Evidence Preservation Is a System, Not a Last-Minute Scramble
Preservation usually begins as a scramble: a report lands, counsel asks who holds the files, and the answer depends on someone remembering what was deleted. Evidence preservation becomes defensible when the trigger, the scope decision, the systems touched, and the release are recorded as operating steps rather than reconstructed after the fact.
Evidence preservation fails quietly, long before anyone asks about it. A credible report arrives and the matter is discussed on a call. The conversation moves quickly to interviews and conclusions. What does not happen is a record: the date the organization knew, who decided what would be held, which systems were suspended, and when the hold would be reviewed. By the time a regulator, an auditor, or opposing counsel asks, the retention schedule has run and the answer depends on recollection.
The Commentary on Legal Holds, Second Edition: The Trigger & The Process from The Sedona Conference sets out the standard most organizations are measured against. It does not demand perfection. It asks whether the organization made reasonable, good-faith decisions at the time, and it treats a documented process as evidence of exactly that.
1. The obligation attaches before the complaint does
Most organizations treat preservation as a litigation activity that begins when a claim or subpoena arrives. The standard attaches earlier, when an organization is on notice of a credible probability of litigation or a government investigation. Inside an investigation, the practical equivalent is a credible allegation with external implications: a whistleblower report naming senior people, a pattern reaching across a business unit, a customer harm that may attract regulatory attention.
The operating question is not whether leadership is certain about what comes next. It is who is authorized to make that judgment, and where the judgment is written down. An organization that has decided in advance which roles can recognize a trigger, and which route that judgment travels, answers the question in an hour. An organization that has not answers it in a week, and the gap is where data disappears.
2. Scope is a decision, not a blanket
“Save everything” feels safe and functions as an avoidance. It is a scope decision nobody made, and it fails twice: it buries the material that matters, and it teaches the organization that a hold is a distribution list rather than an instruction. The Sedona guidance asks for a reasoned weighing of the issues in the matter, the accessibility of the information, its probative value, and the relative burden of preserving it.
At program level, that means naming the question the matter has to answer, listing the sources that would answer it, identifying the custodian groups, and recording why sources were included or excluded. Proportionality is defensible only when someone with authority set it and wrote down the basis.
3. The mechanical failures arrive first
Preservation rarely breaks in the legal argument. It breaks in the plumbing: auto-delete cycles on messaging platforms, retention schedules that purge mailboxes on a fixed clock, departed custodians whose accounts are decommissioned within days, data held by a vendor under a contract that says nothing about preservation, and collaboration tools never treated as a record.
The guidance covers information within the organization’s possession, custody, or control, including sources held by outsourced providers and software vendors, and it notes that modern chat applications retain conversations in a preservable form and should not be assumed inaccessible. Each of those is a system with a setting and an owner, not a reminder email. A preservation step that depends on an employee manually moving messages out of an auto-delete path will fail quietly.

4. Independence determines what actually gets preserved
If a matter implicates the person who normally approves holds, or a function whose records fall inside the scope, the decision has to move. An approval chain that runs through the subject of the review creates both a preservation gap and an appearance problem that outlasts the matter. Independence here is a routing decision: a named alternate approver, a records custodian outside the affected function, and an audit trail showing who authorized the scope and when.
5. Put the record inside the workflow
The record itself is unremarkable: matter reference; the date the trigger was recognized and by whom; the scope decision and its basis; custodians notified and acknowledgements received; systems and settings suspended; the next review date; the authority that will release the hold. The Sedona guidance describes a comparable documentation set, recommends monitoring compliance rather than assuming it, and requires a route for releasing a hold when the obligation ends.
Where that record lives determines whether it survives. A preservation log on a shared drive is updated when someone remembers. A register embedded in the workflow that carries intake, scope, findings, and remediation is updated because the matter cannot move without it. SignalVault™ records preservation against the matter itself, retrievable years later by someone who was never part of the team.
6. Release is part of the system
A hold that never lifts becomes its own liability. Retained data carries cost, complicates future matters, and across borders raises data protection questions the guidance addresses directly. Release should require a cross-check against other open matters, a named authority, and a date.
The leadership test is short. For any matter from the past two years, can the organization show when a preservation obligation was recognized, who set the scope and on what basis, which systems were suspended and confirmed, and who released the hold? Where that record exists, preservation stops being an act of memory under pressure. Organizations that want investigation records built this way can request an organizational license for SignalVault™.
This content is general business and compliance information, not legal advice.