Skip to content

As of

How to group duplicate tickets automatically when hundreds report the same outage

To group duplicate tickets automatically during an outage, link reports by the cause they share, not the words they use, then fix once and close every grouped ticket together. STLabs does this with an AI agent that groups tickets using Axiom™, a live model of the people, devices, apps and access behind each report.

Key takeaways

  • In an outage, people describe the symptom they can see, so one failure arrives as hundreds of tickets that share almost no words.
  • Text similarity matching catches reports worded alike and misses the rest, and major incident parent/child linking is the right record but usually depends on someone linking tickets by hand.
  • AIOps event correlation groups monitoring alerts before people notice, and is the better first alarm for infrastructure you monitor, but it does not see who reported what.
  • Grouping by shared cause works because it follows how each reporter's apps, devices and sign-in connect to the thing that failed.
  • Once tickets are grouped, the outage needs one fix, one update to everyone affected, and every grouped ticket closed together when the fix lands.

Reporting a problem used to take a phone call or a form. Now it takes a message in Slack or Teams, and everyone sends one. When a shared service fails at 9:02, the queue can hold hundreds of new tickets by 9:15, most of them about the same thing.

The service desk already knows what to do: declare a major incident, give it one owner, link the duplicates under a parent record. That process is sound.

The trouble is the linking. It happens by hand, when the team has the least time, and depends on someone recognizing that a timesheet complaint and a sign-in complaint are the same outage. Tools that do it automatically take three different approaches, and only one groups tickets the way the failure spread.

Why one outage turns into hundreds of different tickets

People report what they see, not what broke. An expired certificate on your identity provider doesn't reach anyone as "the certificate expired". It reaches them as the app they were in the middle of using.

So the queue fills with "Workday won't load", "I can't sign in to Outlook", "Is SSO down?" and "Can't submit my timesheet". The last one never mentions sign-in at all. Handled alone, each one gets triaged, assigned and answered by someone who may not know the others exist. That is where the first half hour of an outage goes: sorting the queue while the cause sits unfixed.

What tools can group duplicate tickets automatically?

Three categories of tool group duplicate tickets: text similarity matching inside the service desk, major incident parent/child linking, and AIOps event correlation. They solve different parts of the problem, and most teams end up with some of each.

Text similarity matching compares the words in a new ticket with recent ones and suggests or links the close matches. It runs on any queue and catches "Workday is down" next to "Workday not loading". It misses "Can't submit my timesheet", because the words don't overlap. It can also group too much: during a sign-in outage, a ticket that says "Outlook not working" from someone whose mailbox is simply full looks like a match.

Major incident parent/child linking is a record structure. One parent incident holds the cause, the owner and the status, and every duplicate becomes a child that inherits the resolution. This is the right shape for a ticket storm. The catch is that the structure doesn't decide what belongs in it. On most service desks an analyst links each child by hand, or accepts text-match suggestions one at a time.

AIOps event correlation works upstream of tickets. It reads alerts from monitoring and observability tools and groups the ones that share a cause, often before anyone files anything. If your outages mostly start in infrastructure you already monitor, this is the better first alarm, and nothing below replaces it. What it doesn't see is the people: who reported what, in which words, and which of their apps depend on the failing service. A software as a service app you don't monitor, or an access change, can fail without raising an alert at all.

Why grouping by shared cause beats matching text

Text matching asks whether two tickets sound alike. Cause matching asks whether they depend on the same thing, and that is the question an outage actually poses.

To answer it you need to know how each reporter's work connects to the failure: their apps, their devices, how those apps sign them in, and what changed recently. With that, "timesheet", "Workday" and "SSO" point to the same identity provider, even though no two of those tickets share a word.

Take a change to one firewall rule. Within the hour, one team reports a slow internal app, another can't reach a file share, and people working remotely say the VPN keeps dropping. Three symptoms, three likely categories, three different queues. Text matching will never put them together. A model that knows those three paths cross the same rule will.

The same knowledge stops over-grouping. The person with the full mailbox signs in fine and reports a different problem, so their ticket stays out of the outage and gets handled on its own.

