Best Practices
These best practices help you get faster, more accurate results while building durable system understanding across your organization.
Always Anchor Investigations in Time
Time is the single most important input for accurate investigations.
Best Practice:
Include a clear start time whenever something “broke.”
Use the time when symptoms were first observed—this is the anchor for the investigation.
For example: If an incident occurs on a bridge, use the moment the impact was noticed. When in doubt, the ServiceNow ticket start time is a good proxy.
Why this matters: Traversal looks mostly backwards from this time. Set it too early or days off, and key context may be missed.
Good
“Investigate checkout failures starting 09:12 UTC.”
Risky
“Why is checkout broken?” (no time context)
If results seem vague or low confidence, incorrect or missing time context is the most common cause.
Getting the best results from Traversal’s AI SRE
If you don't know where to start, begin by inputting an inquiry in the Investigations page search bar. You don’t need to choose the “right” mode — Traversal does that for you.
You can start asking Traversal questions immediately after integrating your stack.
Best Practice:
Ask questions naturally: Traversal's AI will automatically determine the reasoning level required to provide an accurate and effective response.
Use follow-ups to refine, drill down, or challenge hypotheses.
While you can and should ask questions naturally, specificity is encouraged.
IDs are very helpful. To the extent they are available, input specific IDs, error logs, or anything else into your query.
Incident investigation
Alert context
Service health
Dependency exploration
Onboarding into unfamiliar systems
Know when to start a new investigation
Use follow-ups when:
You want more detail on an existing RCA
You’re exploring evidence or timelines
You’re validating or disproving a hypothesis
Start a new investigation when:
The incident time changes
You switch to a different alert or service
New information fundamentally reframes the problem
Share Tribal Knowledge with Traversal
Traversal learns autonomously, but it becomes even smarter when guided.
Best practice
Use Knowledge Bank to encode:
Runbooks
Preferred debugging paths
Business-critical services
Known failure modes
Give feedback when an investigation misses practical nuance.
Traversal is strongest when paired with your team’s institutional knowledge.
Use Traversal for learning and onboarding, not just for incidents
Traversal isn’t only for outages.
Best practice
Use it to explore past incidents.
Ask “what usually goes wrong here?”
Onboard new engineers by letting them ask questions freely.
Over time, Traversal becomes a shared source of system understanding, not just an incident tool.
Traversal is designed to do the exhaustive, machine-scale work of investigation so humans can focus on judgment.
The more you:
Ask good questions
Provide clear time bounds
Give feedback
Encode what your team already knows
…the more Traversal becomes a true extension of your best engineers.
Last updated