Research scoped around the decision you need to make.

Should we build this, and for whom?

When it fits

Before a new product, feature or market, when the team has opinions and no evidence yet. Also before automating a workflow, when nobody has written down how the work happens.

What I do

• A kickoff workshop to name the decisions the research must inform
• Stakeholder interviews, from the leadership team to the people doing the work
• Customer interviews and contextual inquiry, in the place the work happens
• Synthesis from theme to evidence to impact

What you leave with

• Findings backed by evidence, every quote traced to its source
• Opportunity areas and the risks beside them
• Recommendations sorted into quick wins, pilots and roadmap items
• A readout for your leadership team

In practice: stakeholder ethnography for Hearthy Foods, in-context interviews for a SaaS beta, an operations audit before automating a business.

You have a direction. Test it before you commit a quarter to it.

When it fits

A concept exists, the team is divided or over-confident, and the next step is expensive. A sprint turns the argument into a prototype and the prototype into evidence in a week.

What I do

• Facilitate the sprint, from mapping the problem to sketching and deciding
• Recruit the right participants while the prototype is built
• Run the test sessions on the final day and synthesize that evening
• Coach your designers and researchers to run the next one

What you leave with

• A tested prototype and the recordings behind it
• A clear go, stop or change-course recommendation
• The questions the sprint raised that need a longer study

In practice: a multi-day design sprint for ResMed's digital-health teams, with the designers and junior researchers coached to run the next one.

Who are our users, really?

When it fits

Marketing, product and engineering describe different people, or decisions keep stalling on "it depends who." Behavioral personas built from evidence give everyone the same person to design for.

What I do

• Review your analytics and existing data first
• A segmentation survey, typically 50 to 150 responses
• Interviews across the segments the survey reveals
• A card sort, if navigation or categories are in question

What you leave with

• Three or four behavioral personas, built from what people do rather than demographics alone
• A current-state journey map with the pain points and the moments that matter
• An opportunity map and a readout

In practice: three personas behind a strategy covered in Bon Appétit; a 50-person segmentation that ended a deadlock at JPMorgan Chase.

Is the AI feature actually helping? Do people trust it? What should it be allowed to do?

When it fits

You've shipped or prototyped an AI feature and need to know whether it works for the people using it. Or you're about to automate part of a business and want to understand how the work happens before a tool touches it.

What I do

• Evaluation: a rubric that scores preserved intent rather than valid-looking output, and separates how often errors happen from how much they matter
• Adoption: interviews, usage data and prototype tests on trust, checking behavior and where people stop using the feature
• Automation: contextual inquiry across roles and systems, then a map of what AI may read, draft and never touch

What you leave with

• A reusable evaluation rubric and the evidence from running it
• A definition of "useful" your team can hold the feature to
• For automation work: a workflow map, a problem list with owners, and a governance rule for what a person must approve

In practice: study design for a SANS-published evaluation of two AI models; defining "useful" for version 1 of LLM Suite at JPMorgan Chase; five governed agents for a precious-metals dealer.