ITSMWORK CONSULT.
← Blog

ITSM & operations · 2026-11-17

Recurring incidents: moving from recovery to prevention

A practical method for connecting incidents, problems and corrective actions without turning your service desk into an administrative burden.

Restore the service, then understand the recurrence

An incident interrupts or degrades a service. Incident management aims to restore an acceptable level of operation, while problem management investigates causes and helps prevent recurrence. These objectives complement each other. A team can therefore close an incident after recovery while continuing an investigation in a linked problem record. This distinction, explained in Atlassian’s documentation, helps prevent every new outage from starting the analysis from scratch.

The following approach is intended for a team with limited time that needs to choose its first improvements. It is a proposed working method, not a customer success story or a universal configuration. Roles, thresholds and stages should be adapted to your organisation.

Start with one family of incidents

Choose a specific service, such as access to a business application. Gather the tickets concerning that service over a defined period, then read their descriptions. Two tickets containing the same keyword do not necessarily describe the same failure. Check the symptoms, timing, affected users and recovery conditions before grouping them together.

Our practical recommendation is to create a short record covering the affected service, observed symptom, linked incidents, business consequences and investigation owner. Record missing information as open questions. This provides a useful starting point even when your technical inventory is still incomplete.

Separate facts, hypotheses and actions

Consider an illustrative scenario: a portal becomes unavailable every Monday morning and recovers after a restart. The restart is a recovery action, not proof of a root cause. Capacity exhaustion, a scheduled task or an external dependency remain hypotheses to investigate. Document what was measured and what is still assumed, including the time and source of each observation.

Give each action an owner, a deadline and a success criterion. An action such as “monitor the portal” is too vague. “Collect load measurements during the next two affected periods and compare them with a normal period” makes the work verifiable. Production changes should continue to follow your usual approval and change-control rules.

Keep a useful record in your tools

In your service management tool, link incident tickets to the problem record and any resulting change. In your knowledge base, describe the validated workaround, its limitations and the person responsible for maintaining it. A colleague should be able to pick up the investigation without reconstructing its history from scattered messages.

For an initial review, we suggest three questions: are incidents recurring less often, can support staff use the workaround, and are agreed actions actually progressing? Keep the scope and comparison period consistent. A lower ticket volume does not, by itself, prove that reliability has improved: a change in usage could also explain the reduction.

Choose a concrete first deliverable

Aim for one complete investigation covering a family of incidents before launching a large transformation: documented evidence, linked tickets, an identified owner and an explicit next decision. ITSMWORK supports ITSM process design, governance and reporting. An engagement can start with this focused assessment before extending the practices to other services. The immediate goal is a shared record that supports action and learning, rather than an additional document that nobody uses.

Sources