From assumptions to audience-first research
Discovery for a mobile-first, new-to-the-internet audience
Context
A global technology company had developed a visual search app for an emerging audience: people new to the internet, typically accessing it for the first time through a mobile phone. They came to us with two asks: help them understand this audience better, and evaluate the first version of their prototype.
I led the research across both phases, bringing first-hand understanding of this audience from my background working in India as a development professional, which shaped how we defined the recruitment criteria and framed the research from the start.
The brief
Explore the information needs of people new to the internet. Map how they search, what they search for, and what they expect from a visual search product and evaluate the prototype against those findings.
The real problem
What we uncovered through research was a more fundamental gap than the brief had anticipated.
The prototype had been built on an initial set of assumptions about how this audience would behave online: reasonable starting points given what was known at the time. But the research revealed that the mental model underlying those assumptions didn't match how this audience actually experienced the internet.
These users weren't explorers but they were more searchers: highly intentional, goal-driven, coming to the internet with a specific question and leaving once they had an answer. Or more often, leaving without one; because the information they needed was hyper-local, hyper-specific, and simply didn't exist in any form the internet could deliver.
The product had been designed for discovery. The audience needed solutions and specific answers.
That gap between assumed behaviour and actual behaviour became the central finding of the research.
Considerations for the work
Reaching a hard-to-reach audience
This audience was not well researched. Low digital literacy, limited internet access, and culturally specific contexts meant standard research approaches wouldn't work. Every methodological decision had to be made with the audience's actual capabilities in mind beyond just the right or feasible research methodology.
The pandemic constraint
The research was conducted remotely during the pandemic, ruling out in-person fieldwork entirely and considerably reducing our options. Finding the right tool to conduct research with a low-literacy, low-digital-access audience remotely was the first design challenge we had to solve before we could even begin researching.
My India background as a research asset
Having worked in India as a development professional, I brought firsthand understanding of the socio-economic, cultural and urban-rural context of this audience. That grounding shaped how we defined the recruitment criteria and helped us ask better questions from the start.
Balancing prior knowledge with new evidence
The client had existing knowledge about this audience from previous projects. Part of our role was to build on that foundation while being rigorous about where the evidence pointed in a different direction which required careful relationship management alongside honest research.
Discovery research — Framing the right questions
We began by defining the audience more precisely. The client understood that this group had limited internet experience but during recruitment we discovered that most Android phones come pre-installed with Google products, making a truly "never used search" audience almost impossible to find in practice.
Rather than chasing a user who didn't exist, we reframed the criteria; focusing on distinguishing needs-based, sporadic searchers from habitual, always-on users. That distinction turned out to be far more useful and honest than the original segmentation.
After ruling out multiple tools, we chose WhatsApp as our primary research platform. It was already familiar to the audience, its always-on nature enabled diary studies and digital ethnography alongside interviews, and it didn't require users to learn a new tool just to participate. This was a methodological decision in itself choosing the right tool for the audience, not the most convenient tool for the researcher.
What we did
Diary studies over WhatsApp (n=50)
Participants documented their information-seeking behaviour in their own words and their own time giving us a window into how they actually used the internet rather than how they said they used it.
Moderated focus groups (n=10 x 3)
Smaller group discussions to explore information needs, search behaviour and the role of technology in participants' daily lives.
Digital ethnography
Observing users in their natural digital environment understanding the context in which they searched, what constraints they faced, and what the internet actually meant to them.
Remote usability interviews (n=15 x 2 rounds)
High-fidelity prototype testing conducted remotely evaluating not just whether users could use the product, but whether they would, when, why, and instead of what.
What we found
The mental model mismatch
When people in this audience came to the internet, their queries were highly specific and practical focused on immediate, everyday needs. Not browsing. Searching. And often for things the internet simply couldn't answer.
A user wanting to know why potatoes in a village one kilometre away were cheaper couldn't find that answer anywhere online. The internet didn't contain the information they needed and a better designed app wouldn't change that.
The always-on assumption was wrong
The product was built around a discovery-oriented, always-on browsing mindset but this audience was intentional and goal-driven. They came with a question, looked for an answer, and left. The product's design assumed a behaviour pattern that this audience simply didn't have.
Features that delighted but didn't solve
Users genuinely liked voice-to-text and voice search, typing was slow and effortful, so voice felt useful. Categorised results and video content delighted them. None of these features resolved the fundamental problem: the information this audience needed most often didn't exist in a form any app could deliver.
Reframing the audience
The assumption that this audience had never used Google search turned out to be almost impossible to satisfy android pre-installation meant most had encountered it. The more useful distinction was between needs-based, sporadic users and habitual browsers. That reframing changed how the client thought about who they were building for.
What changed
The research findings led to a fundamental reconsideration of the product proposition. Development was halted saving the client significant time and resource by avoiding further investment in a direction the evidence didn't support.
The insights were embedded into the understanding of over 100 UX researchers across the client organisation giving a much broader team a grounded, evidence-based picture of who this audience actually was and how they behaved online.
The brief itself shifted: The question was no longer how to improve the product. It was whether this product, for this audience, at this moment, was the right thing to build at all.
100+ UX researchers
Evidence embedded across the organisation
Reflection
This project reinforced something I've come to believe across every piece of research I've done is that the most important contribution research can make is not validating what a team already suspects, but surfacing the gap between assumed behaviour and actual behaviour before that gap becomes expensive.
The client came to us with reasonable assumptions based on real prior knowledge. Research didn't prove them wrong, it showed where the assumptions ran out, and what needed to be known before the next decision could be made well.
That's what good research does: it doesn't just answer the question you asked, it also shows you which question you should have been asking.