Guide
How to make your best research methods repeatable
A practical way to turn a senior researcher's approach into a skill the whole insights team can reuse, without flattening the judgment that made it good.
Every insights team has a person whose analysis everyone trusts. They know which comparison is fair, which difference is noise, and how to say what a finding means without overclaiming. Their method is rarely written down. It lives in their head and in the decks they have made.
That is a risk and a bottleneck. When they are busy, the work waits. When they leave, the method leaves with them. This guide describes a way to capture a method as a reusable research skill, so more of the team can apply it and the result still looks like your team's work.
What is a research skill?
A research skill is a written, reusable definition of how a specific kind of research work should be done: which inputs to use, which steps to follow, which rules to apply and how the output should be structured. It turns "how our best analyst does brand health" into something any researcher, or an AI research partner, can follow the same way each time.
Step 1: Pick work that recurs
Skills pay off on work you do again and again. Good candidates:
- Tracker waves, such as quarterly brand health.
- Recurring reports, such as a monthly customer insights summary.
- Standard analyses, such as concept test read-outs or churn reviews.
One-off strategic projects are poor candidates. The method changes too much from one to the next.
Step 2: Write down the decisions, not only the steps
Ask your senior researcher to walk through the last time they did the work, and capture each point where they made a choice. The steps are usually obvious. The decisions are where the expertise is.
For a brand-health analysis the decisions might be:
- Which KPIs matter. Awareness, consideration and preference, in that order.
- What to compare against. The previous wave and the category average, never a single competitor.
- What counts as a real difference. Only differences that pass a stated significance threshold.
- What to leave out. Segments with too small a base to report.
Each decision becomes a rule in the skill.
Step 3: Define what a good output looks like
A method is not finished until it says what it produces. Specify the structure, for example headline, evidence, implication. Specify the length, the chart conventions and the terms to use. The fastest way to do this is to attach one or two past deliverables that the team considers good and say why.
Step 4: State the limits
Write down when the skill should not be used and what it cannot conclude. A skill for comparing tracker waves should say that it describes change and does not explain it. This is the part teams skip, and it is the part that keeps a repeatable method from producing confident nonsense.
Step 5: Test it on old work
Run the skill against a study the team has already analyzed. Compare the result with what the senior researcher produced at the time. The gaps show you which decisions you failed to capture. Revise and repeat until the two agree on substance.
Step 6: Keep a named owner
Methods drift. Assign one person to own each skill, review it when the business changes, and retire it when it stops being used. A skill nobody owns turns into a rule nobody remembers the reason for.
What you gain, and what you do not
A skill gives you consistency and capacity. Junior researchers produce work that follows the house method, and senior researchers stop being the queue for routine analysis.
It does not replace judgment. Someone still has to decide whether the question was the right one, review the result, and stand behind it.
Skills in Rose
Rose supports universal research skills and organization-specific skills that define how specialized work should happen. Teams can build a skill around their own KPIs, segment definitions, comparison rules and reporting standards, and apply it to new studies as they arrive. See the product page for how skills fit with the research library.