Table of contents
Research context
This protocol asks how much of an offshore incident is spent waiting for a severity decision. The unit is one incident in the window, identified by the incident identifier and the detection time. The study measures the time from detection to the first severity assignment, from assignment to the first substantive response, and from response to resolution. It does not judge whether the final severity was correct.
Define detection as the first evidence that a customer-visible symptom occurred, not the time the ticket was created. A support agent may notice a pattern before a ticket exists, and that earlier time changes the delay. Record the source of detection and the person who saw it. The decision clock should start from the earliest reliable evidence.
Set the scope to one client, one support queue, and one severity policy. Record the levels, the decision owners, the on-call hours, and the channels at the start. If the policy changes during the window, keep the earlier definition for open incidents and analyze the later cohort separately. A new severity label can look like a faster decision when it is only a different scale.
Measure the decision as a recorded event, not a recollection. A severity set in a call and written down two hours later has a different delay from one recorded at the moment of the call. Record both the verbal time and the recorded time where they differ. The gap is a data quality finding in its own right.
Define the population as every incident created in the window, including incidents that were downgraded, merged, or still open at the observation date. The unit is the incident, not the ticket, because one ticket can produce several incidents. A second severity decision after new evidence is a new event linked to the same incident. Test the definition on two recent incidents before collection begins.
Record the detection time, the source, the initial symptom, the first severity, the decision owner, the decision time, the first substantive response, and the resolution time. Add the shift, the responder, and whether the decision required a call. Keep the incident timeline with the record. A second reviewer should be able to rebuild the sequence.
Practical application
Separate a delay caused by missing evidence from one caused by an unavailable decision owner. A clear symptom with no owner is a staffing problem, while a vague symptom with an owner is an evidence problem. Report the two categories separately so the client can choose the right fix. A new severity guide will not solve an unavailable approver.
Check whether the offshore pod had authority to set a temporary severity. A pod that must wait for the client on every incident will show a longer decision time even when its classification is correct. Record the temporary rule and whether it was used. The review should distinguish authority from speed.
Report the number of incidents, the median time to severity, the median time to first response, and the share that waited longer than the policy target. Show the counts by severity and by shift. Present the share of decisions that required a call and the share with no recorded owner. Use the same denominator for every rate.
A useful additional view is the share of incidents where the severity changed after the first response. A later change can be correct, but a pattern may mean the first decision was rushed or the evidence was incomplete. Record the old level, the new level, and the reason. The review should separate a learning update from a correction of a bad first call.
Inspect the slowest decisions and describe the stage without naming the responder. A cluster at shift handoff may point to a missing summary, while a cluster in one symptom may point to an unclear example. Compare the result with the on-call hours and the incident volume. Report incidents that were detected outside business hours separately.
Operating controls
Do not compare decision times across severity levels without showing the mix. A critical incident may be decided faster because it is obvious, while a minor issue may wait for a weekly review. Report the level and the policy target beside the median. A single average can hide the area that needs attention.
The record cannot show whether the customer was satisfied or whether the severity was appropriate. It can show only the timestamps and the recorded decisions. A fast decision is not proof of good judgment, and a slow one is not proof of neglect. The protocol also cannot capture a severity agreed in a side conversation and never written down.
Check whether the incident had a named communicator separate from the decision owner. A single person who decides and communicates can delay both steps. Record the roles and the first customer update. The review should show whether the communication waited for the severity or proceeded in parallel.
Give the result to the support lead and the client owner and ask them to choose one observable change. The change might be a temporary severity authority, a named backup, a clearer example for one symptom, or a parallel communication template. Record the owner, the effective date, and the measure that will show whether the change worked. Repeat the protocol with the same definitions.
Close the review with one change and a date to re-measure. The change should address the stage that consumed the most time, not the stage that is easiest to document. Record the expected effect and the evidence that will prove it. A severity matrix that is never timed is only a policy document.
Sources
Frequently asked questions
What should happen first?
Begin with a bounded ticket, approved access, explicit acceptance evidence, and a named reviewer.
Who approves production changes?
The client's technical owner keeps final merge and release authority.
Related research
What share of offshore delivery decisions are recorded in writing?
How long do revoked offshore credentials stay usable after an exit?
