Researching Inside
Institutional Complexity
Without Breaking It
How I applied HCI rigour to an expert-driven dashboard project — navigating ethics, stakeholder politics, and scope constraints in a public health sector.
PROJECT
Public Health Surveillance Dashboard - Monitoring AMR From Wastewater
MY ROLE
Lead User Researcher
KEY USERS
PHW clinicians, microbiologists, AMR specialists, WWTP staff
ORGANISATION
University of Sheffield (In collaboration with Public Health Wales)
METHODS
Contextual Inquiry | Stakeholder Interviews | Survey
STANDARDS
GDS Service Standard
01
The challenge I walked into
A dashboard was already in progress, but built without user understanding, risking assumption-driven design that could ultimately go unused.
!
The upstream problem
Development had accelerated through collaboration with a single epidemiology expert. Early HCI steps: user profiling, workflow mapping, stakeholder analysis, were skipped in favour of speed. As a result, design decisions were based on assumptions rather than evidence.
?
What this meant for research
The project initially planned to evaluate an early dashboard prototype. However, without understanding who the real users were, how wastewater data was interpreted, or what decisions the dashboard should support, usability testing would have evaluated the wrong thing.
↔
Stakeholder landscape
Access to participants was limited. Public health professionals were busy and difficult to engage, meaning the research approach had to be lightweight and carefully designed to obtain meaningful insight without adding burden.
⚠
What was at risk
Without grounding the design in real user workflows, the dashboard risked being built on assumptions rather than evidence. A misframed project at this stage could lead to:
-
a product that does not fit real public health workflows
-
wasted development time and research effort
-
erosion of stakeholder trust in the project
→
How I reframe the research problem
Moving straight into usability evaluation was premature: the team lacked clarity on who the real users were, how wastewater data fit into their workflows, and what decisions the dashboard should support. I communicated this risk to stakeholders—that evaluating a prototype built on unverified assumptions could produce misleading results and wasted effort—and proposed a lightweight discovery approach to quickly identify users, workflows, and decision contexts without delaying delivery.
02
The research pivot: from V1 survey to expert-first approach
The original long-form survey proved unrealistic for busy public health professionals. I restructured the research plan, introducing expert scoping sessions to establish users, workflows, and decision contexts before scaling data collection.
I started with a survey to quickly surface an understanding of users’ roles and workflows. After revising two versions within a day with stakeholder feedback, the difficulties of engaging busy public health professionals became clear. This rapid iteration surfaced the core constraints of the project: limited participant availability and pressure to deliver quickly.
Rapid survey design iteration (day 1-3)
Problem Identified
Stakeholder Buyin
GDS Standard 1
The survey iterations surfaced key unanswered questions and helped stakeholders recognise the value of addressing them. In response to this feedback, I separated expert scoping from broader survey work, introducing expert sessions to establish users, workflows, and decision contexts before redesigning the survey approach.
Structural insight (Day 4)
Pivot Decision
Two sessions with an epidemiologist as proxy expert established: primary user identification, concrete use scenarios, co-sketched workflows, and information requirements at each decision step.
Expert scoping sessions (1-hour each)
Outputs from the expert sessions directly informed a refined study design: a shortened V3 survey combined with a workshop for structured user feedback. This approach generated richer qualitative insight than a survey alone while remaining feasible within the project’s delivery constraints.
V3 Design (current) — two-instrument approach
Research Strategy
Formative Research
Contextual Inquiry
GDS Standard 8
Evidence Generation