Every incident question that matters (how they got in, what they touched, when, and whether they are still here) is a question about time, yet evidence arrives as dozens of sources with different clocks, formats and meanings. Timeline analysis is the method that puts those sources on one axis so an analyst can see the sequence of attacker activity and test hypotheses against it. This guide covers what a super timeline is, how to choose sources, how to normalise time, how to build timelines with open source tooling such as Plaso and Timesketch, how to pivot and correlate, how to handle timestamp manipulation, and how to report a timeline that will survive challenge. It goes deeper than our Digital Forensics (DFIR) Guide on this single method, and is written for DFIR analysts, incident responders and threat hunters.
Key takeaways
- Targeted beats exhaustive. A full super timeline of a disk can hold millions of events; start from a pivot point and a question, and widen only when the question needs it.
- Normalise to UTC and record offsets. Plaso's psort outputs UTC by default; know which sources store local time and which clocks were skewed before you correlate.
- A timestamp is a claim, not a fact. Each artefact's timestamp means something specific (created, modified, executed, logged) and can be wrong or manipulated, as ATT&CK T1070.006 Timestomp describes.
- Correlate across independent sources. One event confirmed by file system, event log and network evidence is far stronger than any single artefact.
- Keep two timelines. The raw machine timeline is evidence; the curated incident timeline is the narrative, with every line traceable back to a source event.
What timeline analysis is and when to use it
A timeline is an ordered list of events from one or more sources. A super timeline, the term associated with the log2timeline tool, combines file system metadata, operating system artefacts, logs and application data from a system or set of systems into one sorted dataset. NIST SP 800-86 frames forensics as collection, examination, analysis and reporting across files, operating systems, network traffic and applications; timeline analysis is the main technique in the analysis step that joins those sources together.
| Timeline type | Sources | Best for | Trade-off |
|---|---|---|---|
| File system timeline | $MFT, USN journal, inode metadata | Quick view of file creation and change around a known time | No context on who or what process acted |
| Targeted artefact timeline | Selected artefacts (event logs, Prefetch, registry keys, browser history) via a filter file | Triage of many hosts; answering a specific question | May miss activity in sources you did not select |
| Full super timeline | Everything a parser preset can read on an image | Deep investigation of a key host such as patient zero | Large, slow, noisy; needs strong filtering |
| Enterprise or cross-host timeline | Central logs (identity provider, EDR, firewall, cloud audit) plus host extracts | Lateral movement, account misuse, scoping | Clock and format normalisation across many systems |
Choose the targeted or enterprise view for scoping and the full super timeline for the few hosts where you need to know everything. See the Incident Scoping and Containment Strategy Guide for using timelines to set scope.
Sources and what their timestamps mean
Each source answers a different question. Knowing what a timestamp records is the difference between evidence and assumption. Plaso's parser list covers most of these formats, including mft, usnjrnl, winevtx, prefetch, lnk, registry plugins such as amcache, userassist and bagmru, browser history plugins, text/syslog, utmp and systemd_journal.
| Source | Key timestamps | What it can show | Caveat |
|---|---|---|---|
| NTFS $MFT | $STANDARD_INFORMATION and $FILE_NAME times: modified, accessed, MFT entry changed, born (MACB) | File creation, change and movement | $SI is easy to alter from user mode; last access updates are often disabled or delayed |
| USN change journal ($UsnJrnl:$J) | Time of each change record with reason flags | Create, rename, delete sequences, including deleted files | Rolls over; short history on busy volumes |
| Windows Security log | Event time, for example 4624 logon with logon type and Logon ID | Who logged on, how (type 3 network, type 10 RemoteInteractive) and from where | Workstation and IP fields depend on protocol; logs may be cleared or not retained |
| Prefetch | Last run times and run count | Program execution on workstations | Often disabled on servers; limited history |
| Amcache and registry keys | Key last-write times, recorded values | Presence and likely execution of programs, persistence, user activity | Key last-write time is per key, not per value |
| LNK files and jump lists | Target file times, link creation | Files and folders a user opened, including on removable or network drives | Shows user interaction, not necessarily attacker intent |
| Linux logs and journal | syslog, auth logs, utmp/wtmp, systemd journal entries | Logons, sudo, service changes, cron | Traditional syslog lines may lack the year and time zone |
| EDR telemetry | Process, network and file events | Parent and child process chains with command lines | Retention windows vary by licence |
| Identity provider and cloud audit logs | Sign-in and API call times | Token use, consent grants, role changes, key creation | Ingestion delays; some events have separate activity and logged times |
Host artefact detail is covered in the Windows Forensic Artefacts Guide, the Linux Forensics Guide and the File System Forensics Guide.
