The Junior's Research Workflow: From Assignment to Defensible Answer

A senior looks up from a brief and says: find out whether we can do this. The junior goes away, works for six hours, and returns with eleven judgments, a summary of each, and no answer. The senior reads the first two, asks a question the junior has not considered, and does the research themselves in twenty minutes. Both people leave the exchange dissatisfied, and neither is at fault.

This scene repeats in chambers across the country every week, and it is expensive in a way that never appears on any statement of account. The six hours were paid for by somebody, in salary or in the senior's unbilled time or in the client's fee, and they produced nothing usable. Multiply by the number of juniors and the number of assignments and the loss is the single largest inefficiency in most litigation practices.

The instinct is to attribute it to inexperience. That is only partly true. Most of it is the absence of a workflow. Nobody told the junior what a finished piece of research looks like, how long each stage should take, or that stating uncertainty honestly is part of the deliverable rather than an admission of failure. This piece sets out that workflow, with time budgets, in the order the work should actually be done.

Why Unstructured Research Fails

Before the pipeline, the diagnosis. Research done without structure fails in a small number of consistent ways, and every junior will recognise at least three.

Notice that only two of these are about legal knowledge. The rest are about process, sequencing and honesty, all of which can be taught in a morning and enforced by a template.

Stage One: Scope the Question. Twenty Minutes.

The highest return twenty minutes in the entire process, and the stage juniors most often skip because going back to ask feels like an admission that they did not understand. It is the opposite. A junior who returns with three clarifying questions is demonstrating that they have thought about the assignment.

The question a junior was asked and the question a junior answered are different documents. Two lines sent before starting is what keeps them the same.

Stage Two: Search Wide Before You Search Deep. Forty Minutes.

Coverage first, conclusions later

The purpose of this stage is coverage, not answers. You are trying to establish the shape of the landscape: is this settled, is it contested, has your own High Court decided it, is there a divergence between High Courts, and has there been a recent development.

• Start with the statutory provision. Read the section as it currently stands, and confirm whether it has been amended, when, and whether the amendment applies to your facts. A surprising share of research errors are amendment errors rather than case law errors.
• Search by proposition, not by keyword alone. The vocabulary in your head and the vocabulary in the judgment are often different, and keyword search is unforgiving about that. Searching by the concept you actually need is what surfaces the judgment that uses different words for the same idea.
• Move down the hierarchy deliberately. Supreme Court first, then your own High Court, then other High Courts, then subordinate forums and tribunals where relevant. Record what you find at each level, including the fact that you found nothing, because absence at a level is itself a finding.
• Look for the contrary line. Actively search for the position against you before you get attached to the position for you. If there is a divergence between High Courts, you need to know it now rather than in court.
• Note the recent decisions. Judicial positions move. A line of authority that was settled five years ago may have been revisited, and a recent decision of your own High Court is worth more than an older decision from elsewhere.
• Keep a search log. What you searched, where, and what it returned. This costs nothing while you work and it is what lets you say honestly, at the end, how thoroughly the ground was covered.

Forty minutes is enough for this stage if the tooling is right and hopeless if it is not. A junior searching four separate portals with different interfaces will spend the whole budget on navigation. Unified search across the Supreme Court, 25 High Courts, District Courts and Tribunals is what makes a wide first pass a forty minute exercise rather than an afternoon.

Stage Three: Read Properly and Verify Currency. Ninety Minutes.

Now narrow. Of everything the wide pass surfaced, four or five judgments will actually matter. Read those properly, and read them for a specific purpose rather than for general understanding.

Currency verification is the stage that most distinguishes reliable juniors from unreliable ones, and it is entirely mechanical. It requires no judgement, only the discipline of checking every authority before it enters the answer. A junior who is known never to cite something without checking becomes the junior who is trusted with the important questions.

Stage Four: Write the Answer, Not the Search Results. Forty Minutes.

The deliverable is an answer. A list of judgments with summaries is a search result, and handing over a search result means the synthesis, which was the actual work, has not been done. The format below is what seniors want and rarely ask for explicitly.

Confidence Markers: The Habit That Builds a Career

This is the part of the workflow that juniors resist most and that pays the highest dividend. Stating your confidence honestly, in explicit terms, in every research output.

The resistance is understandable. Uncertainty feels like weakness, and a junior wants to look capable. But the senior is going to rely on this answer in front of a court, and the difference between settled and probable and arguable is the difference between an argument they will lead with and one they will hold back. A junior who says the position is settled when it is merely arguable has not been confident, they have been dangerous.

• Settled. Directly decided by the Supreme Court or by a binding decision of the High Court concerned, on comparable facts, and not since disturbed.
• Well supported. Consistent authority pointing one way, though not on identical facts, or from High Courts other than the one you are in.
• Arguable. Support exists but so does a contrary line, or the authority is distinguishable on facts that matter.
• Unsettled. No clear authority, conflicting High Court views, or a question pending before a larger bench.
• Could not determine. The honest answer when time or sources ran out, stated with what remains to be checked.

The Time Budget, and What to Do When It Runs Out

Twenty minutes to scope, forty to search wide, ninety to read and verify, forty to write. Roughly three hours for a standard question, and the budget is the point rather than the total. It forces a decision at each boundary instead of letting a question absorb whatever time is available.

When the budget runs out without a clean answer, the correct move is to report, not to continue. Send what you have with the confidence marker set honestly and the open questions listed. A senior with a partial answer at four in the afternoon can decide what to do. A senior with nothing at nine that evening cannot. Juniors consistently overestimate the value of a complete answer delivered late and underestimate the value of an honest partial answer delivered on time.

Making It a Chamber Habit

A workflow that lives in one junior's head disappears when they leave. Three things make it institutional. A one page template for the research output, used by everyone, so the format is not renegotiated each time. A stated expectation that scope is confirmed in writing before work starts, so juniors are not penalised for asking. And a place where completed research is kept, indexed by proposition, so that the question answered in March is not researched again from scratch in September by someone who never knew it had been done.

CourtMesh is built to support this pipeline. Unified search across the Supreme Court, 25 High Courts, District Courts and Tribunals makes the wide first pass a single search rather than a tour of portals, and it covers the levels a junior would otherwise skip. AI case analysis helps get through a long judgment quickly to locate the paragraph that matters, with the verification step remaining exactly where it belongs, with the person who will put their name to the answer. And because research sits alongside the matters in My Cases, the output of one assignment stays retrievable for the next, which is how a chamber stops paying three times for the same question.