← All resources

Escalations

Customer escalation vs. incident: choose the right response

A customer writes, “We need someone senior on this now.” Support raises the priority. Engineering asks whether the service is actually down. The account manager asks who is going to call the customer. Everyone is responding to a different question.

A customer escalation and a service incident can happen together. They still need different decisions. Treating them as interchangeable can send the wrong people into the wrong meeting while the original issue sits untouched.

An escalation changes who needs to act. An incident changes how you respond.

A customer escalation asks for a different level of attention, authority or coordination. Perhaps the issue has stalled between teams. Perhaps a deadline makes the normal queue unsuitable. Perhaps the customer needs a decision the current owner cannot make.

An incident concerns a disruption or degradation of a service. The immediate job is to understand the impact and restore service through the organization's incident process. Atlassian's incident-response handbook, for example, sets severity by impact and separates recovery from later follow-up.

These are two dimensions, rather than competing labels:

How the same customer issue can require different responses
Situation What needs to happen
A customer cannot complete a task; the cause is unclear. Establish impact and start an investigation. Escalate under the relevant policy when warranted; uncertainty is not a reason to dismiss the report.
Evidence points to a service disruption. Use the incident process to assess severity and coordinate recovery. Keep customer-specific communication attached to it.
The service works, but a commercial or ownership decision is stalled. Escalate to someone with the authority to decide. An outage response may not solve this problem.

One affected customer can still have a serious incident. A large contract does not, by itself, establish technical severity. Use documented impact criteria, including contractual requirements where applicable, rather than the loudness of the message.

Find out what the customer cannot do

“The integration is broken” is a starting point. It is not enough to choose a response. Before creating a bigger coordination effort, capture a short, concrete account of the impact:

  • The blocked task: what the customer is trying to finish.
  • The observed failure: what happened, when it started and the evidence available.
  • The known scope: which users, accounts or workflows are confirmed affected, and what is still unknown.
  • The immediate constraint: a business deadline, an unsafe workaround or another reason the normal response may be insufficient.

Do not delay urgent response to complete a perfect form. Suspected security, data-loss or other serious events should enter the appropriate process promptly. The point is to keep the uncertainty visible while work begins.

Separate technical recovery from keeping the customer informed

There are usually two pieces of work. Someone needs to investigate or restore the affected workflow. Someone also needs to explain what is known, obtain missing information and keep the customer updated.

One person can do both. Making both responsibilities explicit is what matters. Otherwise the engineer assumes support is updating the customer, while support waits for a technical answer before saying anything.

For the export example, a useful working note could read:

Technical next step
Check one failed export and compare it with a successful attempt. Investigation owner: product engineering; acceptance of the handoff is still pending.
Customer next step
Support confirms the report deadline and sends the next update at 14:00, even if the cause remains unknown.
Shared evidence
The original case, relevant timestamps and permitted diagnostic references. No credentials or unnecessary customer data copied into the note.

An update time is not a resolution estimate. “We will update you at 14:00” is something the communication owner can control. “It will be fixed by 14:00” needs supporting evidence and agreement.

If engineering needs a structured handoff, use the support-to-engineering handoff guide and free template. It covers the receiving team’s question without requiring the customer to repeat the whole story.

A restored service does not automatically finish the customer work

After recovery, check the affected workflow. Can the customer complete the export? Does a failed attempt need to be rerun? Is there an unanswered question about missing output? Technical recovery and the customer's next successful task may be separate steps.

That does not mean keeping an incident open indefinitely. Close the incident according to its criteria, and name the remaining customer follow-up separately. Give it an owner, a next step and a way to verify completion.

The useful question at the end is simple: what, if anything, is still preventing this customer from moving forward? An answer with evidence is more useful than another status label.