---
{
  "title": "Digital forensics",
  "summary": "Learn how digital forensics preserves evidence, reconstructs events, and separates supported findings from uncertain interpretations.",
  "topic": "forensics",
  "level": "Beginner",
  "updated": "2026-10-12",
  "units": [
    1
  ],
  "aliases": [],
  "sources": [
    {
      "title": "NIST: Integrating Forensic Techniques into Incident Response",
      "url": "https://csrc.nist.gov/pubs/sp/800/86/final"
    }
  ],
  "published": "2026-10-02",
  "editorialVersion": 2
}
---

## Evidence should answer a question

Digital forensics applies disciplined methods to acquire, examine, analyze, and report digital evidence. The purpose is to answer an investigative question with supported findings. It is not simply running many tools and treating every result as a conclusion. Authorization, documentation, preservation, and reproducibility shape the work from the start.

Evidence can come from disks, memory, network records, application events, and cloud services. Each source has limits. A local device may retain one view of an action while the cloud service records another. The investigator needs to understand what the source records, how it was obtained, and what could be missing.

## A document-access investigation

Imagine a fictional organization asks whether a sensitive document was accessed through a compromised account. An investigator obtains authorized copies of relevant service logs and a preserved workstation image. The question is narrower than “Was this computer hacked?” and helps determine which artifacts deserve attention.

A cloud event may show a download associated with an account. A browser artifact may support that the document's page was visited from a particular device. Neither observation by itself necessarily identifies the human at the keyboard. The investigator combines evidence while retaining that distinction.

If the relevant service does not record downloads, the absence of such an event cannot prove no download occurred. Recording this limitation is part of a useful finding, rather than a failure to produce a confident answer.

## Preserve before interpreting

[[evidence-acquisition]] concerns obtaining appropriate material while controlling and documenting changes. Where practical, investigators preserve original material and examine controlled copies. [[chain-of-custody]] records who handled evidence and when, supporting accountability for its movement and storage.

Hashes can help demonstrate that a collected file has remained unchanged between recorded points. They do not establish that the original collection was complete, that a clock was correct, or that a person's interpretation is accurate. Preservation and interpretation solve different problems.

Some sources are volatile. Memory can change rapidly, while cloud records may have limited retention. Acquisition decisions therefore depend on the question, the environment, and the authorized response plan. There is no universal instruction to switch off every device before collecting evidence.

## Build a timeline with explicit assumptions

An [[incident-timeline]] organizes events, but timestamps need context. Different systems may use different time zones, clocks, or meanings for a recorded time. A file's modification time and a service's event-reception time are not automatically comparable. Note the adjustments and uncertainties used in your reconstruction.

Try a small exercise with fictional events: a login, a document request, and an administrative account reset. Put them in order, then state what each directly demonstrates. Add one alternative explanation that would require more evidence to resolve.

## Report the reasoning

[[forensic-reporting]] should explain the question, sources, methods, findings, and limits in language the intended reader can understand. Another qualified person should be able to follow how the evidence supports the conclusion.

NIST SP 800-86 remains a useful introductory reference for integrating forensic techniques into response. Apply it alongside current documentation for the specific systems involved, since cloud services and modern platforms can produce evidence that older examples do not describe.