What-if Analysis for Business Professionals: Current Practices and Future Opportunities

Research Background and Problem Statement

• What problems or challenges did the authors identify?

  1. Current What-if Analysis (WIA) tools and research primarily target technical data analysts, neglecting business professionals in non-technical domains.
  2. Business professionals, despite needing data-driven decision-making, lack user-friendly analytical tools and are forced to rely on manual or rudimentary methods for advanced analysis.
  3. Commercial tools lack features that support the entire workflow from data exploration to predictive and prescriptive analysis, failing to answer the question, "What should we do next?"

• Why is this problem important?

  • WIA is critical for optimizing business decisions in areas such as marketing, sales, and product management. However, business professionals are not technical experts and cannot effectively access or use current complex analytical tools.
  • Business decisions require quick execution, and any delays or complex dependency paths (e.g., reliance on data teams) can hinder decision-making efficiency.

• Research Motivation and Related Work

  • The authors emphasize the need to empower non-technical users (e.g., business managers) to independently perform advanced analytical decision-making, reducing reliance on data teams.
  • Related work focuses on visualization dashboards, machine learning tools, and advanced analytics but lacks a WIA design framework tailored for non-technical business users.

Solution

• What methods or solutions did the authors propose?

  1. Conducting a two-phase user study with 22 business professionals from marketing, sales, product, and operations departments.
     • Phase 1: Interviews to understand their current WIA practices and key challenges.
     • Phase 2: Developing an interactive visualization prototype to test four typical WIA techniques.
  2. Providing design recommendations to improve future business analytics systems.

• What is innovative about this solution?

  • The study focuses on business professionals, a group that has been largely overlooked.
  • A multi-step user research design is employed to explore the practicality and improvement areas of WIA techniques.
  • By developing a prototype, the study directly captures participants' feedback on using advanced analytical techniques in practice.

• What are the implementation steps and key technologies used?

  1. User Interviews:
     • Identifying participants' goals, data usage, and the shortcomings of current tools.
     • Mapping their analytical needs in specific business scenarios (e.g., optimizing ad spend).
  2. Prototype Development:
     • Implementing four WIA techniques (driver importance analysis, sensitivity analysis, goal-seeking analysis, and constraint analysis).
     • Using intuitive interactive interfaces (e.g., bar charts, small multiples, multi-view coordination).
  3. Task Evaluation:
     • Observing participants as they use the prototype to complete tasks and gain decision support in given business scenarios.
  4. Feedback Collection and Analysis:
     • Gathering opinions on perceived usefulness, scalability, and suggestions for improvement.

Research Outcomes

• What specific outcomes were achieved?

  • Current WIA tools are overly complex and unintuitive, with most participants relying on spreadsheets for manual calculations (inefficient and error-prone).
  • The four WIA techniques provided in the prototype (especially sensitivity analysis and constraint analysis) significantly improved decision speed and confidence.
  • Business professionals were open to adopting these techniques and eager to apply them to their own business scenarios.

• What advantages does it have compared to existing solutions?

  1. The design better meets the needs of business professionals, requiring no programming or statistical knowledge.
  2. A faster "hypothesis-validation-decision" cycle reduces reliance on analytics teams.
  3. User-defined constraints and real-time feedback enhance flexibility.

• What were the experimental or evaluation results?

  • Participants found sensitivity analysis the most useful, followed by constraint analysis and goal-seeking analysis.
  • The representative prototype not only enabled participants to handle business scenarios more effectively but also helped them identify multiple potential improvement opportunities.
  • Users emphasized the need for more data preparation support, greater transparency in risk assessment, and stronger integration of domain knowledge.

• Limitations and Future Directions

  • Limitations:
    • The study sample was limited, and some techniques and use cases may not have been fully covered.
    • Prototype testing was restricted to a single common business scenario, and its cross-industry generalizability remains unverified.
    • No comprehensive evaluation was conducted against existing advanced commercial tools.
  • Future Directions:
    • Expanding the study to other industries and functional departments to enhance the broad applicability of the techniques.
    • Strengthening integration with data preparation tools while supporting more external constraints (e.g., social and economic conditions).
    • Exploring interfaces based on NLP and generative AI to further lower the technical barrier.