There is a version of research that happens at a desk. It is careful and clean: interviews are scheduled, questions prepared, workflows mapped. You build a picture of a system: logical, sequential, legible. A patient arrives, a clinician assesses, a diagnosis is made. The pieces fit. Then you step into an actual clinic, and the picture looks rather different.
The Gap
We arrived in Kenya with a plan: facilities to visit, questions to ask. Within days, we came to understand that meaningful fieldwork doesn't follow plans: it follows people.
A clinician would finish a sentence and add, almost as an afterthought: "You should speak to my colleague at the other facility. She sees this every day." A door would open that no recruitment brief had anticipated, and important conversations happened in corridors, between consultations, arranged through trust rather than protocol.
This is what researchers call the snowball method. In practice, it felt less like a method and more like a way to understand a system that isn't designed to be observed.
What people say and what people do
In interviews, workflows tend to sound structured. Ask a clinician to describe their process and they'll give you a sequence: patient arrives, history taken, examination done, treatment prescribed. It's how the work is meant to go.
Shadowing tends to tell a different story.
Decisions get made under pressure, with incomplete information and frequent interruption. A consultation pauses for a phone call. A photo of a skin lesion gets sent over WhatsApp to a specialist two towns away. Patient notes are written up after hours, long after the patient has left. Tools are improvised. Gaps are filled, without anyone formally acknowledging that this is how it works. But these are not exceptions. They are the system.
What people say they do reflects how they understand their work. What they actually do reflects how the work gets done. And the space between those two things, that gap, is perhaps where design most needs to pay attention.
The real system
Look at healthcare from a distance and it appears structured: institutions, roles, protocols, tools. Get close enough, and much of what makes the system function turns out to be invisible.
A messaging app becomes a diagnostic support network. A handwritten notebook becomes the de facto patient record. A trusted colleague becomes a second opinion, a reference point.
In under-resourced settings, people don't wait for the right tools to arrive. They build what they need from what they have. These informal systems aren't failures of the formal one, they are adaptations to it. They tend to be fast, flexible, and often remarkably effective.
Designing new technology without first understanding these existing practices carries real risk. You may disrupt the very mechanisms keeping the system running. The goal, then, is not to replace what people have built, it is to understand it well enough to design something that fits alongside it.
Context is not a constraint
From the desk, context can look like a list of limitations: unreliable connectivity, equipment shortages, inconsistent training.
In the field, context is the ground everything stands on. A tool that assumes stable internet access will fail. A model trained on data that doesn't represent the population it serves will produce outputs no one can trust. A workflow that assumes standardised terminology will break the moment two clinicians use different words for the same condition.
Designing for real-world contexts doesn't mean lowering the bar, it means raising the questions. What does this person actually have in front of them? What do they know, and when do they need to know it? What does the room look like, and how long does the consultation last?
Good design in constrained environments is about accurate alignment with reality.
And fieldwork in healthcare serves as a further reminder that these environments are not only clinical, they are also deeply social.
In some contexts, a disease doesn't just signal illness; it carries stigma, triggers fear. People arrive with beliefs about why they are sick that have little to do with clinical science and everything to do with lived experience.
These factors shape when patients seek help, how they receive information, and whether they follow advice. They also shape how clinicians communicate: what they say, what they leave unsaid, how carefully they choose to frame a diagnosis. Designing for healthcare means taking all of this into account. Not just the symptom and the treatment, but the hesitation before the patient speaks.
What the field actually changes
Fieldwork provides data, but that is not quite the whole story.
What fieldwork does, perhaps more than anything, is change how you see. It dismantles the assumption that the system you imagine is the system that exists. It reveals the invisible labour, the workarounds, the human ingenuity operating beneath every formal process.
At the desk, it is easy to design for the world as we wish it were: coherent, predictable, rational. In the field, you encounter it as it is: adaptive, constrained, and irreducibly human.
And that, in the end, might be the most important shift. Because the best design doesn't impose a new order onto a system, it listens to the one already there, and finds a way to make it a little more true to the people living inside it.
The work described here was conducted as part of field research in Kenyan healthcare settings for the SkincAIr project, exploring how clinicians navigate diagnostic and documentation challenges in resource-constrained environments, with a focus on skin neglected tropical diseases.
