Methodology: built from Atlassian, ServiceNow, Freshworks and Zendesk documentation and ServiceNow's partner selection workbook, checked 13 Oct 2026. The good-answer and red-flag judgments are ours. No vendor reviewed or paid for this article. We did not run hands-on tests.
A data migration RFP for an ITSM platform should make vendors explain, in writing, how they will move your service history, prove it arrived intact and cut over without losing tickets. The 20 questions below do that, grouped by scope, method, data integrity, security, cutover, pricing and references, each with a good answer and a red flag.
Copy the tables into your RFP. They are built for the Head of Service Management, platform owner or procurement lead who must compare proposals that all promise the same outcome, then defend the choice.
Key takeaways
- Configuration is rebuilt, not migrated. Atlassian disables migrated automation flows in cloud by default (Atlassian migration docs, as of 13 Oct 2026).
- Matching counts can hide missing files. Freshservice's Bulk API guide says a ticket is still created when its attachment over the 40 MB limit is rejected (modified 24 Nov 2025, checked 13 Oct 2026).
- SLA history can break on import. Atlassian says Jira Service Management calculates SLAs from issue history, so SLAs keep counting on closed issues imported by CSV (Atlassian KB, updated 26 Sep 2025).
- Nobody can quote downtime blind. Atlassian says it "can't provide an exact estimation of downtime" and recommends a test migration (as of 13 Oct 2026).
How to run a data migration RFP in five steps
Comparable answers need the same facts.
- Send a data pack. Source and target platforms, record counts by type and year, attachment volume, integrations, compliance constraints and the go-live window.
- Ask for written answers to the 20 questions. Written answers can go into the statement of work (SOW).
- Run a test migration with the shortlist on a copy of your data.
- Review the reconciliation report with your process owners.
- Call references and decide.
Scope and discovery questions
| # | Question to ask | Good answer | Red flag |
|---|---|---|---|
| 1 | What do you need from us in discovery, and what do you hand back? | A written inventory: counts by record type, state and year, attachment volume, custom fields and a draft field map. | A firm price and date before anyone has seen your data. |
| 2 | How much history should we move, and what should stay in an archive? | Asks about audit and reporting needs first, then sets out options: full history, open work only, or an archive plus selected years. | "We move everything," with no question about why. |
| 3 | What do you rebuild on the target as configuration, and who does it? | Workflows, automations, business rules, SLA definitions, macros and running SLA timers are rebuilt, with a named owner. | "Your workflows come across too." |
Question 3 matters most here. ServiceNow says update sets "capture configuration details but exclude task or process data" (ServiceNow docs, as of 13 Oct 2026). Between two platforms, configuration is rebuilt. A vendor who blurs the two in the proposal will blur them in the SOW.
On question 2, Brian Parks, writing on itsm.tools, recommends you "archive the old, migrate what matters to the business" (checked 13 Oct 2026). Audit needs can argue for full history. The good answer starts from your requirements.
Method and tooling questions
| # | Question to ask | Good answer | Red flag |
|---|---|---|---|
| 4 | How will you extract and load, and which platform limits shape the plan? | Names the APIs or import methods on both sides, their limits and the workaround. | One CSV export and one CSV import for everything. |
| 5 | If a load fails halfway, what happens on the rerun? | Loads keyed on unique source IDs, so reruns update instead of duplicating, plus a failure log. | "We wipe the target and start over." |
| 6 | How many test migrations are included, on which data, and who signs off? | A full test on a copy of production data, messy records included, reviewed by your process owners. | A demo on sample data, or tests billed as extras. |
Push back on CSV. Atlassian says CSV import "isn't a recommended migration method due to its limitations" (as of 13 Oct 2026). Zendesk's JSON export leaves out comments for any ticket with more than 1 MB of data, and its CSV export excludes comments entirely (Zendesk help center, checked 13 Oct 2026).
Question 5 tests rerun discipline. In ServiceNow imports, when several records match a key, "only the first matching record in the target table is updated" (ServiceNow docs, as of 13 Oct 2026). So if the key a vendor matches on is not unique, a rerun can update one record and silently skip the others. Ask which key they use.
Data integrity and reconciliation questions
| # | Question to ask | Good answer | Red flag |
|---|---|---|---|
| 7 | How do you rebuild relationships: parent and child, incident to problem to change, ticket to CI, CI to CI? | Explains load order, keeps a source-to-target ID map, samples links in reconciliation. | "Links come across automatically." |
| 8 | What happens to original created dates, authors and internal versus public notes? | Shows where original timestamps land on the target and how note visibility is tested. | All notes arrive public, or created dates become the migration date. |
| 9 | How do you handle SLA history? | Keeps historical SLA results as data, rebuilds definitions and running timers, and explains how the target calculates SLAs. | "SLAs migrate." |
| 10 | What does the reconciliation report compare, and what are the acceptance criteria? | Counts by type, state and year, attachment counts, sampled links and fields, signed by both sides. | Row counts only. |
Ask to see question 9 answered on your test run. Atlassian says CSV-imported closed issues have no history beyond creation, so "the SLA keeps counting" (KB updated 26 Sep 2025). ServiceNow's SLA repair "uses the history from the Task" (as of 13 Oct 2026). The history loaded decides what your SLA reports show.
