Clarifying incident handoffs between teams in IT operations
An IT service provider supports complex customer environments with specialist teams, but unclear handoffs cause delays and duplicated work.
- Technology
- Customer service
KEY METRIC
Time to first qualified response
6039Min.
Average time to first qualified response per ticket.
At a glance
Starting point
A mid-sized IT service provider manages several hundred servers, databases and business applications for its customers. The central service desk takes in about 300 incidents a day by phone, portal and email, and creates tickets in the ITSM system. Separate teams handle the work, such as network, applications and databases, plus an on-call service for critical systems. Handoffs between these groups happen mostly as free text in the ticket, with extra emails on top. Priorities, ownership and next steps are often not clearly documented, which leads to follow-up questions and waiting times.
- 300
- tickets per day
- 5
- IT teams involved
The challenge
Incident handling suffers less from missing expertise than from unclear handoffs. From mixed descriptions and monitoring alerts, the service desk has to decide which team is responsible and how urgent a case is. Tickets are often passed along several times without anyone knowing what information is missing or which checks have already been done. Specialist teams receive incomplete or contradictory details and have to ask follow-up questions while the customer waits for a response. During critical incidents, several teams work on the same issue in parallel because each thinks it is responsible. Managers see only overall numbers in reporting, with no visibility into friction losses at the handoff points.
Typical symptoms in the company
Why this remained unsolved
The problem was known, but it was treated as an organizational issue rather than an IT process in its own right. Earlier improvements focused on checklists and training for the service desk. These helped only to a limited extent, because every incident is different and can’t be fully squeezed into rigid forms. A classic workflow with fixed fields failed because free-text descriptions and monitoring alerts vary too much. There was also no capacity to bring several systems, such as ITSM, monitoring and on-call planning, together in a single project. So handoffs stayed manual: too small a problem for a major IT project, but too big to ignore.
Our solution
An agentic workflow was introduced that structures incident handoffs between the service desk and specialist teams. The workflow reads new tickets, monitoring alerts and relevant master data from the ITSM system. From these it creates a structured handoff proposal covering ownership, urgency and next steps. When information is missing, it guides the service desk with targeted follow-up questions and records the answers directly in the ticket. If needed, the workflow starts queries to the customer or internal clarifications, for example whether a change is in progress. Specialist teams thus receive consistent handoffs with clear tasks. They can report their own steps back through the workflow, which then updates status and communication in the ITSM system.
Where generative AI is at work
Generative AI understands the free text in tickets and emails, combines it with monitoring alerts, and matches it to typical incident scenarios. It writes structured handoffs and specific follow-up questions in clear language, tailored to either the service desk or the specialist team, and condenses even incomplete information in a meaningful way.
- ITSM system connection with read and write access to tickets, status fields and comment functions, including tenant logic.
- Integration of monitoring systems to evaluate alerts, metrics and events that are linked to tickets or added to them automatically.
- Context model of customer environments, services and responsibilities, which derives ownership suggestions and escalation paths for the different teams.
- Generative assistance layer for the service desk and specialist teams that proposes handoffs, follow-up questions and status updates in a structured form.
- Control logic for agentic workflows that triggers follow-up actions in the ITSM system, such as queries, escalations and team changes.
How the incident workflow runs
Capture the ticket
The service desk creates a ticket in the ITSM system as before. The workflow automatically reads in the free text, category and monitoring alerts.
Build a handoff proposal
The workflow analyzes the content, suggests the responsible team, urgency and initial checks, and fills in missing information through targeted follow-up questions.
Specialist team handles it
The responsible specialist team receives a structured handoff and documents its steps through the workflow, which updates the status and comments in the ticket.
Manage escalation
For critical or stalled cases, the workflow suggests escalation paths, such as bringing in more teams or the on-call service, and coordinates the handoffs.
Improve reporting
Finally, handoffs, follow-up questions and handling times are analyzed to reveal bottlenecks at the handoff points and to refine the rules.
Measured 6 months after rollout
Zeitraum: 6 Monate nach Einführung.
| Metric | Value | Basis |
|---|---|---|
| Time to first qualified responseKey metric | −35% | Before/after time trackingmeasured |
| Share of tickets passed along multiple timesQuality measure | −40% | System measurementmeasured |
| Handling time for critical incidents | −20% | Before/after time trackingestimated |
Economics
Limitation
The solution depends heavily on the quality of the master data on services and responsibilities. If that data isn’t maintained, or changes often without documentation, the suggestions become less accurate and manual rework is still needed.
Integration and operation
Transferability
The pattern of an agentic workflow for handoffs between teams transfers well to other areas where many people work on a single case and information is spread across several systems. The key is to bring free-text descriptions and structured signals together, and to have generative assistance suggest clear next steps. This can also ease handoff points outside IT operations, without having to model every variant of a process in detail upfront.