What-If Analysis with Simulation

What-If Analysis Using Simulation

What-if analysis uses simulation to compare multiple scenarios before you invest time and resources in a change. By creating scenarios and comparing their results, you can make informed decisions about process changes. You can run a what-if analysis on your own process with a free trial.

Comparing Scenarios

Simulation alone provides valuable predictions, but the real insights come from comparison:

• Compare before and after a proposed change
• Compare simulated results with actual historical data
• Compare multiple alternative scenarios

Comparing Datasets in ProcessMind

Because simulation output is a standard event log dataset, ProcessMind provides versatile tools for comparing any datasets, whether simulated, historical, or both.

How to Compare Datasets

1. Make sure the datasets you want to compare are mapped to your model (see mapping datasets).
2. Set one dataset as the primary dataset and another as a comparison dataset (see comparing data).
3. Open your dashboards: charts show the primary and comparison datasets side by side.

What Can You Compare?

Metric	Description
Throughput Time	Total time from case start to completion
Waiting Time	Time cases spend waiting for resources
Processing Time	Actual work time for activities
Case Count	Number of completed cases
Queue Length	Number of cases waiting for a resource
Path Distribution	Which paths cases take through the process

Before/After Analysis

The most common what-if scenario is testing a process change:

Scenario: Adding Resources

Question: What happens if you add one more approval staff member?

Approach:

1. Create a baseline simulation with current staffing
2. Create a modified simulation with increased capacity
3. Compare the results

Example Results

Metric	Baseline	+1 Staff	Improvement
Avg. Throughput Time	5.2 days	3.8 days	27% faster
Avg. Waiting Time	2.1 days	0.9 days	57% reduction
Cases/Week	150	195	30% more
Avg. Queue Length	64 cases	16 cases	75% shorter queues

Comparing with Actual Data

A key feature of simulation is comparing simulated predictions with real historical data.

Why Compare to Actual Data?

1. Validate your model: If simulation matches historical data, your model is accurate
2. Identify gaps: Differences reveal where your model needs refinement
3. Measure improvements: Compare new actual data against simulated predictions

How to Compare Simulated vs. Actual

1. Run a simulation for a historical time period
2. Upload or select your actual historical data for the same period
3. Use the Compare feature to view differences

Interpreting Differences

If Simulation Shows…	Possible Cause
Faster than actual	Model missing delays, underestimated complexity
Slower than actual	Overestimated processing times, unnecessary constraints
Different paths	Gateway probabilities don’t match reality
Resource overload	Capacity constraints configured incorrectly

Multi-Scenario Comparison

Compare more than two scenarios to find the optimal configuration:

Scenario: Finding Optimal Staffing

Question: How many approval staff do you need?

Approach: Create multiple simulations with varying staff levels

Example: Staffing Level Analysis

Staff	Throughput	Queue Length	Wait Time	Cost
2	100 cases/wk	64 cases	4.5 days	$120K
3	145 cases/wk	37 cases	1.8 days	$180K
4	160 cases/wk	11 cases	0.5 days	$240K
5	165 cases/wk	5 cases	0.2 days	$300K

Insight: Adding a 3rd staff member provides the best value. The 4th and 5th show diminishing returns.

Common What-If Scenarios

Process Redesign

Test structural changes to your process:

• What if you remove a manual approval step?
• What if you add a quality check?
• What if you parallelize sequential activities?

Demand Changes

Prepare for different demand levels:

• What if demand increases by 50%?
• What if you experience seasonal peaks?
• What if a marketing campaign doubles incoming cases?

Resource Changes

Optimize your workforce and systems:

• What if you cross-train employees for multiple activities?
• What if you automate a manual task?
• What if you outsource certain activities?

Process Timing

Explore schedule and timing impacts:

• What if you extend operating hours?
• What if you add a weekend shift?
• What if you stagger lunch breaks?

Best Practices for What-If Analysis

1. Establish a Baseline

Always create a validated baseline simulation before testing changes. This provides your point of comparison.

2. Change One Variable at a Time

When possible, isolate variables to understand their individual impact. Changing multiple things simultaneously makes it hard to attribute improvements.

3. Use Realistic Parameters

Base your simulation parameters on actual data when available. Unrealistic inputs lead to unreliable outputs.

4. Consider Multiple Metrics

Don’t optimize for a single metric. Faster throughput might come at the cost of quality or employee burnout.

5. Document Your Scenarios

Keep clear records of what you changed in each scenario. This makes results reproducible and shareable.

6. Validate Before Acting

When possible, validate simulation predictions with small-scale tests before full implementation.

Presenting Results

When sharing what-if analysis results with stakeholders:

• Lead with the question: What business problem are you solving?
• Show the comparison: Visual side-by-side metrics make differences clear
• Quantify the impact: Translate to business value (time saved, cost reduced, and revenue gained)
• Acknowledge uncertainty: Simulation is a prediction, not a guarantee

Next Steps

Interface Reference

Complete reference for all simulation settings and options.