Phase 0

Decide & Plan

Decisions that lock scope before any work begins.

Confirm the Freshdesk plan tier and Omnichannel variant. Free, Growth, Pro, and Enterprise gate features differently. Sandbox requires Enterprise. Custom Objects require Enterprise (20 objects, 100 fields, 5 associations cap).
Risk if skipped: A plan-tier surprise mid-migration forces a contract amendment or strips features like sandbox testing and custom objects from scope.
Decide whether Workspaces or multi-product is required. Newer Freshdesk accounts use Workspaces to isolate brands; legacy Estate/Forest tiers used multi-product. Pick one model before configuration begins; conversion mid-flight is painful.
Risk if skipped: Configuring single-workspace then discovering a second brand needs isolation requires rebuilding ticket fields, SLAs, and automations.
Define migration scope by object. Document explicitly which objects move: Tickets, Contacts, Companies, Agents, Groups, Solutions articles, Custom Objects, Time entries, Satisfaction ratings, Scenario Automations, Dispatch'r/Observer/Supervisor rules.
Risk if skipped: Scope ambiguity around Custom Objects or time entries surfaces during cutover and forces emergency mapping work.
Decide the historical-data window for tickets. Two to five years of ticket history is common. Document an explicit cutoff (e.g. tickets created on or after 2022-01-01). Closed tickets older than the window stay in archived source exports.
Risk if skipped: Pulling all-time history balloons run time; pulling too little erodes agent context for long-tail accounts.
Confirm the Freshdesk data center region. Freshworks hosts in US, EU, AU, and IN regions. Region is set at account creation and cannot be changed. Match to data-residency obligations (GDPR, regional contracts) before provisioning.
Risk if skipped: Provisioning in the wrong region requires opening a new account and reloading every byte.
Capture GDPR, HIPAA, and SOC2 obligations. If PHI is in scope, execute a Business Associate Agreement with Freshworks before any health data lands. Document retention-policy changes that take effect at cutover.
Risk if skipped: Loading PHI without a signed BAA creates a notifiable HIPAA breach.
Assign a single accountable migration owner (DRI). One named person owns scope, schedule, and sign-off. Multiple stakeholders may contribute, but only one person decides.
Document explicit out-of-scope items. List what is NOT migrating: spam tickets, archived inactive end users, deleted KB articles, voicemails older than X months. Get sponsor sign-off on the exclusion list.
Risk if skipped: Unwritten exclusions become user complaints in the first week after cutover.
Set the cutover window and source-system freeze date. Pick a low-ticket-volume window. Freeze must include source-system writes, automations, and integration triggers. Communicate the date to agents at least two weeks ahead.
Define written success criteria. Concrete pass/fail thresholds: ticket count delta < 0.1%, KB article count exact match, attachment availability > 99.5%, zero failed agent imports, custom field coverage 100% on a 50-record sample.
Risk if skipped: Without numeric criteria the migration is never formally complete and the team keeps firefighting.

Phase 1

Pre-Migration Prep

Everything that must be true before the first byte leaves the source.

1a. Source-system audit and cleansing

Export current source record counts as a baseline. Capture counts for tickets (open, pending, resolved, closed), contacts, companies, agents, groups, KB categories/folders/articles, and automation rules.
Risk if skipped: Without a baseline you cannot prove the migration was complete during Phase 6 reconciliation.
Run deduplication on source contacts using email as the match key. Normalize email case to lowercase before deduping. Freshdesk will overwrite an existing contact when a CSV import row matches an existing email, so source duplicates create silent overwrites.
Risk if skipped: Duplicate contacts in the source produce silent merges in Freshdesk, mixing ticket histories across distinct people.
Standardize phone numbers to E.164 format. Freshdesk's phone channel requires a country code prefix. Convert 4088481234 or 408-848-1234 to +14088481234 before import.
Risk if skipped: Phone numbers without country codes are not recognized by Freshcaller and routing breaks for inbound calls.
Identify contacts with more than one email address. Freshdesk supports one primary plus up to nine secondary emails per contact (10 total). Source rows must use the correct multi-email format in the import CSV.
Risk if skipped: Multi-email contacts collapse to a single email and the other addresses are lost, breaking inbound email matching.
Resolve missing required-field records in the source. Find tickets, contacts, and companies missing values for any field that will be required in Freshdesk. Decide: fix in source, default at transform, or drop.
Risk if skipped: Required-field gaps surface mid-import and halt entire batches.
Archive or exclude spam, test, and bot-generated tickets. Filter out resolved tickets older than the chosen window, test tickets, spam, and bot pings before export. Less data means faster migration and a cleaner Freshdesk on day one.
Inventory inline images and embedded attachments per ticket. Note which tickets carry inline images in the body (not just attachments). These often require special handling so the inline rendering survives in Freshdesk.
Risk if skipped: Inline images converted to broken links produce unreadable historical tickets agents cannot escalate from.

