Migration guides

ITSM Migration 2026: A Step-by-Step Help Desk Data Migration Plan

Plan an ITSM migration in six phases: what moves, what is rebuilt, platform limits, mapping rules, and the cost, scope and vendor questions to settle first.

Ayush BothraFounding Member, MigrateX

Published Updated 13 min read

Five steps of an ITSM migration: the old system, sorting tickets, a test run, checking the counts, and the new system
On this page

An ITSM migration, often called a help desk migration, moves your service history from one ITSM platform to another: incidents, service requests, problems, changes, knowledge articles, CIs and the links between them. It runs in six phases: discover, map, test migration, cutover with delta sync, reconcile and hypercare.

Workflows, automations, business rules, SLA definitions and running SLA timers do not move; they are rebuilt on the target. If you own a service desk that is switching platforms, this guide covers the plan, the limits that shape it, and the decisions to settle before you sign.

This guide is the briefing. For the working document, use our ITSM migration checklist: 40 checks across the same six phases, each with a done test, an owner and a go/no-go gate, down to the freeze timestamp and the attachment failure log.

Why teams start an ITSM migration in 2026

Three vendor retirements and the renewal cycle push teams toward an ITSM migration in 2026.

TriggerWhat the source says (as of 23 Oct 2026)
Atlassian Data Center end of lifeNo new customers after March 30, 2026; existing customers can expand until March 30, 2028; end of life, Jira Service Management Data Center included, on March 28, 2029, and no renewal runs past it (Atlassian)
Cherwell end of lifeIvanti lists Cherwell under "Cherwell products, now discontinued" but gives no date (Ivanti). StrataCom, an ITSM consultancy, reports December 31, 2026, announced October 31, 2023 (StrataCom); confirm it with Ivanti in writing
Opsgenie shutdownEnd of sale June 4, 2025; Opsgenie shuts down on April 5, 2027, and its capabilities "are being moved to Jira Service Management" (Atlassian)
Renewal and rightsizingA renewal prompts a check on whether the platform still fits the desk's size and process maturity

On Microsoft's stack, System Center 2016 Service Manager reaches end of support on 11 Jan 2027; our System Center Service Manager end of life guide covers the dates and options.

Before you start: scope and inventory

Settle three things before any data moves: the kind of migration, the record counts and how much history comes across.

Know which job you are doing

A cross-platform move is not a hosting move within one vendor. Atlassian's Jira Cloud Migration Assistant carries SLAs, workflows, queues and request types from Jira Data Center to Jira Cloud (Atlassian, as of 23 Oct 2026). Between platforms, configuration models differ, so data moves and the process layer is rebuilt as a second workstream.

Count everything

Inventory every record type: incidents, service requests, problems, changes, knowledge articles, CIs and relationships, users, groups, attachments and notes. Count per state, and date the counts, because reconciliation compares against them. Flag duplicates, orphaned records and unused fields in the same pass.

Decide how much history moves

This is the biggest scope decision, and it belongs to audit, compliance and reporting owners as much as to IT. There are three options:

OptionWhat it meansTrade-off
Full historyEvery incident, change, problem and asset movesThe most work and cost: every record type, link and attachment has to be mapped and reconciled
Active work onlyOpen incidents, unfinished changes, active assets, live knowledgeOld records stay behind, such as a past incident or an audit trail someone later needs
Archive plus selective migrationKeep a full archive of the old system, migrate the records that matterYou report across two sources for a while

Whichever you pick, name who needs old records and how they will reach them.

Before you sign: cost, scope, proof and vendor questions

The commercial decisions are made before mapping starts, and they shape everything below. Five of our guides each answer one:

Across all five, the rule is the same: write history scope, test pass criteria and the cutover method into the contract. If you bring in a provider, our guide to ITSM data migration services covers how to judge one. To see how MigrateX runs these steps, starting with a free 100-record sample of your own records, read how it works.

What moves, what gets rebuilt, what's at risk in an ITSM migration

The one-sentence rule: records move, configuration is rebuilt, and anything computed from history is at risk.

