eCloudMobile Business Intelligence

The Problem: Bridging the “Insight Gap”

Small-to-mid-sized businesses generate large amounts of transactional data, but turning that data into actionable insight is often difficult and time-consuming.

This project explores a core hypothesis:

"SMEs are data-rich but insight-poor—not due to lack of data, but due to the high cost of interpreting it."

In many common workflows, business owners and managers:

• Export data into spreadsheets for manual comparison
• Review multiple reports across different timeframes or channels
• Rely on intuition when time or expertise is limited

While these workflows provide access to data, they introduce significant friction in the decision-making process.

Strategic Direction: Designing for Speed to Insight

To address this gap, I defined two key constraints that shaped the product approach:

Economic Constraint: existing BI tools (e.g., Tableau, Looker) are cost-prohibitive for SMEs.

Cognitive Constraint: most BI tools are designed for analysts—not business operators.

Core Design Principle

Prioritize Speed to Insight over Customization Complexity

Instead of offering highly flexible but complex dashboards, the product focuses on minimizing the time it takes for users to:

• Understand what’s happening
• Identify why it’s happening
• Decide what to do next

This principle guided key trade-offs, including:

• Limiting the number of surfaced metrics to reduce cognitive load
• Structuring information around business questions rather than raw data
• Reducing reliance on manual chart interpretation

Target Users

SME owners and store managers who need to make fast operational decisions—such as inventory planning, staffing, and promotions—without a background in data analysis.

Design Decisions: Reducing Friction from Data to Decision

Each design decision was guided by a single goal:

"Minimize the time and effort required for users to move from data to action."

Rather than maximizing flexibility, the system prioritizes clarity, focus, and guided interpretation.

1. Information Architecture: Structuring Around Decisions, Not Data

Traditional BI tools often organize information around datasets or reports, requiring users to navigate and interpret multiple layers before reaching a conclusion.

To reduce this friction, the dashboard is structured around four core business questions:

• 銷售金額 (Revenue) — How much are we earning?
• 銷售筆數 (Transactions) — How are sales performing?
• 銷售人數 (Customer Traffic) — How many customers are visiting?
• 平均客單價 (Average Order Value) — How much is each customer spending?

This approach allows users to:

• Quickly assess overall performance at a glance
• Navigate directly to relevant insights without interpreting raw data structures
• Maintain context while exploring specific metrics

Design Trade-Off

To support faster comprehension, the system intentionally:

• Limits the number of surfaced metrics
• Avoids deep hierarchical navigation
• Reduces reliance on custom report building

While this reduces flexibility for advanced users, it significantly lowers the cognitive load for time-constrained operators.

2. Data Visualization: Optimized for Immediate Interpretation

Instead of offering a wide variety of chart types, visualization choices were intentionally constrained to reduce interpretation effort.

Each chart type was selected based on how quickly users can extract meaning:

• Line Charts → Highlight trends, patterns, and anomalies over time
• Bar Charts → Enable fast comparison across products or categories
• Pie Charts → Provide a quick understanding of proportional distribution

Design Rationale

The goal was not visual variety, but interpretation speed.

More complex visualizations (e.g., stacked charts, multi-axis graphs, dense tables) were avoided because they:

• Require higher analytical literacy
• Increase cognitive load
• Slow down decision-making in time-constrained scenarios

Design Trade-Off

By simplifying visualizations:

• Users can interpret data faster with minimal training
• However, advanced analytical depth is reduced

This reinforces the product’s positioning as a decision-support tool, rather than a full analytical platform.

3. AI-Assisted Interpretation: Reducing Cognitive Effort

A core limitation of traditional BI tools is that data visualization does not guarantee understanding. Users are still required to interpret charts, identify anomalies, and determine next steps.

To address this, the platform introduces an AI-driven insight summary designed to translate data into actionable guidance.

Instead of requiring users to analyze multiple charts, the system:

• Detects significant changes or anomalies in key metrics
• Provides contextual explanations (e.g., trends, seasonality, or recent patterns)
• Suggests potential next steps to support decision-making

Design Rationale

This reduces the need for users to:

• Compare multiple timeframes manually
• Identify anomalies across charts
• Translate observations into decisions

Instead, the system supports a more direct workflow:

Observation → Understanding → Action

Design Trade-Off

Given the risks associated with AI-generated content, the feature was designed with several constraints:

• Transparency: insights are supported by underlying data and visualizations to allow verification
• Relevance Filtering: only high-signal changes are surfaced to avoid overwhelming users
• Clarity over complexity: explanations are written in simple, operational language rather than technical terms
• Decision Support, Not Automation: the system suggests actions but does not make decisions on behalf of users

Final Interface

The final interface transforms complex business data into a clear, actionable workflow—helping users move from raw data to insight and decision-making without manual analysis.

Example Scenario: From Data to Decision

A store owner opens the dashboard in the morning to review yesterday’s performance.

Instead of scanning multiple charts, they immediately see:

• A drop in revenue compared to last week
• An AI-generated explanation linking the decline to a holiday period
• A suggestion to adjust staffing or promotions

They move from data review → understanding → action without manual analysis.

Collaboration & Technical Execution

To ensure the design translated effectively into a working product, I worked closely with engineering throughout development:

• Validated data-heavy components for performance across large datasets
• Ensured responsive behaviour across devices
• Conducted hands-on QA in staging environments to verify interactions such as tooltips, comparisons, and scaling

Impact (Pre-Launch)

The redesigned experience reduces the effort required to interpret business performance by:

• Surfacing key metrics immediately on load
• Providing contextual explanations for anomalies
• Eliminating the need for manual data comparison

This enables users to move from data review to decision-making quickly, without needing to manually compare reports.

Reflections

This project reinforced that designing for non-experts is not about simplifying data—it’s about simplifying decisions.

By removing unnecessary complexity and focusing on clarity, we created a system that respects the user’s time and attention while still delivering meaningful insights.

Next Steps: Post-Launch Validation

Following launch, the focus will shift to validating real-world impact:

• Behavioral Analysis: identify which metrics users rely on most and where friction occurs in exploration
• Merchant Collaboration: work with pilot users to ensure insights translate into operational improvements
• Continuous Iteration: refine the system based on how effectively it reduces decision-making time in daily workflows