1b. Freshdesk destination setup

Provision the Freshdesk production account. Sign up at the correct data center region, confirm timezone, business hours, and primary language at first login. These settings affect ticket timestamps and SLA clocks.
Provision the Freshdesk sandbox (Enterprise plan). Go to Admin > Support Operations > Sandbox and click Build Sandbox. A replica of production configuration is created. Wait for the admin email before testing.
Risk if skipped: Skipping sandbox testing puts every field-mapping bug into production where rollback is expensive.
Create all agents and confirm seat counts. Include inactive or legacy agents needed for ticket ownership history. Set agent type (occasional / full-time), roles, and group memberships.
Risk if skipped: Tickets owned by agents who do not exist in Freshdesk fall back to a default owner and break historical workload reporting.
Create all groups and assign agents. Group names and IDs become the routing target for migrated tickets. Match source group names exactly to keep mapping simple.
Configure global account settings before any import. Set timezone, business hours, default ticket language, currency, and date format (YYYY-MM-DD). These influence how migrated timestamps display.
Risk if skipped: Changing timezone after import shifts every ticket timestamp and breaks SLA reporting.
Recreate all ticket custom fields with matching types. Build dropdowns with case-sensitive values that exactly match the source values. Build dependent fields. Lock the schema before importing data.
Risk if skipped: Importing a value into a missing custom field silently drops it; importing a value with mismatched case fails validation.
Recreate all contact and company custom fields. Build the same field types (text, dropdown, multi-select, number, checkbox, date) on Contact and Company entities to mirror the source schema.
Create Custom Objects and associations (Enterprise only). Up to 20 Custom Objects per account, 100 fields each, 5 associations per object. Native objects supported for associations: Tickets, Contacts, Companies.
Risk if skipped: Creating Custom Object schema after data import means rebuilding lookup relationships row by row.
Lock immutable field types before import. Some Freshdesk field types cannot be changed after creation (e.g. dropdown to multi-select). Decide types up front; document the schema as the contract.
Risk if skipped: A wrong field type discovered after import forces field deletion and re-import of every affected ticket.
Disable all automation engines before import. Turn off Dispatch'r (ticket creation rules), Observer (ticket update rules), Supervisor (time-based rules), Scenario Automations, and all email notifications. Re-enable in Phase 7.
Risk if skipped: Active rules during import spam customers with status-change emails, trigger reassignments, and inflate Freshcaller call attempts.
Disable outbound email notification triggers. Confirm under Admin > Notifications that 'New ticket created' and 'Agent reply' notifications are off during the load window.
Risk if skipped: Migration imports trigger thousands of stale notification emails to end users, generating immediate complaint tickets.
Configure ticket fields, statuses, types, and priorities. Match source ticket workflow values exactly. If source had 'Waiting on Customer', create that status in Freshdesk; do not assume defaults.
Connect support email channels and verify forwarding. Add support@yourdomain mailboxes, configure DNS forwarding, and verify each address routes to Freshdesk. Email channels are UI-only and cannot be imported.
Risk if skipped: Email reconnect on a previously-connected mailbox can pull the last 48–72 hours of inbox history and create duplicate tickets in Freshdesk.

