A Practical Legal Research Workflow — From Question to Verified Authority

Why a workflow matters

Legal research is not Googling. A search that returns a plausible-looking case from the wrong jurisdiction, the wrong decade, or the wrong procedural posture is worse than no result at all — it creates false confidence. The difference between good research and dangerous research is usually not intelligence. It is method.

What follows is a seven-step workflow that works whether you are a solo practitioner, a law student, or someone representing yourself. It starts with a question and ends with a verified citation you can hand to a court or a colleague. At several stages, Freecase's tools can save you time. At every stage, your judgment is the final filter.

Step 1 — Write down the question before you search

Most bad research starts with a vague prompt typed into a search box. "Negligence" is not a research question. "Can a plaintiff recover damages for emotional distress in a medical malpractice case in Illinois where there is no physical impact?" is one.

Before you touch a search engine, write down three things: the jurisdiction (which state or federal circuit), the legal issue (what doctrine or statute controls), and the key facts (what actually happened). If you cannot articulate these, you are not ready to search — you are ready to read a treatise or a practice guide to figure out what question you are actually asking.

This step sounds elementary. It is also the step most often skipped, and it is the single biggest source of wasted research time.

Step 2 — Search for the landscape, not the answer

Your first search should not be for the one perfect case. It should be for the terrain: What courts have addressed this issue? What language do they use? Are there multiple lines of cases that disagree with each other?

On Freecase, you can search with natural language or with the same terms-and-connectors syntax that experienced legal researchers already know. If you are familiar with Westlaw-style Boolean searching, Freecase supports the same operators:

• AND / OR / NOT — combine or exclude terms. • Proximity — "emotional distress w/3 malpractice" finds the terms within three words of each other, in either direction. • Sentence and paragraph connectors — "damages /s emotional" requires both terms in the same sentence; "/p" requires them in the same paragraph. • Phrases — "good samaritan" in quotes matches the exact phrase. • Field restriction — limit searches to specific courts, jurisdictions, or date ranges.

Start broad. Run a few queries with different framings of the same issue. One query will often surface cases that another misses, because courts describe the same doctrine using different language.

For statute research, Freecase covers 20 searchable jurisdictions. If your issue is governed by statute rather than case law, search the statutes first — the controlling statutory language will frame everything that follows.

Step 3 — Use AI to orient, not to conclude

This is where Freecase's AI features can save real time — if you use them correctly.

When you open a case in Freecase, the AI can generate a summary of the opinion: the key facts, the holding, and the legal reasoning. This is useful as a map. It tells you what the case is about and whether it is worth reading in full. It is not a substitute for reading the opinion.

Think of the AI summary the way you would think of a colleague saying, "This case is about X — the court held Y." Useful orientation. Not the final word.

Here is how to use AI summaries well:

• Scan the summary first. If the case is clearly off-point — wrong jurisdiction, wrong issue, wrong procedural posture — move on. This is where AI saves the most time. • If the case looks relevant, read the actual opinion. The summary may miss nuance, omit a limiting parenthetical, or flatten a multi-part holding into a single sentence. • Never cite a case based on the AI summary alone. Read the language yourself. Confirm that the court actually said what the summary claims it said.

AI-generated content can contain errors. It can hallucinate holdings that do not appear in the text. It can miss the one paragraph that limits the holding to a specific factual context. These are not hypothetical risks — they are documented limitations of every large language model in production today. Using AI summaries as a starting point is smart. Using them as an endpoint is malpractice risk.

Step 4 — Read the opinion, not just the result

There is no shortcut for this step. Once you have identified a case that looks relevant, open it and read it.

Confirm the court. An Illinois Appellate Court opinion does not bind a Missouri trial court. A federal district court opinion from 2019 may have been superseded by a circuit opinion in 2023.

Confirm the date. The law changes. A case from 2005 may state a rule that the same court modified in 2022.

Read the actual holding — not the dicta, not the dissent, not the procedural history, unless one of those is what you need. The holding is what the court decided. Everything else is background.

Read the language the court used. Courts choose words carefully. The difference between "we hold" and "we note" is the difference between binding authority and a passing observation.

Freecase provides the full text of opinions and generates reformatted PDFs that you can save, annotate, and share. The PDF export preserves the opinion text with page-marker breaks and paragraph reflow for easier reading.

Step 5 — Check whether the case is still good law

Finding a case that says what you need is only half the job. The other half is confirming that no higher court has overruled it, that no subsequent legislation has abrogated it, and that no later case has distinguished it into irrelevance.

Freecase's citator is an automated tool that can detect some forms of negative treatment — overruling, reversal, criticism, and distinction. It uses multiple detection channels including caption matching, citation graph analysis, and full-text frequency analysis.

Important limitations: the citator is experimental. It is not Shepard's Citations and it is not KeyCite. A "no flag" result does not mean the case is good law — it means the citator did not find negative treatment. An estimated 7–8% of real negative treatment may be missed. A flag is a research lead: it tells you to look more closely, not that a determination has been made.

Until the citator is fully validated and coverage-complete, always verify treatment status using an established citation service. This is not a disclaimer to skim past. Citing overruled authority is one of the fastest ways to lose credibility with a court and to face a sanctions motion.

Step 6 — Build your research trail

Good research leaves a paper trail. For every authority you plan to rely on, record three things:

1. The full citation — case name, reporter, year, court. 2. The controlling quotation — the exact language from the opinion that supports your point. 3. Why it matters — a one-sentence explanation of how this authority fits your argument.

This record serves three purposes. First, it lets you return to the source later without re-running the search. Second, it lets you compare authorities side by side when multiple cases address the same issue. Third, it gives you the raw material for a brief, a memo, or a conversation with a colleague or client.

Freecase lets you export opinions as PDFs for offline annotation. Keep those PDFs organized by issue, not by case — when you sit down to write, you will think in terms of "what does the law say about emotional distress?" not "what did Smith v. Jones say?"

Step 7 — Verify before you rely

Before you file, cite, or advise based on your research, run a final check:

• Is the case from the right jurisdiction? • Is the opinion published (not unpublished or designated "not for publication")? • Is the case still good law? • Does the holding actually say what you think it says? • Did you read the limiting footnotes, the concurrence that narrowed the holding, or the procedural context that limits applicability? • Is there a statute that controls and overrides the common-law rule?

This is the step that separates research from results. A judge or opposing counsel who catches a miscite will not just reject your citation — they will question everything else you have submitted.

Where Freecase fits

Freecase is a research tool, not a lawyer, not a legal advisor, and not a substitute for professional judgment. It provides free access to case law and statutes. It uses AI to help you read faster — to summarize, to orient, to surface what is relevant. It is building tools to help you check authority.

But the responsibility for the research is yours. No tool — not Freecase, not Westlaw, not Lexis — can tell you whether a case is good law with certainty. No tool can read the opinion for you and decide whether the holding fits your facts. No tool can replace the judgment of a licensed attorney who understands the court, the judge, and the client.

Use the tools. Verify the results. Trust your training.