Normalising time before you correlate
- Use UTC throughout: store and analyse in UTC, and convert only for reporting where needed. psort writes UTC by default and accepts
--output-time-zonefor formats such as dynamic and l2tcsv. - Record each system's time zone: from the registry or system configuration, before processing. Sources that store local time without an offset are the most common cause of one-hour and multi-hour errors.
- Measure clock skew: compare a known event seen on two systems (a logon recorded on the client and on the domain controller, a request in the proxy and the web server) and record the offset per host. Hosts that were not synchronising time are a finding in their own right. See the NTP Security Guide.
- Watch resolution: sources range from sub-second to whole seconds or coarser. Do not infer order between events in different sources that fall inside the same second unless something else supports it.
- Distinguish event time from record time: a log may show when something happened and, separately, when it was ingested or written. Use the event time for the timeline and keep the other as a field.
Building the timeline
- Define questions and pivot points: for example "first malicious logon on FS01", with a known indicator time from an alert. Questions keep the timeline from becoming a data swamp.
- Acquire and verify evidence: work from verified images or collected triage packages, hash them, and record chain of custody. See the Evidence Handling and Chain of Custody Guide.
- Extract events: Plaso's log2timeline writes a storage file that holds extracted events and metadata. It picks a parser preset during preprocessing; you can override it with
--parsers, limit extraction with a filter file (-f) for targeted triage, and choose partitions and Volume Shadow Copies to process. Record the Plaso version and options used. - Check what was processed: use pinfo on the storage file to confirm sources, parsers and event counts, so you can state later what the timeline does and does not contain.
- Filter and export: psort sorts, filters and outputs events. Use a date range filter around the incident window, keep its default deduplication unless you need every copy, and use
--sliceto see events around a pivot time (five minutes either side by default). Output to CSV or JSON lines for analysis, or to Timesketch. - Add central logs: export identity provider, EDR, firewall and cloud audit events for the same window and normalise them into the same fields (time, host, user, source, description).
- Tag: psort's tagging analysis plugin applies tags from a rules file, which helps mark known-good and known-bad patterns at scale.
- Analyse collaboratively: Timesketch imports Plaso, CSV and JSONL timelines into shared sketches with tagging, starring, stories and analyzers including Sigma.
Tool choices are compared in the Open Source DFIR Tools Landscape Guide, and Plaso is covered in detail in the Plaso and log2timeline Guide. For fast first-pass hunting on Windows event logs before a full timeline, see the Hayabusa and Chainsaw Guide.
Small team versus large investigation
- Single host, small team: a targeted filter file, psort to CSV and a spreadsheet or timeline viewer is enough. Keep the raw output and the curated list separately.
- Many hosts or a long incident: central timeline platform (Timesketch or a SIEM index), consistent field naming, shared tags and a named owner for the curated incident timeline.
Analysis: pivoting and correlating
- Start at the anchor: slice around the pivot event and read outwards. What happened in the minute before (logon, download, process start) and after (new files, services, connections)?
- Pivot on identifiers: follow a Logon ID from 4624 to the processes and privilege events that share it; follow a file name or hash across hosts; follow a source IP across VPN, firewall and host logs.
- Look for clusters and gaps: bursts of file creation in temp folders, new services, or a sudden absence of logs (which may mean log clearing or a stopped agent).
- Test hypotheses: state "the attacker used RDP from VPN address X at 02:14" and look for what must also be true (VPN session, 4624 type 10 on target, RDP-related artefacts) and what would disprove it.
- Use frequency: sort by rarity. A binary or service seen on one host in the estate is more interesting than one on every host.
- Map to techniques: label confirmed activity with ATT&CK techniques (for example T1078 Valid Accounts, T1021.001 Remote Desktop Protocol) so the narrative and detection gaps are clear.
Timestamp manipulation and evidence gaps
Attackers alter timestamps to blend in. ATT&CK T1070.006 Timestomp describes changing modify, access, create and change times, notes that the $STANDARD_INFORMATION attribute is the common target because it can be changed through user-level API calls, and that "double timestomping" also alters $FILE_NAME to defeat mismatch checks. On Linux and ESXi the same goal is reached with touch.
| Indicator | What to check | Strength |
|---|---|---|
| $SI times earlier than $FN times | Compare both attribute sets in the MFT | Strong where present; absent after double timestomping |
| Sub-second fields all zero | NTFS precision is finer than a second; tools that set times may write whole seconds | Supporting, not conclusive |
| USN journal order contradicts file times | File "created" years ago but its creation record is in this week's journal | Strong |
| MFT entry number out of place | A file with an old date but a recently allocated MFT record | Supporting |
| Execution evidence before file existence | Prefetch or EDR shows a process before its file was created | Strong indicator that times were changed |
| Log clearing or gaps | Event log cleared events, missing ranges, stopped agents | Strong indicator of defence impairment |
Treat missing data as a finding. ATT&CK v20 groups log tampering and disabling of security tools under the Defense Impairment tactic (TA0112). Record which sources had rolled over or been cleared, because the report must state what could not be determined. See the Forensic Readiness Guide for retaining the sources timelines depend on.