1c. People and process prep

Communicate the source-system freeze to all agents. Announce the freeze date, time, and what 'frozen' means: no new tickets, no field edits, no automation changes. Pin the announcement in the team chat.
Schedule end-user training on Freshdesk basics. Cover the agent inbox, ticket properties, canned responses, scenario automations, and how to find migrated history. Record sessions for absent agents.
Designate first-responder agents for the first 48 hours post-cutover. Two named agents who will triage user-reported data issues and escalate them. Their tickets get expedited routing during the post-cutover window.
Risk if skipped: Without a triage owner the first surface bug becomes a Slack thread with no resolution path.
Provision an admin-rights user dedicated to the migration. All bulk imports run under this user. Keeps the audit log clean and the migration identity revocable without affecting daily users.
Document the rollback plan and decision threshold. Define what conditions trigger rollback (e.g. >5% record loss, agent-blocking field error). Name the person who decides. Document the source-system unfreeze procedure.
Risk if skipped: Without a documented rollback plan, a bad cutover devolves into ad-hoc decisions made under pressure.

Phase 2

Source Export

Get the data out of the source system in a form you can transform.

Export tickets with all standard and custom fields. Include ticket ID, requester, agent owner, group, status, priority, type, source channel, created_at, updated_at, resolved_at, closed_at, all custom field values, and tags.
Risk if skipped: Missing the resolved_at or closed_at column means migrated tickets cannot be re-stamped to the right historical state.
Export ticket conversations as a separate file. Each conversation entry needs ticket ID, author, body (HTML), public/private flag, created_at, and attachment references. Public notes become customer-visible replies; private notes become internal notes.
Risk if skipped: Conversations exported without the public/private flag will all import as the same visibility, leaking internal discussion to customers.
Export attachments as files plus a metadata index. Index columns: source filename, MIME type, file size, ticket ID, conversation ID, inline-or-attached flag. Confirm total volume against the 20 MB per-conversation Freshdesk limit.
Risk if skipped: Attachments exceeding 20 MB per conversation are silently rejected at import and the corresponding ticket loses evidence.
Export inline images with rewritten body references. Walk every ticket and conversation body, identify <img> tags, download each binary, and add the file to the attachment index keyed to its parent message.
Risk if skipped: Inline images become broken-image placeholders if not re-uploaded and rewritten with new URLs.
Export contacts and companies as separate CSVs. Contacts: include primary email, secondary emails, phone in E.164, company ID, custom fields. Companies: include company name, domain, custom fields. Record the company-contact association.
Export agents with roles, groups, and reporting structure. Agent names, emails, roles, group memberships, and agent type (occasional / full-time) are needed to recreate the team in Freshdesk before any tickets land.
Export the knowledge base hierarchy. Solutions in Freshdesk follow Category > Folder > Article. Export from source preserving that depth: categories with descriptions, folders with visibility settings, articles with HTML body, status, author, and tags.
Export article translations and language mappings. If the source KB has multilingual articles, capture the primary article ID and each translation's locale, title, and body. Freshdesk supports per-folder language mapping via uploaded .yml files.
Risk if skipped: Translated articles imported without a locale tag default to the primary language and the multilingual experience breaks.
Export time entries linked to tickets. If time tracking is in scope, export with agent ID, ticket ID, billable flag, duration, executed_at timestamp, and notes.
Export Satisfaction (CSAT) ratings. CSAT can be imported, but the source export needs ticket ID, rating value, feedback text, and rated_at timestamp.
Capture screenshots and configs of source automations. Source triggers, automations, SLAs, dispatch rules, and macros will be rebuilt manually in Freshdesk. Capture exact conditions and actions for each rule.
Risk if skipped: Without captured automation configs, the team forgets edge-case rules and customer-facing behavior subtly changes after cutover.
Export the canned response / macro library. Macros and canned responses are rebuilt in Freshdesk as Scenario Automations or Canned Responses. Capture title, body, and folder assignment.
Export reports and dashboard definitions. Native Freshdesk reports differ from source-platform reports. Capture filter logic, grouping, and chart type so analytics can be rebuilt in Phase 7.
Record the exact timestamp of each export job. Note start and end time of every export. The delta between this timestamp and the cutover freeze is what the final delta export must cover.

Phase 3

Transform & Map

Take raw source exports and produce Freshdesk-shaped files.

