Internal Research Workflow Template

A defined route from research request to checked output: how a brief is taken, what gets agreed before work starts, how the work is done and checked, and how it comes back. Most bad research output is a briefing failure rather than a research failure.

What this is for

Research requests arrive informally, in a message or across a desk, with the requester holding most of the context in their head. The work comes back not quite right, and the second attempt costs more than the first.

A defined workflow front-loads the disagreement to the point where it is cheap.

When to use it

When output regularly comes back needing rework
When several people request research and each does it differently
Before handing research to anyone new
When nobody can say how long a piece of research should take

What it covers

The brief. Captured in writing, in one place, covering: what decision this supports, what specifically is needed, the scope boundary, the format, the deadline, and which sources are acceptable. A verbal brief is not a brief.
Scoping check. Before work starts, the researcher confirms back what they understand the request to be and flags anything ambiguous. Five minutes here saves the rework.
Source agreement. Which licensed sources, and what to do when they disagree. Agreed at the start, not argued about at the end.
The work. Against the agreed scope. Anything outside it gets flagged, not absorbed.
Recording the unknowns. Gaps are logged as the work proceeds rather than filled in at the end.
Quality check. Against the brief and against your process, by someone other than the person who did the work. Checking whether it answers the question asked, not just whether the facts are right.
Return. In the agreed format, with sources, dates and open questions visible.
Feedback loop. Where output needed rework, the reason gets recorded and the documentation updated, so the same gap does not recur.

Where it usually goes wrong

The unwritten brief. The commonest failure by a distance. If it is not written down, two people are working to two different requests.

No scope boundary. Research expands to fill the time unless somebody says where it stops.

Self-checking. The person who did the work is the worst person to check whether it answers the question.

Silent gap-filling. A researcher who cannot find something and quietly estimates it has introduced an error nobody knows about.

Feedback that goes nowhere. Rework that is not recorded is rework that will happen again.

What good looks like

A request can be picked up by any trained person and come back the same. The requester can tell from the output what is verified and what is not. Rework is rare, and when it happens the reason is documented.

Related resources

What Is Data & Research Operations?

Building and maintaining a target universe, and the factual research that comes off it.

Company Profile Template for Deal Calls

A one page structure for company profiles, covering what a deal team actually needs before a call.

Taxonomy and Classification: Getting It Right

Why classification decisions matter more than they look, and how to make them consistently.

Is your research process the same every time?

Talk to us