Custom Dashboards & AI Reporting

D-Tools Cloud shipped with a static dashboard that showed field technicians, project managers, and company owners the exact same charts. Because it didn't answer anyone's daily questions, most users ignored it. I redesigned the workspace around role-based starter templates, customizable widgets and views, and one-click AI reporting tied to on-screen metrics, driving 71% multi-dashboard adoption.

My role

Product Design Lead (working with 1 PM and 4 developers)

  • Owned discovery research, information architecture, and the 12-column widget grid
  • Worked with 4 full-stack engineers on widget data schemas and role permissions

Key outcomes

  • 71% multi-dashboard adoption

71% of active integrator accounts maintain two or more separate dashboards.

Executive BI dashboard overview showing role-based sales metrics, close rate gauge, and trend analysis
Role-specific KPI counters, pipeline velocity gauges, and margin trend cards configured for commercial contractor reviews.

The Problem: Trapped in Pre-Canned Dashboards

For a 10-person residential AV company, high-level sales gauges were useless clutter. For a commercial contractor tracking $2M bids, daily service ticket counts got in the way. Meanwhile, leadership wanted to jump straight to a conversational AI chatbot. My job was to solve the immediate reporting pain first and channel the AI push into something integrators would actually trust with their job margins.

Goals

  • Give each role (Field Dispatch, Sales, Accounting) a starter template that answers daily questions in under 30 seconds
  • Let users resize, filter, and save multiple dashboard views without feeling like they are configuring a complex BI tool
  • Keep AI reporting strictly tied to the numbers visible on screen so financial data never hallucinates

Research Insights

To understand why teams abandoned existing analytics, I led 10 semi-structured discovery interviews: 6 with active power users (integrator ops managers, project leads, and accounting controllers) and 4 with internal stakeholders across sales, executives, and customer success.

Research Findings

  • Finding 1
    Core insight: Many users simply ignored dashboards entirely.
  • Finding 2
    Workflow friction: integrators needed fast routines: customizing a daily view couldn't feel like building a report from scratch.
  • Finding 3
    Sales constraint: Sales reps needed immediate visual punch for live demos.

Design Decisions

  • Decision 1
    Role-based templates: Replaced existing static layouts with starter templates matching specific company role.
  • Decision 2
    Fast & easy card manipulation: Drag-and-drop rearrangement and simple AI reporting from active widgets.
  • Decision 3
    Demo-ready views: Pre-populated executive layouts that sales could pull up instantly during discovery calls.

What I got wrong first

Early in wireframing, I pitched three static department layouts: Sales, Service, and Operations. But during user walkthroughs, the approach broke down immediately. A 12-person residential integrator runs completely differently than a 60-person commercial shop; their daily priorities had almost zero overlap. Hardcoded department views would have recreated the exact problem we were trying to fix. That pushed us toward modular templates that managers could reorganize, customize, and rename quickly, easily, and whenever they wanted.

User Research Findings 10 User & Stakeholder Interviews
Research synthesis comparing residential and commercial integrator daily workflows
Integrator Cohort Their Question Where Hardcoded Views Failed
Residential Integrator
10-15 staff
"Is this month's revenue better than last month's?" Current revenue totals lacked the prior-month context needed to show whether performance was improving.
Commercial Integrator
50+ staff
"Where are active change orders eroding our project margins?" Margin impact was buried across individual change orders, making it difficult to see which active projects were losing profitability.

Design Approach & Decisions

  • 1. Responsive 12-Column Widget Grid

    Rather than building isolated reporting pages, I designed a modular 12-column grid supporting three widget sizes: a 1x1 counter, 2x2 trend chart, and 3x2 data table. Users could start from a role-specific template or a blank canvas, dragging and configuring cards to match their morning routine.

  • 2. Grounded AI Reporting

    Project managers and executives don't trust AI guesses when job margins and billable hours are on the line. I worked with backend engineering so the summary only reads the data returned by the active widgets on the user's screen, never running open-ended database queries.

Widget Center modal: integrators browse and add specific operational metrics (such as change orders and sales volume) directly to their active view.
Executive report: compiling active dashboard widgets into a consolidated summary brief with sections for strengths and weaknesses.

Defining Decision: Dashboards Before AI

Leadership originally wanted to skip visual dashboards and build a chat prompt interface. In usability tests, however, integrators froze when faced with a blank chat box. AV integrators don't log in at 7:00 AM to write prompts. They need an immediate visual scan of how they can improve their business that day.

That testing convinced leadership to keep visual cards as the primary morning workspace and (for now) use AI strictly as a quick report writer for weekly/monthly/quarterly leadership meetings.

Execution Constraints: Balancing Different User Needs

The dashboard had to serve people with very different priorities. Executives needed a quick read on business performance, while operations teams needed to know what required attention right now. With limited space and a lot of potentially useful information, the challenge was deciding what deserved prominence without turning the dashboard into a wall of metrics.

From Reporting to Action

The existing dashboard leaned heavily toward high-level reporting, which could push more immediate operational needs further down the page. I redesigned the hierarchy around the questions users needed answered throughout the day, surfacing actionable information while keeping the broader business picture accessible.

Prototyping & Validation

Iteration: higher data density, tighter control

I tested the interactive Figma prototype with five power users. Integrators immediately rejected our early wireframes. They wanted compact counters (unassigned service calls, overdue milestones, low-stock flags) docked at the top of the screen above trend charts. Adding a compact 1x1 stat card and drag-and-drop reordering took one extra sprint, but turned the dashboard into a daily second-monitor workspace.

Customizing dashboard cards: adjusting dimensions and managing active widgets to fit daily scanning workflows.
  • 4/5 users said the role templates matched how they already organized their week
  • All 5 participants customized and saved a new dashboard without prompting

Impact: Replacing Weekly Exports with Daily Habit

Measured at 30 days post-launch, 71% of active accounts maintained multiple custom dashboards, and weekly dashboard views rose 64% across consecutive quarters.

71% multi-dashboard adoption
+64% sustained dashboard views

Users applied an average of 2.4 custom filter rules per card. Usage data also showed a clear split in daily habits: visual charts peaked between 6:45 AM and 8:15 AM as managers routed field crews, while AI reporting clicks surged at month-end and quarter-close when leads compiled owner reports.

Rename and manage dashboards for different workflows
Customized views: integrators store custom dashboard presets for weekly pipeline reviews, inventory checks, service ticketing, or whatever they want.

Lesson

Executive excitement for conversational AI rarely matches how field integrators actually work. Integrators managing tight schedules and thin margins don't want to chat with their software; they want fast, reliable numbers they can scan at 7:00 AM and a one-click summary when it's time to brief the owner.

What's Next

Next on the roadmap: I left D-Tools before some of our future dashboard dreams were put into production. The next big thing we had lined up was user-scheduled email digests.