3a. Mapping spreadsheet

Build a master field-mapping spreadsheet. One row per source field with destination Freshdesk field, transformation rule, nullability, and required-on-import flag. This becomes the contract for the rest of the migration.
Risk if skipped: Without a written mapping, field decisions get re-debated mid-import and waste cycles.
Resolve all dropdown value mismatches. Freshdesk dropdown values are case-sensitive. Source value 'Closed Won' will not match Freshdesk choice 'closed won'. Normalize every dropdown column before import.
Risk if skipped: Case mismatches reject entire rows silently in batch with no partial save.
Decide the upsert key per object. Contacts: email (lowercased). Companies: name or external_id. Tickets: external_id (stored in a custom field) to allow idempotent re-runs.
Decide handling for source fields with no destination equivalent. Per field: drop, concatenate into the ticket description, or create a new custom field. Document the decision in the mapping spreadsheet.
Map source statuses to Freshdesk's status set. Freshdesk default statuses are Open, Pending, Resolved, Closed. Add custom statuses to match source before mapping; do not collapse 'Waiting on Customer' into 'Pending' unless agents agree.
Risk if skipped: Status collapsing skews SLA and reopen-rate reports for years after cutover.
Map source priorities to Freshdesk's Low / Medium / High / Urgent. Document the exact source-value-to-Freshdesk-value table. Tickets without a priority should default consistently.
Map source agent IDs to Freshdesk agent IDs. Build a lookup table from source agent email or ID to the Freshdesk agent_id returned after Phase 1 provisioning. Use this table everywhere agent attribution appears.
Risk if skipped: Missed agent mappings dump tickets onto a default agent and erase historical workload attribution.

3b. Data transformation

Convert all timestamps to Freshdesk's expected format. Dates use YYYY-MM-DD; full timestamps use ISO-8601 with timezone (e.g. 2026-05-27T14:30:00Z). Confirm the destination account timezone before conversion.
Risk if skipped: Mixed timezone strings produce ticket events where first_responded_at appears before created_at, breaking SLA calculations.
Lowercase and trim every email address column. Freshdesk treats email as a contact match key. Mixed case or trailing whitespace causes duplicate contacts on import.
Re-encode all text fields as UTF-8. Non-UTF-8 bytes in ticket bodies and notes break the import parser. Re-encode CSVs as UTF-8 with byte-order-mark if your loader expects one.
Risk if skipped: Bad encoding causes silent character corruption (mojibake) across tens of thousands of records.
Escape commas, quotes, and newlines inside fields. Wrap fields containing commas in double quotes; escape embedded quotes by doubling them. Newlines inside ticket descriptions must be preserved with proper quoting.
Truncate fields that exceed Freshdesk character limits. Walk every text field's max length in Freshdesk's schema. Truncate at a UTF-8-aware boundary to avoid mid-character cuts.
Risk if skipped: Source rows exceeding Freshdesk text limits reject the entire row, not just the long field.
Reformat phone numbers to E.164. Strip all formatting characters, infer country code from contact metadata, and prefix the country code. Reject any number that still cannot be parsed.
Convert HTML in ticket bodies to Freshdesk-safe markup. Strip <script>, normalize <a href> attributes, and ensure inline image src attributes point at re-uploaded Freshdesk URLs (assigned in Phase 5).
Validate every dropdown value against its destination picklist. For every row, confirm the dropdown column's value exists in the Freshdesk field's choice list with matching case. Reject rows that fail before submission.
Risk if skipped: A single bad dropdown value can reject the entire ticket and block the import batch.

3c. Relationship and audit-trail decisions

Decide strategy for original ticket timestamps. Freshdesk stamps newly created and updated times on most objects automatically. Capture original created_at, updated_at, resolved_at into a 'Source Timestamp' set of custom fields on Tickets to preserve audit history.
Risk if skipped: Without preserved source timestamps, all migrated tickets look like they arrived on cutover day and historical reporting collapses.
Lock in the contact and company created_at strategy. Freshdesk does not allow backdating contacts/companies created_at; the field is system-managed. Store original creation date in a custom field 'Source Created At' for each contact and company.
Risk if skipped: Teams that assume created_at is overridable discover at validation that every contact looks brand-new and account-tenure reports are wrong.
Decide ticket-ownership strategy for inactive agents. Two options: assign original owner if the agent exists in Freshdesk, or assign to a single 'Data Migration' agent for departed staff. Document the choice and apply consistently.
Decide attachment migration strategy. Re-upload binaries as Freshdesk attachments (counts against per-conversation 20 MB limit) or preserve as external storage links in a custom field. Pick based on total volume.
Decide how to preserve the source ticket ID. Add a custom ticket field 'Source Ticket ID' and populate during import. This becomes the upsert key for re-runs and the audit anchor for the next decade.
Risk if skipped: Without a source-ID custom field, idempotent re-runs become impossible and any retry duplicates tickets.

