow into an enterprise-grade one.
Non-goal
This was not a replacement for Tableau or PowerBI. Advanced analytics reporting — pivot tables, cross-report joins, regex filters — would be available to enterprise tier, but not the core experience.
Target Users
Safety Manager / Fleet Ops Manager
Proficient in Excel (formatting, filtering, formulas), but not familiar with database concepts. Wants to build reports to regularly share out with managers.
Safety/Compliance Director
Expert in BI tools and Excel, may understand database concepts. Wants to build reports to share with execs and managers they work with.
Executives VP
Viewer of reporting through PDF exports or emailed versions of the reports.
Data Analyst
Familiar with SQL and database concepts.
Research
Audited 9 organizations and 30 customers to benchmark data manipulation expectations.
We ran usability testing across DHL, Iron Mountain, Harris Teeter, Artera, Univar, and five other enterprise customers. The sessions revealed that safety managers weren't intimidated by filters or sorting — they were intimidated by starting from nothing.
Industry benchmarking across PowerBI, Databricks, Airtable, Tableau, Salesforce, Mode, HubSpot, and Mixpanel helped us define what “basic,” “standard,” and “enterprise” actually meant in this context.
Design Principles
Start with a dataset, not a blank canvas
Safety Managers aren't database engineers. Anchoring them in familiar data objects (Assets & Vehicles, Safety Events) before asking them to build removes the most common failure point.
Build off existing patterns
Samsara's product already had mental models users trusted. The framework inherited those patterns and extended them, reducing the learning curve for the 80% case.
Progressive disclosure
Standard users see a clean table editor. Advanced users unlock multi-column sorts, grouping, and charts. Enterprise users get cross-dataset joins. Complexity scales with capability.
Process
Guided Setup Flow
We chose Option 2 — contextualizing the dataset at setup. Showing field previews alongside the dataset picker removed ambiguity about what data would appear in the report, reducing failed first attempts in usability testing.
Main Edit Pattern
Inline editing won. The sidebar added a mode-switching cognitive load that slowed down expert users. Inline editing kept the report in context while editing — closer to the spreadsheet mental model our primary user already had.
Data Preview
Live preview with a clear 'limited data' label. Users needed to see real data to validate their choices, but a full query on every change was too slow. A 5-row preview hit the confidence/performance tradeoff correctly.
Final Design
Inline editing with live filter context
Filters, grouping, and chart editing all happen in-context on the report surface. No mode switching, no sidebar that obscures the data you’re editing against.
Real data, limited rows, clear signal
A 5-row live preview with a clear “limited data” banner gave users enough to validate their column choices without the cost of a full query on every change.
Scaling chart components across the product
Every chart type, color scale, and data label we designed was contributed back to Samsara’s design system, unlocking visualization capabilities for other product teams.
Outcomes
of target accounts built a report post-launch
drop in limitation-related support tickets
in new revenue attributed within 6 months
“The framework shipped as Standard in January 2025. Advanced features were already queued as post-GA fast follow — cross-dataset joins, regex filters, and scheduled chart delivery.”