Verify AI Variance Reporting
Lead Product Designer | 6 Months
Capterra Visual Design Overhaul
Lead Product Designer | 6 Months
Capterra Visual Design Overhaul
Lead Product Designer | 6 Months
Please Note
NDA requirements prevent me from showing some key aspects of work involved in this project. If you'd like to know more please ask about my design process! 🙂
Please Note
NDA requirements prevent me from showing some key aspects of work involved in this project. If you'd like to know more please ask about my design process! 🙂
Background
At Verify, I led design efforts for an AI-assisted variance reporting workflow built for multifamily operators, asset managers, and ownership groups.
The core challenge was not simply financial reporting. It was a breakdown in stakeholder communication.
Every month, property managers, regional managers, and ownership stakeholders struggled with a fragmented, repetitive workflow where critical operational context was scattered across spreadsheets, emails, meetings, and disconnected systems.
Our goal was to use AI as infrastructure for improving clarity, alignment, and decision-making between stakeholders.
Background
Research employee, franchise, and customers’ needs to inform Restore’s customer experience strategy. Our deliverables include an initial website expression, a hero flow across major touch points, and a service framework.
The Problem
Multifamily reporting is highly collaborative, but the communication systems surrounding it are often extremely inefficient.
A single monthly reporting cycle might involve:
Property managers explaining variances
Regional managers requesting clarification
Ownership groups reviewing summaries
Finance teams validating accuracy
Executives trying to identify portfolio-level risk
The issue was that each stakeholder required a different level of detail, but everyone relied on the same fragmented data sources.
This created several problems
The Problem
Multifamily reporting is highly collaborative, but the communication systems surrounding it are often extremely inefficient.
A single monthly reporting cycle might involve:
Property managers explaining variances
Regional managers requesting clarification
Ownership groups reviewing summaries
Finance teams validating accuracy
Executives trying to identify portfolio-level risk
The issue was that each stakeholder required a different level of detail, but everyone relied on the same fragmented data sources.
This created several problems
Communication was Repetitive
Property managers repeatedly answered the same questions from different stakeholders.
Context was Lost
Critical operational knowledge lived in:
Slack messages
Email threads
Meetings
Institutional memory
Very little context persisted in the reporting workflow itself.
Reporting Was Reactive
Stakeholders spent more time interpreting and clarifying information than acting on it.
Existing Tools Were Passive
Most reporting systems surfaced raw numbers, but did little to help users understand:
Why something changed
Whether it mattered
What action should be taken
Communication was Repetitive
Property managers repeatedly answered the same questions from different stakeholders.
Context was Lost
Critical operational knowledge lived in:
Slack messages
Email threads
Meetings
Institutional memory
Very little context persisted in the reporting workflow itself.
Reporting Was Reactive
Stakeholders spent more time interpreting and clarifying information than acting on it.
Existing Tools Were Passive
Most reporting systems surfaced raw numbers, but did little to help users understand:
Why something changed
Whether it mattered
What action should be taken
Discovery and Research
Through internal reviews and customer conversations, several patterns emerged.
Insight 1: Most Variances Were Operational Stories, Not Financial Ones
A negative variance often represented:
Delayed turns
Staffing shortages
Weather events
Vendor increases
Leasing challenges
The most interesting data wasn't financial, it was operational.
Insight 2: Stakeholders Consumed Information Differently
Different users required different levels of abstraction.
Property Managers
Needed:
Detailed transactional context
Ability to explain anomalies quickly
Regional Managers
Needed:
Prioritization
Operational risk visibility
Approval workflows
Ownership Groups
Needed:
Concise summaries
Confidence in reporting accuracy
Insight 3: AI Could Reduce Interpretation Friction
We identified an opportunity for AI to:
Synthesize context
Identify patterns
Assist explanation generation
Surface important anomalies
Preserve historical operational knowledge
Importantly, users did not want AI replacing decision-making, but we did see an opportunity to reduce cognitive load utilizing AI.
Discovery and Research
Through internal reviews and customer conversations, several patterns emerged.
Insight 1: Most Variances Were Operational Stories, Not Financial Ones
A negative variance often represented:
Delayed turns
Staffing shortages
Weather events
Vendor increases
Leasing challenges
The most interesting data wasn't financial, it was operational.
