Research question and scope

This protocol measures the share of after-hours escalations that reach the named contact. The unit is one escalation attempt for one incident. Reaching the contact means the named person or the named backup acknowledged the message and could make the first decision. A call that rings without an answer is not a reach.

Set the scope to one support team, one after-hours window, and one escalation tree. Record the triggers, the contacts, and the channels used at the time. If the tree changes during the period, split the sample at the change date. A new contact or number can change the result.

Keep the measure separate from a judgment about the incident. A serious incident can have a fast reach, and a minor one can fail. The purpose is to test the path, not to rank the people on it. State that purpose in the report.

Population and unit of analysis

Define the population as every escalation attempt made during the after-hours window. The unit is the attempt, because one incident may require several calls before a decision owner is reached. Count each attempt and link it to the incident. A per-incident rate can be reported separately.

Include attempts that were made to a backup after the first contact did not answer. Include attempts that used a different channel, such as a chat message or a call. Record the channel and the order. Excluding a failed attempt would overstate the reach.

Exclude incidents that were resolved by the on-duty person without an escalation. Report them separately as self-resolved. They are part of the workload but not part of the reach measure. Keep the two groups distinct in the analysis.

Definitions that make the measure reproducible

Define a reach as an acknowledgment from the named contact or backup within the response target. Define a no-answer as an attempt with no acknowledgment in the target. Define a wrong-contact as an attempt that reached a person without the authority to make the first decision. Publish one example of each.

Define the response target before the sample. A target of ten minutes will produce a different result from a target of one hour. Record the target and the clock used. Use the same target for every contact in the comparison.

Define the first decision as the action the contact is allowed to take under the tree, such as approving a rollback or calling the client. A contact who only listens is not a decision owner. The definition should match the tree, not an assumption about seniority. Record that permission in the tree so the role is not confused with the title.

Sampling and data collection

Draw the sample from the incident log and the escalation channel. Record the export date, the filters, and the number of incidents. Include incidents that were closed quickly and incidents that were reopened. A sample of serious incidents only will not represent the path.

Capture the incident identifier, the trigger, the attempt time, the channel, the contact, the acknowledgment time, the decision time, and the outcome. Add the reason if the contact was unavailable. Store the timestamps with the record. The sequence is the evidence.

Check the contact list against the tree on the observation date. A number that is correct in the tree but wrong in the channel is a different finding. Record the version of the list used. A reach measure without the list version is hard to reproduce.

Ask the on-duty person whether they knew which contact to call. A person who guessed may have reached the right number by luck. Record the answer separately from the outcome. The knowledge gap is a process finding even when the call succeeded.

Measuring escalation reach

Report the share of attempts that reached a decision owner, the share that reached a contact without authority, and the share with no acknowledgment. Show the counts behind each share. The three groups need different fixes and should not be combined into one rate. A rate without the attempt count is easy to misread.

Break the result down by trigger, contact, channel, and time of night. A channel that fails at one hour may work at another. A contact with several misses may need a backup rather than a reminder. Compare only contacts with similar roles and hours.

Report the median time to acknowledge and the median time to the first decision. A fast acknowledgment without authority is not a successful reach. Show both times beside the reach rate. The pair tells the reader whether the path is fast and useful. A rate without the attempt count is easy to misread.

Limitations and next action

The measure cannot show an escalation that was never attempted because the on-duty person decided to wait. It also cannot show a contact who was reachable but unwilling to decide. Treat the result as a measure of the recorded path. Note the incidents that were not escalated.

Give the result to the team lead and the contacts with one change to test. The change might be a verified backup number, a shorter response target, or a different first contact for one trigger. Record the effective date and repeat the measure with the same definitions. A controlled change is easier to judge.

Review the tree with the people who answer the calls. They see the gaps that the log does not capture. A short review after each quarter is enough for a stable service. The goal is a path that works at the hour it is needed.