From fact file to research plan · Lesson 4

Build a repeatable research plan

6 min read

Lesson objectives

  • Build a repeatable research plan: produce a checkable result and explain its limit.
Reading tools

An element map is not completed research. If the handoff says only “I searched and found nothing,” a colleague cannot tell where you searched, which version you used, or whether a narrow filter hid the material. The final lesson of this module turns the map into a plan and a log that another researcher can repeat or challenge. The goal is not to record every click. Record choices that could change the result.

Define the deliverable and its limits first: “A two-page preliminary memorandum on the delivery obligation and possible remedy, conditional on identifying governing law, based on documents available at the review date.” For each element, name an initial source and a fallback. Once the legal system and topic are known, start with an official primary collection, using indexes or specialist secondary writing to discover terms and cited authorities missed by the first search. If the system is unknown, do not combine rules from several countries into one analysis. Resolve scope from the accepted agreement and connections first, and label comparative reading as background only. A deliberate order saves effort and prevents the memo from depending on the first easy result.

Make log columns for element; source collection; query or navigation path; jurisdiction, date, and type filters; result or exclusion reason; document identifier; version and check date; next step. An entry might read: “meaning of delivery | official collection after scope is confirmed | accepted term | transaction type and version applicable at the event | unresolved because agreement is incomplete | no identifier yet | unverified | request accepted agreement.” This is not a legal finding. It shows why a jurisdiction-specific rule has not yet been retrieved. A separate entry can record a university index used for terminology, clearly labeled secondary and not binding.

Give each path a stopping rule. One favorable result is insufficient: look for a material contrary source or a plausible alternative reading. Equally, research cannot remain indefinite. A provisional stop is possible when each consequential element has an identified primary source and known version, significant conflicts are recorded, and a conditional conclusion can state its remaining gaps. If that point cannot be reached, do not write “there is no law.” Write “no source was located within the stated collections and queries,” then name the next search or escalation to a locally qualified professional. An empty query result measures the query and the collection, not the absence of a rule.

The research log differs from the final authority list. The final list contains items cited in the memo; the log also explains relevant items rejected or left unresolved. When new documents arrive a week later, it reveals which part must be revisited: scope, a statutory version, or the weight of a decision. A visit date alone is inadequate if the page's substance has changed. Retain the title, stable identifier, exact provision or paragraph, and version where possible, subject to the organization's records and privacy rules. A real client's confidential material should not be moved into a training area or unauthorized tool.

Add a “not done” field to the search log. If only a case summary was opened, say so. If a point-in-time version could not be reached, do not enter today's access date as though it were commencement. A polished table should not make unperformed research look complete. A document request may properly precede database work: the accepted contract could prevent hours of searching the wrong jurisdiction. Search order itself is a reasoned professional choice.

Hand the log to someone who did not search with you. Can that person repeat the query in the same collection, understand an exclusion, and see when scope was fixed? If not, add collection identifier, filter, and reason rather than a claim of “comprehensive research.” Eighty-three hits say little about coverage; they may be duplicates or outside the relevant period. The stopping condition is coverage of the material elements and a reasonable contrary check, not a target count of links.

Lesson summary

Build a repeatable research plan output and verification limit.