Insight 2: Stakeholders Consumed Information Differently
Different users required different levels of abstraction.
Property Managers
Needed:
Detailed transactional context
Ability to explain anomalies quickly
Regional Managers
Needed:
Prioritization
Operational risk visibility
Approval workflows
Ownership Groups
Needed:
Concise summaries
Confidence in reporting accuracy
Insight 3: AI Could Reduce Interpretation Friction
We identified an opportunity for AI to:
Synthesize context
Identify patterns
Assist explanation generation
Surface important anomalies
Preserve historical operational knowledge
Importantly, users did not want AI replacing decision-making, but we did see an opportunity to reduce cognitive load utilizing AI.
Core User Flow
Through our generative feedback sessions we mapped out a core user flow.
Property Manager: Receives notification → Reviews flagged variances → Adds comments → Responds to AI questions → Submits report
Regional Manager: Reviews explanations → Asks clarifying questions → Approves or requests more information → Forwards to executive
Executive: Views portfolio-wide summaries → Identifies trends → Takes strategic actions
Core User Flow
Through our generative feedback sessions we mapped out a core user flow.
Property Manager: Receives notification → Reviews flagged variances → Adds comments → Responds to AI questions → Submits report
Regional Manager: Reviews explanations → Asks clarifying questions → Approves or requests more information → Forwards to executive
Executive: Views portfolio-wide summaries → Identifies trends → Takes strategic actions
Design Process
I mapped the full information flow across user roles and created low-fidelity wireframes to align stakeholders around a simple, core experience. In Early iterations I explored more complex multi-screen flows, but validation with customers revealed users needed a more direct path.
I reduced the experience to five key steps:
Receive variance notification
Review flagged items
Add structured explanations
Submit for review
Access portfolio insights
Working closely with the product manager, we defined an MVP that prioritized core functionality while laying the foundation for future analytics and sharing features.
Design Process
I mapped the full information flow across user roles and created low-fidelity wireframes to align stakeholders around a simple, core experience. In Early iterations I explored more complex multi-screen flows, but validation with customers revealed users needed a more direct path.
I reduced the experience to five key steps:
Receive variance notification
Review flagged items
Add structured explanations
Submit for review
Access portfolio insights
Working closely with the product manager, we defined an MVP that prioritized core functionality while laying the foundation for future analytics and sharing features.
Current Experience
Variance Commenting: Table interface with contextual variance details, required fields, and progress indicators.
Manager & Executive Dashboards: Portfolio health summaries, key metrics, and variance trends across categories like maintenance, utilities, and income.
Admin Configuration: Drag-and-drop approval chains, customizable thresholds, and property assignments.
Current Experience
Variance Commenting: Table interface with contextual variance details, required fields, and progress indicators.
Manager & Executive Dashboards: Portfolio health summaries, key metrics, and variance trends across categories like maintenance, utilities, and income.
Admin Configuration: Drag-and-drop approval chains, customizable thresholds, and property assignments.
Core Features:
Variance Table Interface: Familiar spreadsheet-like layout with comment fields and color coded priorities.
Approval Workflow: Review and approval stages for regional and executive stakeholders.
Executive Dashboard: Multi-tab view for AI generated portfolio summaries, trend analysis, and action tracking.
Admin Controls: Configurable thresholds, roles, and workflows to adapt to different organizational structures.
Core Features:
Variance Table Interface: Familiar spreadsheet-like layout with comment fields and color coded priorities.
Approval Workflow: Review and approval stages for regional and executive stakeholders.
Executive Dashboard: Multi-tab view for AI generated portfolio summaries, trend analysis, and action tracking.
Admin Controls: Configurable thresholds, roles, and workflows to adapt to different organizational structures.
Impact
Significant reduction in time spent on variance reporting for current customers (anecdotal)
100% completion rate for variance explanations
Actionable insights delivered to executives within 3 days vs. 2+ weeks
Successfully deployed across multiple property management companies with 20+ properties each
Impact
Significant reduction in time spent on variance reporting for current customers (anecdotal)
100% completion rate for variance explanations
Actionable insights delivered to executives within 3 days vs. 2+ weeks
Successfully deployed across multiple property management companies with 20+ properties each
