When the Duty to Preserve Begins

Preservation is the one discovery obligation that begins before a court is involved. Nothing is served, no scheduling order exists, and no request has been made — yet the decision to leave a mailbox on its ordinary deletion cycle can become the central fact of a case years later. The federal sanctions rule for electronic evidence assumes the duty rather than creating it, which puts the whole weight of the question on when anticipation began.

What reasonable anticipation means

Rule 37(e) speaks of information that "should have been preserved in the anticipation or conduct of litigation." The phrase does two things at once. It confirms that the obligation can arise before a case exists, and it declines to say when. Courts have filled the gap with an objective test: whether a reasonable party in the same position, knowing what this party knew, would have foreseen litigation over the subject matter.

Foreseeability is not certainty. A party need not conclude that suit is probable, only that it is reasonably likely over an identifiable dispute. Nor is it a general sense of legal risk. Every large organization faces the abstract possibility of being sued, and treating that as a trigger would freeze every system permanently. The test asks about a specific controversy with identifiable subject matter, not about background exposure.

The assessment is made on what was known at the time. A later revelation that a dispute was more serious than it appeared does not retroactively move the trigger backward. Courts have been explicit that hindsight is the wrong lens, though the record of what an organization did after an event is often the best available evidence of what it actually foresaw.

There is a second reason the standard is objective. A subjective test would reward the organization that documented nothing and punish the one that wrote down its concerns. Courts have declined to build that incentive, and the result is that an internal memorandum assessing legal exposure tends to be read as confirmation of anticipation rather than as the thing that created it.

Events that start the clock

Certain events recur as triggers because they combine a known dispute with a known adversary. A letter threatening suit, a charge filed with an agency, a subpoena in a related matter, or an internal report of conduct that would obviously support a claim all point the same way. Others are weaker on their own and depend heavily on context.

The pattern across these is that a trigger requires both a subject and an opponent, at least in outline. A vague awareness that some claim might someday arrive does not identify what to keep or whose files to freeze, and a duty that cannot be scoped cannot be discharged.

Whose knowledge counts

An organization knows what its people know, but not equally. Knowledge held by in-house or outside counsel is attributed to the organization almost without argument. So is knowledge held by managers with responsibility for the subject matter, or by executives to whom the event was escalated. The awareness of a single employee at the scene, never reported upward, is a harder case and courts have divided on it.

This matters because the trigger and the identification of custodians are the same inquiry viewed from two angles. The people who first knew of the dispute are usually the people whose records matter most. An organization that can show a documented escalation path — who learned what, and when it reached counsel — is in a far stronger position than one reconstructing the sequence after the fact.

Attribution also runs to agents. Where a third party holds records on the organization's behalf under a contract giving it a right to obtain them, the reach of the duty follows the reach of control described in Rule 34. That includes payroll processors, cloud platforms and outside administrators, and it is a common blind spot in early preservation efforts.

How far the duty reaches

Once the duty attaches, its scope is bounded by relevance and proportionality rather than by exhaustiveness. Rule 26(b)(1) limits discovery to matter relevant to a claim or defense and proportional to the needs of the case, and the same boundary shapes what must be kept. A party is not obliged to freeze every system in the enterprise because one department is in dispute.

What is required is the identification of sources likely to hold relevant material and the suspension of processes that would destroy it. That includes the automatic deletion settings that run without human involvement, backup rotation that overwrites on a cycle, and reimaging of departing employees' devices. The obligation is one of reasonable steps, not perfect capture, and the proportionality limits on what has to be kept are argued at this stage rather than at the sanctions stage.

When the trigger is missed

A late trigger is not itself a sanctionable event. The rule for electronic information asks whether material that should have been preserved was lost, whether it can be restored or replaced, and what prejudice followed. An organization that anticipated litigation months before it acted may still face no consequence if nothing relevant was destroyed in the interval, or if what was destroyed exists elsewhere.

The distinction has a practical edge. Argument about the trigger is often a proxy for argument about a particular category of records — the messages from one team over one quarter, the version of a file that existed before a system migration. Framing the dispute around what was actually lost usually produces a sharper record than an abstract argument about foreseeability, and it is the framing the rule itself invites.

Where loss did occur, the trigger date becomes the pivot on which everything else turns, and it is litigated with the same intensity as the merits. Parties depose the people who received the first complaint, subpoena the escalation emails, and compare deletion logs against the timeline. The available responses then run from curative measures through to the graduated sanctions the rule provides, with the severest reserved for a separate finding about the party's purpose.