1. Define the Objective
Every DoE must start with a clearly defined objective. This answers the fundamental question: Why are we doing this experiment?
Typical objectives include identifying critical factors, optimizing a formulation or process, or resolving a specific quality or performance issue. A weak objective leads to weak learning.
2. Identify Factors and Levels
Potential factors that may influence product performance are identified based on scientific knowledge and prior experience. These factors are assigned realistic and justified levels (low, medium, high or target). This step converts formulation or process understanding into testable variables.
3. Select the Experimental Design
The experimental design is selected based on the objective and available resources. Screening designs help identify key factors from many variables, while optimization designs help understand interactions and define optimal ranges. The right design determines the efficiency and quality of learning.
4. Define Responses (Critical Quality Attributes)
Responses are measurable outcomes that reflect product quality or performance. These are typically aligned with CQAs such as dissolution, assay, viscosity, or stability attributes. Clearly defined responses ensure experiments are relevant to real-world quality expectations.
5. Conduct the Experiments
Experiments are executed strictly according to the design matrix. Runs are randomized to reduce bias, data is recorded systematically, and repeat runs are included where necessary to assess variability. Good execution is as important as good design.
6. Analyze the Data
Collected data is analyzed using statistical tools to understand main effects, interactions, regression relationships, and factor significance. Numerical indicators such as ANOVA help quantify which variables truly matter.
7. Interpret Results (Visualization)
Statistical outputs are translated into practical insights using graphical tools such as main-effects plots, interaction plots, contour plots, and 3D response surfaces. Visualization makes complex relationships easier to understand and communicate.
8. Establish the Design Space
Based on numerical optimization and graphical interpretation, the design space is defined. This represents the multidimensional combination of factors where quality is consistently achieved.
9. Confirm the Findings
Confirmation runs are performed to verify model predictions. This step ensures that optimized settings are reproducible, robust, and scalable. It converts statistical models into experimental proof.
10. Implement Control Space and Monitor
Finally, optimized conditions are implemented in routine manufacturing. Monitoring and control strategies are established to ensure consistent performance throughout the product lifecycle. This closes the QbD loop from development to commercialization
Read also: DoE Application in Formulation Development
Resource Person: Moinuddin Syed. Ph.D, PMP®