Where identity fits
In many modern incidents the decisive timeline is an identity timeline: when a token was issued, when a new credential was added to an application, when a service account first authenticated from an unfamiliar network, when an OAuth consent was granted. Host timelines alone will miss activity that happened purely in identity provider and cloud control planes. Build a per-identity timeline for each compromised human and non-human account, joining sign-in logs, audit logs and host logons. See the Identity Incident Response Playbook and the Cloud Forensics Guide.
Reporting the timeline
- Curated incident timeline: one row per meaningful event with time (UTC), host, identity, activity, ATT&CK technique, confidence and source reference (storage file, event ID, record number).
- Confidence levels: confirmed (two or more independent sources), probable (one reliable source), possible (inference). Do not mix them silently.
- Method statement: tools and versions, parsers, filters, time zone handling and skew corrections, so another examiner can reproduce the result.
- Limitations: sources unavailable, rolled over or suspected of manipulation.
- Feed the decision record: timeline milestones such as first access and data access time feed notification decisions. See the Incident Log and Decision Record Template.
Measures that show the capability is working
- Time from evidence receipt to first targeted timeline for a host (target hours, not days).
- Percentage of curated timeline entries with a traceable source reference (target 100 per cent).
- Percentage of in-scope hosts with synchronised clocks at the time of the incident.
- Number of investigations where a key question was unanswerable because a source had rolled over, trended down over time.
Framework mapping
| Framework | Reference | Relevance |
|---|---|---|
| NIST CSF 2.0 | RS.AN-03 | Analysis establishes what took place and the root cause |
| NIST CSF 2.0 | RS.AN-07, RS.AN-06 | Incident data collected with integrity and provenance; investigation actions recorded |
| NIST CSF 2.0 | DE.AE-03 | Information correlated from multiple sources |
| NIST SP 800-53 Rev. 5 | AU-8, AU-9, AU-11 | Time stamps, protection of audit information, audit record retention |
| NIST SP 800-53 Rev. 5 | AU-6, IR-4 | Audit record review and analysis; incident handling |
| ISO/IEC 27001:2022 Annex A | 5.28, 8.15, 8.17 | Collection of evidence, logging, clock synchronisation |
| MITRE ATT&CK v20 | T1070.006, TA0112 | Timestomp; Defense Impairment (log tampering, disabled tools) |
Common failure patterns
- Timeline everything, answer nothing: a full super timeline of every host, with no question, buries the analyst.
- Mixed time zones: a firewall in local time and a host in UTC produce a false sequence, and conclusions about "before" and "after" are wrong.
- Trusting $SI times: a timestomped tool looks years old and is dismissed.
- Reading last access as proof of execution: access times can be updated by scanners, backups or indexing, or not updated at all.
- Host-only view: the attacker's activity in the identity provider or cloud console never appears, so scope is understated.
- No method record: the timeline cannot be reproduced, and its conclusions are challenged.
Practitioner checklist
- Write the investigation questions and pivot points before you process anything.
- Record time zone and measured clock skew for each source.
- Prepare filter files for targeted triage of Windows, Linux and macOS hosts.
- Record tool versions, parsers and filters for every timeline you produce.
- Check pinfo output to confirm what was and was not parsed.
- Join central identity, EDR and cloud logs into the same normalised timeline.
- Check for $SI and $FN mismatches and USN journal contradictions on key files.
- Keep the raw timeline as evidence and the curated timeline as the narrative, linked by source references.
- Assign confidence levels and list limitations in the report.
Standards and references
- NIST: SP 800-86, Guide to Integrating Forensic Techniques into Incident Response
- NIST: SP 800-61 Rev. 3, Incident Response Recommendations and Considerations
- Plaso: Using log2timeline
- Plaso: Using psort
- Plaso: Parsers and plugins
- Timesketch: Collaborative forensic timeline analysis
- MITRE ATT&CK: Indicator Removal: Timestomp
- Microsoft Learn: 4624 An account was successfully logged on
Related NHI Mgmt Group resources: Digital Forensics (DFIR) Guide · Windows Forensic Artefacts Guide · Memory Forensics Guide · Open Source DFIR Tools Landscape Guide