Framework Analysis for UX Research

WHY IT EXISTS: Raw qualitative data is noisy and high-dimensional. Without structure, researchers can drown in transcripts and struggle to justify design decisions to stakeholders who want clear links to business goals or existing theory. Framework Analysis was developed to impose order without starting from zero. By borrowing categories from established models, prior research, or domain standards, teams move faster and produce findings that map directly to metrics and mental models the organization already understands.

THE MENTAL MODEL: Think of it as sorting laundry with pre-labeled bins rather than dumping the basket and deciding what piles to make. You already know you need lights, darks, and delicates. Most items fit neatly, but a few will challenge your categories and force you to add a new bin. The framework is a starting hypothesis, not a prison. It gives you a stencil to trace over the mess, but the outline is allowed to change where the data demands it.

HOW IT WORKS: The method follows five stages. First, familiarization: immerse yourself in the data through transcripts, recordings, and notes until you sense the landscape. Second, framework identification: select or build a set of categories derived from research questions, existing theory, or business objectives. Third, indexing: systematically code data segments against these categories, tagging where they fit and flagging where they resist. Fourth, charting: lift the coded segments into a matrix or table so you can compare participants and themes side by side. Fifth, mapping and interpretation: look for associations, contradictions, and gaps to answer your research questions. Crucially, the framework is provisional. If data repeatedly breaks a category, you split, merge, or add themes rather than distorting the evidence.

WHEN TO USE IT: Use it when your research question is tethered to known constructs, such as evaluating trust against a security framework or usability against Nielsen's heuristics. It excels in large-scale qualitative projects, longitudinal studies, and mixed-methods work where qualitative insights must align with quantitative variables or regulatory requirements. It is also valuable when stakeholders need findings expressed in the language of existing business or academic models.

WHEN NOT TO USE IT: Avoid it in pure generative discovery where preconceived categories could blind you to unexpected behaviors or mental models. If you are exploring an entirely new domain and have no validated framework, forcing an ill-fitting model onto user stories will create blind spots and false confidence. It is also the wrong tool when team consensus on the framework itself is missing; a contested stencil produces inconsistent coding.

ONE CANONICAL EXAMPLE: A healthcare UX team interviews patients about portal adoption and uses the Technology Acceptance Model as their framework. They index transcripts against perceived usefulness and perceived ease of use, charting quotes in a matrix to see patterns across age groups. The framework confirms that older patients struggle with navigation, but it also reveals a gap: privacy anxiety does not fit the original TAM dimensions. The team adds a trust category, demonstrating both the power of the lens and the necessity of adapting it.