Inventory IQ
Overview.
Inventory IQ is an internal inventory counting system designed to streamline cycle counting workflows, where operators regularly verify inventory quantities across production lines. The system improved real-time data entry, discrepancy tracking, and visibility across operations.
My Role.
As the sole product designer, I led the end to end design of an internal inventory counting system used by production operators and supervisors. I collaborated closely with software and data engineering teams to define workflow architecture, simplify operational processes, and design scalable interfaces for real-time discrepancy tracking and review.
My work included workflow mapping, wireframing, prototyping, visual design, and usability testing across desktop and tablet experiences.
Problem.
🧑💻 User Experience
Operators rely on customized Excel sheets to collect and manage machine data, resulting in inconsistent data, limited visibility, and high operational effort.
Operators spend ~20 min per cycle count, repeated up to 3x per shift.
12–14 manufacturing lines relied on disconnected Excel sheets with no input validation.
Errors are only identified after count results are compared, making real-time correction impossible.
💼 Business
Slow and inaccurate inventory discrepancy detection delayed corrective actions, contributing to an estimated $500K in annual production losses.
Solution.
Design a centralized real-time inventory system that catches input errors at the point of entry, calculates average discrepancy rates across cycle counts to monitor performance, and surfaces historical trends to support long-term operational analysis.
Constraints.
📡 Data Latency
With data flowing from 3+ manufacturing systems, true real-time accuracy was not achievable. Because cycle counts took several minutes to complete, the system relied on timestamped validation by capturing data at the moment of entry and comparing it against current system values.
💾 Infrastructure Costs
Leadership raised concerns that large data exports would significantly increase server costs. The solution was adapted to export only essential process data, while detailed historical counts remained accessible within the application.
🔍 Scope Limitations
Advanced analytics features, including trend detection and recurring discrepancy insights, were intentionally deferred to keep the initial release focused on the core counting workflow.
🕓 Time Pressure
As part of a broader smart factory initiative, leadership prioritized rapid delivery as delays and operational mistakes were driving significant production costs, limiting the initial scope to core functionality.
Design Goal.
Because this workflow was being fully digitized from a previously manual process, no baseline metrics existed to measure success.
To guide the project, we defined our own success criteria:
⭐ Reduce Data Entry Errors
Catch mistakes at the moment of entry, not after counting is complete
⭐ Ensure Data Accuracy
Maintain a timestamped history so discrepancies can be traced back to the source
⭐ Improve Data Visibility
Give supervisors real-time status without waiting for end-of-shift comparisons
User Research.
Field Study
Operators either recorded machine readings on paper for later entry or manually entered data directly into Excel on the production floor.
Operators often paused counting workflows to respond to other production requests.
Data entry was frequently delayed by operational interruptions and multitasking.
Interviews
Operators described inconsistent counting workflows, revealing no standardized process across teams.
The most common frustration was losing track of progress after interruptions mid-count.
Delays were frequently caused by switching between tools and waiting on other operators to complete counts.
Most operators had no reliable way to detect errors until they were later flagged by supervisors.
Task Analysis
Users referenced multiple Excel sheets to locate the correct fields and input locations.
Count data was manually transferred between spreadsheets and digital forms.
Workflow execution varied significantly by operator, including direct laptop entry and handwritten note-taking.
Key Insights.
🔑 Workflows vary by operator, not by system
No standardized process existed. Some operators entered data in real time, while others relied on handwritten notes or memory, leading to inconsistent results across the same workflow.
🔑 Accuracy drops when data entry is delayed
Operational interruptions often delayed count entry, increasing reliance on recall-based input and introducing mistakes across 12–14 manufacturing lines.
🔑 Errors go unnoticed at the point of entry
Without input validation, operators had no reliable way to detect mistakes until results were reviewed after the counting process.
🔑 Issue detection happens too late
Input errors became difficult to trace and correct once operators moved on to other tasks.
Design Approaches.
🧩 Support in-the-moment data capture
Because delayed entry increased recall-based mistakes, the experience was designed to make real-time input fast and low-friction, even on the production floor.
🧩 Design around data accuracy, not live data
True real-time synchronization was not achievable across 3+ manufacturing systems. Instead, the system relied on timestamped validation to provide reliable inventory data despite system latency.
🧩 Create a guided counting experience
Operators lacked a consistent workflow. A guided process reduces reliance on individual habits and ensures counts are completed consistently across teams.
🧩 Improve visibility across the workflow
Supervisors previously had to wait until all cycle counts were completed before reviewing results. Submitting each count individually gave them earlier visibility into discrepancies throughout the counting process.
🧩 Extend the system beyond task completion
Cycle counts were only one part of the operational picture. Historical tracking enabled teams to identify recurring discrepancies and monitor long-term performance trends.
