Measurement and validation in an age of abundance.
AI can answer almost anything. The harder question is which answers deserve to guide a decision.
I build systems that combine AI scale with research discipline, producing insights organisations can trust and act on.
For over a decade, I measured what people think. The same problem has now shifted to machines: understanding what models know, where they are reliable, and where they fail.
The future of research pairs people with AI. The skill is knowing what to ask of each. That is the work I do.
Two pieces of thinking in progress. One on where the ground is shifting. One on what comes next.
The tools are converging and the insight category is consolidating. For an increasing set of high-stakes decisions, the AI-mediated layer is where category beliefs now form. No one owns this space.
In practice. One half is measurement applied to brands, categories, audiences, and channels. The other, more open-ended half is improving how human data is collected.
AI brand tracking
An application of synthetic systems research. It measures how AI models represent key category attributes, comparatively across brands and over time. The data is illustrative; the method is the point.
Open the demonstration
Quantitative human-data collection
An embedded prototype. A rapid choice between two options captures preference and strength of feeling, while interaction patterns help verify a real respondent. A faster, more human alternative to survey grids.
Open the prototype
Qualitative human-data collection
A seed. A voice-first companion to the swipe tool: a short “vibe” prompt arrives in your inbox, followed by three unseen prompts in sequence. Thirty seconds to think, sixty to respond, then it disappears. Capturing in-the-moment feeling rather than polished answers.
Read the approach
Audiences and channel analysis
The same measurement discipline applied to people and reach: who is being reached, and how. I treat audiences as populations that can be characterised, and channels as audiences with a subject, each tested as a designed experiment rather than a guess.
Under development
I build systems that turn AI outputs into trustworthy answers you can act on.
Health literacy tool development
At Deakin University I led the development of the Conversational Health Literacy Assessment Tool (CHAT), distilling a long assessment framework into ten questions a health worker can use in clinical conversations. It has since been adopted and adapted by other researchers and health services.
Read the paper, BMC 2018
5 years at RCH National Child Health Poll
I was the analyst for 18 public releases of the Royal Children's Hospital National Child Health Poll, 2015 to 2020. I built reproducible analysis pipelines that turned each national survey around on a tight media cadence, and let me rapidly answer clinicians' follow-up questions, so the discussion could move from the headline number to what it meant for care.
See the polls
AI applied at commercial scale
In a company I co-founded (2023 to 2026), I built and led development of a platform designed to bring research-grade rigour to AI-generated market research insights. I owned the full chain, from product concept and data architecture through to analysis and the web tools that turned complex research into action for non-technical teams.
I'm building this out now. If there's something here you want to deliver on, let's talk.