On question 10, recall the Freshservice case: the ticket arrives, the oversized attachment does not, and these errors "cannot be retried automatically" (Bulk API guide, modified 24 Nov 2025). Matching ticket counts prove the tickets arrived. They prove nothing about what was attached.
Security and compliance questions (in brief)
Your security team will run its own questionnaire. Keep these two here.
| # | Question to ask | Good answer | Red flag |
|---|---|---|---|
| 11 | Where does our data sit during the migration, for how long, and how is it deleted? | Hosting location and retention period in writing, with deletion confirmed at the end. | Data cached "in our cloud" with no retention period. |
| 12 | Which compliance documents cover the migration service itself, and can we read them? | The actual reports or agreements, such as a data processing agreement, scoped to include the migration work. | A website badge with no document behind it. |
Cutover and hypercare questions
| # | Question to ask | Good answer | Red flag |
|---|---|---|---|
| 13 | How do you capture changes after the bulk load, and how do you size the final delta? | Bulk load before the freeze window, delta sync at cutover, window measured in a test run. | A freeze for the whole migration, or a window quoted with no test. |
| 14 | What gets muted during loads, on both platforms, and who does it? | A list: notifications, workflows, business rules and integrations, each with an owner and a re-enable step. | No list. Requesters get emails about tickets closed years ago. |
| 15 | What are the go/no-go criteria and the rollback plan, and who makes the call? | Written criteria tied to reconciliation, source kept available until sign-off, a named decision owner. | "We've never needed a rollback." |
| 16 | What does hypercare cover, for how long, and what is fixed free? | A defined window in the SOW, named defect types, a route for issues found after sign-off. | Open-ended time and materials. |
Question 14 is easy to skip. Freshservice's Bulk API guide says "Slack, Teams, Business Rules, Workflows - Will not be muted" during bulk loads and recommends muting them first (modified 24 Nov 2025, checked 13 Oct 2026).
Pricing and commercial terms questions
| # | Question to ask | Good answer | Red flag |
|---|---|---|---|
| 17 | What is your pricing basis, and what triggers a change order? | Priced per phase after discovery, with drivers named and line items you can compare. | One number before discovery, or open-ended exclusions. |
| 18 | What is excluded: extra test runs, delta runs, rework after failed reconciliation, platform access steps? | A written inclusions and exclusions list, including who requests platform migration access. | "Standard exclusions apply." |
ServiceNow's partner strategy workbook (March 2021, checked 13 Oct 2026) says: "Ask the partner to provide a financial proposal with line-item details." That is how you compare vendors who price differently. Platform access belongs in question 18: Freshservice migration tokens are requested by the migration partner and work only for that partner, that customer account and a defined window (Freshservice article, modified 21 Mar 2025, checked 13 Oct 2026).
References and team questions
| # | Question to ask | Good answer | Red flag |
|---|---|---|---|
| 19 | Can we speak to recent customers who moved a similar source and target at similar scale? | Named contacts you can call, on the same platform pair or a close one. | Logos on a slide. |
| 20 | Who does the work, who is accountable, and is any of it subcontracted? | A named lead and team, a RACI for both sides, subcontractors disclosed. | The sales engineer is the only name you know. |
ServiceNow's workbook asks for references "(preferably three) from clients that had requirements similar to yours within the last three years" and a RACI for each team member (March 2021, checked 13 Oct 2026).
How to score a migration vendor's answers
Score specifics, not confidence.
| Answer pattern | How to treat it |
|---|---|
| Names your platforms, their limits and the workaround | Score high |
| Commits test runs, reconciliation and rollback to the SOW | Score high and copy the wording into the contract |
| Promises ("zero risk", "fully automated") with no method | Red flag; ask for the method in writing |
| No written answer to question 3, 10 or 15 | Knockout |
| Price before discovery | Ask what it assumes; treat as provisional |
FAQ
Methodology
We built the questions from what platform documentation says goes wrong in a move. Facts come from Atlassian, ServiceNow, Freshworks and Zendesk documentation, ServiceNow's partner strategy workbook and one itsm.tools article, checked 13 Oct 2026. The good-answer and red-flag columns are our judgment.
Disclosure: MigrateX, the publisher of this article, runs ITSM data migrations, including between platforms named here. No vendor reviewed or paid for this article.