Lawyers and investigators tend to approach the same case from different directions, which is precisely why the relationship can be so productive when the assignment is framed correctly. Counsel is usually thinking about claims, defenses, burdens, admissibility, discovery, impeachment, deadlines, and what must ultimately be established before a decision-maker; the investigator is working backward from those questions into people, places, records, timelines, observations, contradictions, and sources that can be developed in the real world.
The trouble begins when those two layers are never translated.
“Interview the witness,” “do surveillance,” “canvass the area,” and “see what you can find” sound like assignments, but each leaves the investigator to make decisions counsel probably intended to make: which issue matters most, what information is already established, which facts remain genuinely disputed, what contact restrictions apply, what deadline drives the work, how much collateral development is authorized, and what form of work product will actually be useful when the investigator is finished.
A strong investigative assignment begins with the factual question, not the desired answer.
Instead of asking an investigator to “prove the defendant knew about the condition,” counsel can identify the unresolved factual issue: determine who was responsible for the property during the relevant period, identify witnesses with personal knowledge of prior complaints, locate records or other sources documenting notice, and report information supporting or contradicting the notice theory. Instead of “catch him working,” a surveillance assignment can identify the conduct, location, time period, or factual representation counsel needs independently observed, while leaving the investigator free to document what actually occurs rather than hunting for a predetermined conclusion.
That difference is more than stylistic because an investigator retained by counsel is not operating outside the lawyer's professional obligations. Washington RPC 5.3 expressly identifies investigators among the nonlawyer assistants lawyers commonly employ and requires lawyers using both in-house and outside investigative services to make reasonable efforts to ensure that the investigator's conduct is compatible with the lawyer's own professional obligations, including confidentiality and appropriate supervision.
The same principle matters enormously when witness contact is involved. Washington RPC 4.2 prohibits a lawyer from communicating about the subject of a representation with a person the lawyer knows is represented by counsel in that matter unless counsel consents or another legal authorization applies, and the rule expressly states that the lawyer cannot accomplish a prohibited communication through someone else. Organizational contacts can require additional analysis because the rule separately addresses certain current constituents of represented organizations. An investigative brief should therefore identify known representation issues before anyone starts dialing numbers rather than making the investigator discover halfway through the conversation that the contact should never have occurred.
Work-product treatment deserves equally deliberate handling. In Washington civil litigation, CR 26(b)(4) provides qualified protection for documents and tangible materials prepared in anticipation of litigation or for trial by a party or that party's representative, expressly including attorneys, consultants, insurers, and agents, while also allowing discovery upon the required showing of substantial need and undue hardship and protecting attorney or representative mental impressions, conclusions, opinions, and legal theories. The rule also gives a person access, without that heightened showing, to that person's own qualifying prior statement. In other words, hiring an investigator through counsel may matter significantly to trial-preparation analysis, but the words INVESTIGATIVE WORK PRODUCT at the top of a report are not a magic cloak that makes every fact, statement, recording, or document permanently undiscoverable.
That should influence how the assignment itself is constructed. Counsel should know whether the investigator is being asked to identify facts, preserve evidence, obtain a witness account, test a factual theory, document observable conduct, develop leads, or help counsel understand a complicated factual record, because those are different objectives and may justify different documentation practices. The investigator, in turn, should understand what must be memorialized contemporaneously, what source material needs to be preserved, what should be reported as observation versus inference, and when an unexpected fact is important enough to trigger a call to counsel before the assignment continues in a direction nobody authorized.
A good investigative report should not require counsel to reverse-engineer what happened in the field. It should identify the assignment, investigative actions, dates and sources, material findings, unsuccessful efforts that affect interpretation, limitations, conflicts or inconsistencies, supporting records, and additional leads where appropriate, while keeping the distinction between what the investigator personally established and what another source merely reported visible throughout.
That last distinction is where useful investigation separates itself from volume. A twenty-page report that cleanly tells counsel what was attempted, what was established, what remains unresolved, and where the supporting material lives can move a case farther than two hundred pages of unfiltered search results, interview notes, screenshots, and field photographs dropped into a folder because everything felt potentially relevant.
This resource hub is built as a working desk for that relationship: how to brief a witness interview without scripting the answer, what an investigator actually needs before surveillance begins, and how scene canvassing can develop witnesses, cameras, geography, sight lines, and evidence that never made it into the original file.