User Roles.
Production Operators / Shift Leads
Perform cycle counts on the production floor under time pressure. Primary users of the counting workflow, responsible for entering inventory data across 12–14 manufacturing lines.
Inventory Management Team
Review and resolve discrepancies between physical and system inventory when submitted counts do not match expected values.
Supervisors / Managers
Monitor counting progress, completion rates, and team accountability without directly performing counts.
Inventory Data Analysts
Analyze historical count data to identify recurring discrepancies and operational trends over time.
System / Dashboard Owners
Manage system integrations, configurations, and data reliability across manufacturing systems.
Design Process.
Information Architecture
The system was organized around two core workflows: cycle counting and discrepancy review. Defining the structure early helped align engineering and data teams on system scope before interface design began.
Userflows
Mapped the counting workflow end-to-end to identify unnecessary steps and role handoffs. This helped simplify the process, reduce operational friction, and define the minimum number of screens required.
Wireframes
Built to validate the structure and sequencing of the counting workflow with real operators before moving into high-fidelity design. Early testing helped avoid investing in detailed UI patterns before the workflow was finalized. Because each process followed a different data structure, separate wireframes were created to validate layout patterns and input logic across workflows.
Iterations.
METHODS
🔄 Usability Testing
Operators frequently backtracked between steps to verify previously entered information.
Page transitions interrupted the counting rhythm, especially during active entry.
Testing showed the primary friction came from navigation, not data entry.
🆚 Time-on-Task Comparison
Completion time was measured during testing to evaluate workflow efficiency.
Step-by-step navigation increased task completion time.
Repeated back-and-forth navigation introduced unnecessary delays.
Interruptions during navigation slowed overall input flow.
📝 Stakeholder Feedback Loops
Stakeholders initially wanted to preserve the original step-by-step flow to avoid retraining 40+ operators. Testing data demonstrated faster completion times and fewer errors with the revised workflow, leading to alignment on the updated design direction.
ITERATION OUTCOMES
Decreased steps from 7 pages → 1 continuous flow
The original workflow required users to navigate through 7 separate pages with repeated “Next” actions. The redesigned flow consolidated the process into a single continuous experience to reduce navigation interruptions during counting.
Reduced completion time 20 minutes -> 15 minutes
Measured during wireframe testing with 3 operators. Removing repeated page transitions reduced delays caused by back-and-forth navigation.
Improved Data Visibility
All inputs were consolidated into a single scrollable view, making it easier for operators to review and validate entries before submission.
USERFLOW
Before
Repeated loops and validation steps created unnecessary navigation throughout the workflow.
After
Reduced decision points and eliminated repeated loops, shortening the path from start to submission.
WIREFRAME
Before
Two separate pages, each with unused screen real estate and no visibility into previous inputs.
After
Consolidated into one continuous scroll, with all inputs visible and inline validation throughout.
Final Designs.
Count Overview
Operators can view up to three count submissions in a single interface, using discrepancy indicators and comparison metrics to quickly identify mismatches before escalation or recalibration decisions are made.
Real-Time Entry and Error Prevention
The left sidebar tracks progress across each counting step, helping operators maintain context throughout the workflow. Inline validation flags missing or incomplete inputs in real time before submission.
Review and Continuous Workflow
All counting steps are completed within a single continuous flow, allowing operators to review entries and validate discrepancies before submission.
Additional Workflows
Beyond the primary counting workflow, the platform also supported additional operational processes for compiled summaries, discrepancy reporting, and cross-line review across 6 manufacturing workflows.
Reflections.
Clear requirements reduce rework
As new requirements were introduced mid-project, both the data structure and workflow logic had to be revised multiple times. Earlier alignment on scope and system requirements would have reduced redesign effort and engineering delays.
Reducing navigation improved efficiency
We initially assumed operators needed a step-by-step workflow to stay on track. Testing showed the opposite, removing repeated page transitions made the process faster, easier to review, and less error-prone.
Testing revealed real operator behavior
Field testing exposed frequent interruptions, backtracking, and delayed data entry patterns that were not visible during early planning. These insights directly shaped the continuous workflow and inline validation system.
Visibility reduced operator errors
The most impactful improvement was not adding more functionality, but improving visibility throughout the workflow. Keeping all inputs and review states accessible in a single interface reduced missed entries and unnecessary backtracking.
Next Steps.
Currently undergoing user testing with production operators prior to full deployment.
Validate usability and workflow efficiency improvements in real production environments
Track completion time, discrepancy rates, and input error frequency over time
Continue refining workflows based on operator and stakeholder feedback
Expand historical trend analysis to identify recurring discrepancy patterns across production lines