Skip to main


Persona Automation


Users need...

personalized support for leveraging data and relating tools.

The Canadian Institute for Health Information (CIHI)


Users struggle to assess how CIHI data could be useful for their projects.

Customizing user experiences to improve relevancy is not straightforward. Many users have multiple roles and their needs in CIHI tools can dynamically change per visit.

*Note: Before I joined the project, this analysis was planned to be completed manually with Excel. With my programming background, I identified an opportunity to leverage automation techniques.

Data Scrubbing with Python

  • Extracted responses to open-text, demographic questions from recent exit-intent survey responses
  • Cleaned data for processing
    • Assessed response validity, excluded invalid responses
    • Adjusted for spelling variations
    • Translated French-language responses

Mapping Users to Personas with Natural Language Processing

  • Compiled keyword list from open-text responses relating to persona characteristics
  • Iteratively categorized respondents per persona
    • Automatically mapped keywords to survey response
    • Manually assessed unmapped response for potential keyword list improvements
    • Improved mapping algorithm
      • E.g., Updated word stems, considered medical terminology, and considered vernacular variations
  • Ended processing when 87% of the survey responses could be accurately categorized

Expanding the Persona Framework

  • Once respondents were categorized, full responses were further analyzed
  • Validated the existing persona framework using mapped responses
    • E.g., Persona A most often uses the XYZ tool on the website
  • Extracted new user insights and integrated findings in the persona framework
    • E.g., Persona B is most likely to report XYZ pain points
  • Provided data-driven recommendations for usability improvements to the CIHI website

  • Excel
  • Python
  • VS Code