This only works if the map is current. A dependency map kept by hand tends to be accurate for whatever someone documented last year, rarely the system that just broke. The case for reading it live from the systems that own the facts is in What is an anti-CMDB.

What happens after the duplicate tickets are grouped

The fix happens once. One owner works the cause, and nobody else spends the morning investigating the same certificate from a different ticket.

The update goes out once. Most follow-ups in an outage are some form of "any news?". One status update to everyone in the group answers all of them: what is affected, any workaround, and when the next update comes.

Every grouped ticket closes together, with the same note, when the fix lands. Nobody is left with an open ticket for a fixed problem, and the count on the parent shows the outage's real reach. Why the same failures keep coming back is a different job, covered in How to find the root causes behind your highest IT ticket volume.

How an AI agent groups a ticket storm in STLabs

STLabs groups related tickets by cause and removes duplicates as they arrive, in Automated Triage. It can do that because of Axiom™, the live model of your organization that STLabs builds from your identity provider, HR system, device management and other connected tools. Axiom™ syncs in real time, and every fact in it records which system it came from and when it was observed.

When the reports start, each ticket is tied to the person who sent it and to the devices, apps and access around them. Tickets whose paths lead to the same failing service land in one cluster with the likely cause beside it. It is the same reasoning that lets STLabs trace a slow app, an unreachable file share and VPN drops back to one firewall rule change. Each ticket also arrives with knowledge base articles, past fixes and Axiom™ context attached, so the owner isn't starting from a blank screen.

From there your team works the whole cluster at once, by typing instructions in plain language. Tell everyone in the group that the cause is found and a fix is coming, and duplicate reports of the same issue are answered together. When the fix is a known one, an approved workflow in Workflows can run it and then close every resolved ticket in the cluster as a batch.

None of this asks you to give up control. You can require a person's approval before a sensitive workflow runs, in Policies & Approvals, and every workflow is versioned so it can be reviewed and rolled back. STLabs logs every AI decision and workflow run in a searchable log, so the review afterward starts from what actually happened. This is the part that needs an AI-native design: the grouping, the fix and the record all read from the same model of the organization.

What an AI agent should leave to your incident owner

An AI agent is good at the sorting and repetition in a major incident. Recognizing the 400th report of the same outage and telling that person what everyone else already knows is not work anyone on your team should do by hand.

The decisions stay with people. Whether to roll back, when to call the vendor and what to tell executives belong to the incident owner. So does deciding that a group really is one outage when the cause is new and the evidence thin. The agent puts the evidence in front of that person; the person decides.

How to test this on your own systems

Use your last real outage. It is a better test than any demo.

Pull every ticket opened in the two hours after it started. Count how many your tools linked to the parent incident, and how long after the first report. Read the ones never linked: those are your text-matching misses. Then count how many reporters were told it was fixed, how many asked for an update first, and how many tickets were still open the next morning.

Finally, ask whether your tools could have known the unlinked tickets shared a cause. If that took someone who remembered how sign-in works for that app, the grouping lives in a person's head.

One outage, one ticket's worth of work

You don't have to replace your service desk to try this. STLabs runs alongside it, and teams go from integration to results in five days. The plan is in Moving to STLabs from your current service desk.

An outage has one cause, and it deserves to be worked as one problem, not answered four hundred times.

Frequently asked questions

What is a ticket storm?

A ticket storm is a sudden flood of tickets caused by one underlying problem, usually an outage of a shared service. Most of the tickets are duplicates describing the same issue in different words.

What tools can group duplicate ITSM tickets automatically?

Text similarity matching, major incident parent/child linking and AIOps event correlation all group tickets or alerts. STLabs groups tickets by shared cause, using Axiom™ to follow each reporter's apps, devices and access to the failing service.

Should duplicate tickets be deleted or merged?

Link them under one major incident rather than deleting them. Every reporter then hears when it is fixed, every ticket closes together, and the count shows the outage's real reach.

Does an AIOps tool replace grouping in the service desk?

No. AIOps event correlation groups monitoring alerts, often before anyone reports a problem. Grouping in the service desk handles the employee reports, the updates to each person and closing every ticket when the fix lands.

See it on your tools.