Moves as dataGets rebuilt on the targetAt risk in transit
Incidents, service requests, problems and changes, with their field valuesWorkflows and state modelsSLA history, because targets recalculate it from record history
Comments and work notesBusiness rules and automationsInternal vs public visibility of notes
AttachmentsSLA and OLA definitions, calendars and running SLA timersInline images and very large files
Knowledge articlesNotification templates and assignment rulesOriginal created and resolved timestamps and authors
Users, groups and customer organizationsForms, portal and service catalog itemsLinks between incidents, problems, changes and CIs
CIs and their relationshipsIntegrations, reports and dashboardsMeaning of states and priorities after mapping

SLA history is at risk because targets recalculate it. ServiceNow's SLA repair "removes the SLA record, then recreates and recalculates it" using "the history from the Task" (ServiceNow, as of 23 Oct 2026). Jira Service Management calculates SLAs from issue history, which CSV imports lack. Atlassian's two fixes are to exclude flagged imported issues from SLA goals, or bulk transition them to one Done status so the clock stops there (Atlassian, as of 23 Oct 2026). Neither keeps the original result, so store historic SLA outcomes as imported fields, in an archive, or both.

Help desk migration mapping essentials

In a help desk migration, mapping gives every source value a documented home on the target. Build one signed-off mapping sheet per record type before any test run.

States and statuses

Map every source state, including rare ones. Jira CSV import can only map to statuses that already exist in the target workflow, so build the workflow first (Atlassian, as of 23 Oct 2026). Freshservice's API uses status codes 2 Open, 3 Pending, 4 Resolved and 5 Closed (Freshservice, as of 23 Oct 2026).

Priority matrix

Map impact and urgency, not only the final priority. ServiceNow derives priority from impact and urgency, with sample values from 1 Critical to 4 Low (ServiceNow, as of 23 Oct 2026). Freshservice's API runs the other way, 1 Low to 4 Urgent (Freshservice, as of 23 Oct 2026). A numeric one-to-one copy turns critical incidents into low ones.

Users, groups and assignment

Load users and groups before tickets, and decide how leavers appear: inactive account, placeholder user, or name as text. Freshservice's bulk migration APIs need users and workspaces to exist first (Freshservice, as of 23 Oct 2026).

Notes: public vs internal

Freshservice says private notes are "only visible to agents logged in to your service desk" and cannot be seen by requesters (Freshservice, as of 23 Oct 2026). In Jira Service Management, a CSV import makes all comments public (Atlassian, as of 23 Oct 2026). Test visibility on a sample before any bulk load.

Attachments move after their parent records exist, so give them their own workstream and counts. Links between incidents, problems, changes and CIs, and CI relationships, are rebuilt from a cross-reference of old and new record IDs, so keep that table for the whole project.

ITSM data migration limits to plan around

Each platform caps export size, API calls per minute or work items per import file. These caps set how many runs a full load takes and how long a delta sync needs inside the freeze window.

Platform (vendor docs, as of 23 Oct 2026)Getting data outGetting data inWatch for
ServiceNowUI exports stop at 10,000 records per format by default. The Table API pages with sysparm_limit (default 10,000) and sysparm_offsetImport sets with coalesce fields: a match updates, no match insertsWith no coalesce, every row is new; with several matches, only the first is updated
Jira Service Management CloudCloud backups exclude automation flows, Assets, Opsgenie-powered features and app dataCSV import, about 1,500 work items per file recommended, roughly an hour each by Atlassian's estimateCSV-imported comments become public; SLAs calculate from issue history
FreshserviceTicket exports capped at 10,000 per export, ticket-level fields only; full conversations need the API or an account-wide XML export. API limits run from 100 to 500 calls per minute by plan, account-wideBulk migration APIs for data migration partners: tickets and notes, 50 per request, 10 requests per minute, 40 MB of attachments per entity, all still marked "To be confirmed post load testing"Approved migrations run at 700 calls per minute in a scheduled future window; business rules and workflows are not muted during the load

The practical reading: none of these platforms exports full history with notes, links and attachments in one native file. Plan an API-based extract and test it against these limits before fixing a cutover date.

The ITSM migration plan in six phases

An ITSM migration plan runs in six phases, each ending with an exit criterion. Do not start the next phase until it is met.

1. Discover

Inventory records, configuration and integrations. Agree the history option, the freeze window and sign-off owners, and name one project lead for the whole migration. Exit: a signed scope with baseline record counts.

2. Map

Write the mapping sheets while building target workflows, SLA definitions and automations, because the mapping needs targets to point at. Exit: mapping signed off by the process owners.

3. Test migration

Load a full or representative set into a non-production target. Check counts, open samples in each state, and have agents confirm notes, attachments and links. Exit: a test run that passes reconciliation without manual fixes.

