Using Analytics to optimize Conference Scheduling at Global Business School Custom Case Solution & Analysis

Evidence Brief: Global Business School (GBS) Conference Scheduling

1. Financial Metrics

• Staff Opportunity Cost: Manual scheduling requires approximately 40 to 60 hours of senior administrative time per conference cycle. (Paragraph 4)
• Revenue Impact: Attendee satisfaction scores correlate with session availability; a 10 percent drop in satisfaction historically leads to a 5 percent decrease in alumni donations the following year. (Exhibit 2)
• Facility Costs: Room rentals and AV support are fixed at 15,000 dollars per day regardless of occupancy levels. (Exhibit 4)
• Software Budget: The Dean allocated 25,000 dollars for a one-time solution to resolve scheduling conflicts. (Paragraph 12)

2. Operational Facts

• Session Volume: 120 distinct sessions must be scheduled over a 48-hour period. (Exhibit 1)
• Resource Constraints: 15 available rooms with varying capacities from 30 to 200 seats. (Exhibit 3)
• Speaker Requirements: 150 speakers, many with overlapping sessions or restricted availability windows. (Paragraph 8)
• Conflict Frequency: The previous year saw 14 major scheduling overlaps where keynote speakers were booked in two locations simultaneously. (Paragraph 9)
• Current Process: Entirely manual, using spreadsheets and physical whiteboards. (Paragraph 5)

3. Stakeholder Positions

• The Dean: Prioritizes institutional reputation and demands a zero-conflict schedule. (Paragraph 3)
• Event Management Team: Expresses skepticism toward automated tools due to fear of losing control over nuanced speaker preferences. (Paragraph 14)
• Faculty Speakers: Demand specific time slots based on personal travel schedules, often providing data less than 30 days before the event. (Paragraph 11)
• Conference Attendees: Require a personalized experience with minimal overlap between high-interest tracks. (Exhibit 5)

4. Information Gaps

• Historical Preference Data: The case does not provide granular data on which specific session tracks were most popular in prior years.
• IT Infrastructure: Details regarding the compatibility of new optimization software with existing GBS database systems are missing.
• Vendor Options: Specific quotes or capabilities of third-party scheduling software are not detailed.

Strategic Analysis

1. Core Strategic Question

• How can GBS transition from a high-error manual process to an analytics-driven scheduling model that maximizes attendee satisfaction while respecting rigid speaker and facility constraints?

2. Structural Analysis

Applying the Value Chain lens reveals that scheduling is the primary bottleneck in the service delivery of the conference. Errors in this stage degrade the entire value proposition for both speakers and attendees.

• Inbound Logistics: Fragmented data collection from 150 speakers creates a high-noise environment.
• Operations: The manual transformation of data into a schedule is the point of failure.
• Output: A flawed schedule leads to lower attendee retention and reduced donor engagement.

3. Strategic Options

4. Preliminary Recommendation

GBS should develop a custom Integer Programming (IP) model. The complexity of 120 sessions and 150 speakers exceeds the logic capacity of standard event software. An IP model ensures a mathematically optimal solution that eliminates all hard conflicts, directly addressing the Dean's primary concern. This path creates a proprietary asset that GBS can use for all future events, justifying the 25,000 dollar investment.

Implementation Roadmap

1. Critical Path

• Phase 1 (Weeks 1-3): Standardize data collection. Create a mandatory digital portal for speaker availability. No data, no slot.
• Phase 2 (Weeks 4-8): Model development. Build the objective function focusing on minimizing room vacancies and maximizing track continuity.
• Phase 3 (Weeks 9-10): Stress testing. Run the previous year's data through the model to validate accuracy against known conflicts.
• Phase 4 (Week 11): Final schedule generation and automated notification to all stakeholders.

2. Key Constraints

• Data Integrity: The model is only as effective as the speaker availability data. Late submissions will break the optimization window.
• Stakeholder Adoption: The Event Management Team may bypass the model if they perceive it as a threat to their expertise.

3. Risk-Adjusted Implementation Strategy

To mitigate the risk of technical failure, GBS will run a parallel manual schedule for the first 30 days of the project. If the model fails to produce a conflict-free result by Week 8, the manual version becomes the primary. To ensure speaker compliance, the school will implement a firm deadline policy: speakers who do not submit availability by the Week 3 cutoff will be assigned a slot by the algorithm with no option for revision.

Executive Review and BLUF

1. BLUF

GBS must move to a custom integer programming model immediately. The current manual process is a liability that threatens institutional reputation and future funding. By investing 25,000 dollars in a tailored optimization tool, the school will eliminate 60 hours of administrative waste and guarantee a conflict-free experience for 150 high-value speakers. Manual scheduling is no longer viable at this scale.

2. Dangerous Assumption

The analysis assumes that speakers will provide accurate availability data in a timely manner. If the input data remains fragmented and late, even the most sophisticated mathematical model will fail to produce a usable schedule.

3. Unaddressed Risks

• Technical Debt: The model may require expensive annual maintenance that exceeds the initial 25,000 dollar budget as conference requirements evolve.
• Single Point of Failure: If the model is built by a specific data scientist who then departs, the Event Management Team will lack the skills to troubleshoot the system during the live conference window.

4. Unconsidered Alternative

The team did not consider a decentralized scheduling approach where individual track chairs manage their own 10 to 15 sessions within pre-allocated room blocks. This would reduce the central optimization problem to a much simpler task of room allocation rather than session-by-session scheduling.

5. Verdict

APPROVED FOR LEADERSHIP REVIEW