Reengineering the development of personas through a user-centred human machine teaming approach

Personas are widely used in user-centred design to capture users' goals, needs, behaviours, and pain points. They serve as narrative user models which represent a group of users and play a crucial role in various domains such as system design, software development, marketing, and content dissemination. Personas were initially created as static representations in the form of documents, providing valuable insights into the target user group. However, as technology has advanced and user needs evolved, there has been a shift towards representing personas as dynamic digital entities.

Traditional methods of persona creation rely on data-driven approaches for user research, such as interviews and surveys, require significant resources and training. They often use data that does not reflect the diversity of the user group, nor do they adapt to their changing needs; raising doubts about the value and validity of static personas. Recent advancements in generating digital personas, which use secondary user data and algorithms, offer scalability and efficiency but lack end-user and business team involvement. To address these challenges and limitations, this research focused on re-evaluating and re-engineering the persona creation process from a practitioner-centred perspective. This research targeted students and novice practitioners in the fields of UX, IxD, and HCI, and aimed to enhance the effectiveness and relevance of persona development methods within the user-centred design approach. It considered the evolving nature of personas and the opportunities for collaboration between humans and machines.

This PhD thesis employed a mixed-methods approach to investigate persona creation. First, a comprehensive literature review identified gaps in traditional methods and studies on static persona representations. This went on to inform the creation of a survey which gathered quantitative data on the attitudes and practices of UX/IxD practitioners. Data was analysed using significance testing, odds ratios were detailed, and an exploratory factor analysis was carried out. The findings indicated the benefits of personas, identifying and prioritising challenges related to resources needed for persona creation, such as the availability of trained personnel, data analysis expertise, and access to data. In-depth interviews further validated and expanded on survey findings, employing a thematic analysis to uncover new opportunities for persona development, particularly emphasising the need for personas to be up-to-date, reusable, and trustworthy. Additionally, investigations into advancements in automated persona generation aligned with earlier empirical findings, emphasising the need to re-engineer the process of persona development.

The work presented in this PhD explored the concept of human-machine teaming to understand the role each party (the human and the machine) can play in re-engineering the process of persona construction, resulting in the development of the practitioner-centred Persona Process Model (PPM). The PPM incorporates three distinct approaches for persona construction: data-driven, assumption-driven, and recommendation-based; integrating interactive visual analytics and automated processing methods in dimension reduction and clustering. The Persona Toolkit (PT), a micro-level proof of concept for the PPM, was developed and evaluated against Cooper’s supported by Excel as a reference approach. The evaluation utilised a series of objective and subjective metrics targeting key properties of the PPM micro-model, such as analysis workload, trustworthiness, and efficiency. The evaluation also included a review of the utility of the metrics employed. The study supported the conclusions that: (1) with minimal training, novice UX teams can efficiently create trustworthy personas when using the toolkit, and (2) the Persona Toolkit appears to reduce the data analysis workload persona creation. These findings lend support to the value of the PPM model.

The research outlined in this thesis presents contributions across literature, academia, and industry. The PPM and Persona Toolkit provide a valuable streamlined solution suited to the needs of novice practitioners and students, providing them with a practical and effective approach to persona creation. These findings can have impact on academic curricula and educational programmes, as well as industry practices Additionally, the findings highlight the broader applicability of persona creation across various domains and technologies, suggesting its utility extends beyond traditional user-centred design.

In accordance with the ethical approval that was granted for this study, the datasets used for testing and evaluating the Persona Toolkit in this thesis (in formats such as .csv and .xls) are available upon request. These datasets can be accessed by contacting the author directly. Additionally, the datasets are included in the appendix of this thesis for reference.
The code developed and used for the Persona Toolkit in this thesis is not publicly available.
Access to the code will be granted on a case-by-case basis, subject to approval, and for specific academic or research purposes. Any requests for access to the code should be made in writing to the author, providing a clear explanation of the intended use.
Requests for further information can be directed to e.bashir@mdx.ac.uk.