4. Cutover with delta sync

Load the historical bulk ahead of cutover. In the freeze window, stop changes on the source and run a delta sync for records created or updated since. The load must update rather than duplicate on re-run, as ServiceNow's coalesce fields do. Pause target automations that would fire on imported records, and hold a go/no-go with named criteria before switching intake channels. Exit: go decision recorded; email, portal and integrations point at the target.

5. Reconcile

Compare target and source counts per record type and state. Spot-check samples end to end: fields, note visibility, attachments, timestamps, and links to problems, changes and CIs. Log every difference with a cause and a fix. Exit: reconciliation report signed off.

6. Hypercare

Keep the migration team on call. Watch for SLA breaches on migrated records, missing attachments and broken links, and fix each pattern at its cause. Keep the old platform read-only or archived, per the phase 1 decision. Exit: hypercare closed and source retirement approved.

Common service desk migration failure points

Each is cheaper to catch in the phase shown than after go-live. Our guide to migration risks to plan for covers the commercial side.

Failure pointHow it shows upPhase that should catch it
Unmapped states or prioritiesClosed tickets reopen, critical incidents land as lowMap
Internal notes exposedRequesters see agent-only work notesTest migration
SLA clocks running on closed recordsBreach reports fill with historic ticketsTest migration
Automations firing on imported recordsRequesters get emails about old ticketsTest migration
Duplicates after re-runsDelta sync inserts instead of updatingTest migration
Missing linksIncidents lose their problem, change or CIReconcile
No home for old recordsAuditors ask for a ticket nobody can openDiscover

Checklist summary by phase

The short version, one line per phase. The full ITSM migration checklist breaks each of these into 40 checks with a done test, an owner, the evidence to keep and a gate per phase.

  1. Discover: name the migration owner and sign-off owners, record dated baseline counts per record type and state, choose the history option in writing, and record export, API and import limits on both platforms.
  2. Map: build target workflows, SLA definitions and automations first, then map states, priorities (impact and urgency), users and leavers, note visibility, CIs and relationships, and keep an old-to-new ID cross-reference.
  3. Test migration: load a representative set, prove the load re-runs without duplicates, check references, notes and SLAs on imported records, and time the run.
  4. Cutover: load the bulk early, book and announce the freeze window, record the freeze timestamp, run the delta sync with target automations paused, load attachments as their own workstream, and hold a go/no-go with named criteria before switching intake.
  5. Reconcile: match counts against baseline plus delta, sample records end to end, open the links, close the discrepancy log, and sign the reconciliation report.
  6. Hypercare: staff it with a migration queue, review SLAs and routing daily, fix missing-data reports in bulk, keep the source read-only until audit owners release it, then retire it.

Facts checked October 23, 2026 from Atlassian Data Center end of life, Jira Cloud Migration Assistant docs, Jira CSV import docs and Freshservice API docs. No vendor paid for this article. How we research

ITSM migration FAQ

How long does an ITSM migration take?

There is no neutral public benchmark, and none of the vendor documentation checked here gives a duration for a cross-platform move. Time depends on record volumes, the history option, how many states, priorities and custom fields need mapping, the source's export and API limits, how many test runs it takes to pass reconciliation, and when a freeze window is free.

How much does an ITSM migration cost?

There is no neutral public price list. Cost follows scope: which record types and how many years of history move, how many relationships and attachments come with them, how many test migrations it takes to pass reconciliation, and how the cutover runs. Ask every provider to price the same written scope, test runs and delta sync included.

Can you migrate SLA history to a new ITSM platform?

Not as live SLA records on the platforms checked here. ServiceNow's SLA repair rebuilds task SLAs from the task's history, and Jira Service Management calculates SLAs from issue history, which CSV imports lack (ServiceNow; Atlassian, as of 23 Oct 2026). Keep historic SLA results as imported fields, in an archive, or both, and tell report owners which one to use.

Should you migrate closed tickets or archive them?

It depends on who needs old records and how fast. The three options are full history, active work only, or a full archive plus selective migration. Decide per record type and age with your audit and reporting owners.

What is delta sync in a help desk migration?

Delta sync is a second, smaller load run during cutover. It picks up records created or changed on the source after the bulk load, so agents keep working until the freeze window starts. It only works if the load updates existing records instead of inserting duplicates, which ServiceNow's coalesce fields do (ServiceNow, as of 23 Oct 2026).

Written by

Ayush Bothra

Founding Member, MigrateX

Ayush writes about ITSM platforms, pricing and what it takes to move a service desk to a new platform without losing its history.