3d. Helpdesk-specific transforms

Reassemble ticket conversations into Freshdesk's reply and note structure. Split each thread into two sequences: public messages become customer-visible replies; internal notes become private notes. Preserve the original chronological order across both sequences.
Risk if skipped: Mixing public and private into a single visibility exposes internal triage notes to customers via the portal.
Build the Solutions Category > Folder > Article import payload. Freshdesk Solutions enforces a three-level hierarchy. Create categories first, then folders nested under each category (with visibility = all_users, logged_in_users, or selected_companies), then articles nested under folders. Each article carries title, description, status, and seo_data.
Prepare per-folder translation .yml files for multilingual KB. Freshdesk supports article translations via uploaded YAML files. Build one .yml per language containing labels and picklist values; upload via the Manage Translations slider after the primary article exists.
Risk if skipped: Uploading translations before the primary article exists drops the translation silently.
Plan the three-engine automation rebuild. Freshdesk has three separate automation engines that must each be rebuilt manually: Dispatch'r (ticket-creation rules), Observer (ticket-update rules), and Supervisor (hourly time-based rules). Source platforms typically conflate these into one trigger system.
Risk if skipped: Treating all source triggers as a single Freshdesk rule type creates rules that never fire or fire on the wrong event.
Convert source macros to Scenario Automations. Freshdesk Scenario Automations are the macro equivalent: agent-invoked bundles of field updates and replies. Recreate each source macro as a Scenario Automation under Admin > Workflows > Scenario Automations.
Map SLA policies to Freshdesk's SLA model. Recreate SLA policies under Admin > Workflows > SLA Policies. Match the source policy's conditions (priority, group, source) to Freshdesk's matching criteria. Re-create response and resolution targets per priority.
Risk if skipped: Missing SLA policies break first-response and resolution clocks for migrated open tickets immediately at go-live.
Configure business hours and holiday calendars. Business hours are tied to SLA clock calculation. Replicate source business hours and holidays under Admin > Workflows > Business Hours before SLA testing.
Risk if skipped: Default 24x7 business hours bypass source business hours and inflate SLA pass rates artificially.
Set up multi-channel inbox connections. Freshdesk supports email, Freshcaller, Freshchat, social channels, and WhatsApp. Each channel is configured separately under Admin > Channels. Plan the order of connection to avoid premature inbound traffic.
Risk if skipped: Connecting Freshchat before agents are trained causes live customer chats to bounce around an empty queue.
Recreate the CSAT survey configuration. Freshdesk's Satisfaction Survey settings live under Admin > Workflows > Customer Satisfaction. Configure question, rating scale, and send-on-trigger conditions to match the source survey behavior.
Stage Freddy AI features for post-import enablement. Freddy Copilot, Triage, and predictive prioritization need a representative ticket history before they perform well. Plan to enable Freddy AI in Phase 7 after data lands and indexes.
Risk if skipped: Enabling Freddy AI mid-import causes the model to train on a partial dataset and gives bad recommendations for weeks.

Phase 4

Sandbox Test Migration

Catch every problem in sandbox before it costs you in production.

Load a 1–5% representative sample of tickets into sandbox. Pull tickets across the full date range, all priorities, all sources, all custom field variants. A skewed sample misses bugs that only appear at the edges.
Risk if skipped: A sample biased toward recent tickets misses encoding bugs in legacy threads that surface in production.
Validate field-by-field on 10 random sandbox tickets. Open each ticket in the agent UI and compare every standard and custom field to the source. Confirm dropdowns show the right case, dates show the right timezone, and tags survived.
Test long-string and special-character edge cases. Include tickets with emoji in the subject, embedded quotes in descriptions, 8000+ character bodies, and HTML tables in conversations. These break the most parsers.
Test attachments at the 20 MB per-conversation ceiling. Load a ticket with multiple attachments totaling near 20 MB to confirm Freshdesk's upload pipeline accepts the batch and renders thumbnails.
Risk if skipped: Discovering the 20 MB ceiling on production cutover day forces emergency re-architecture of the attachment loader.
Test private vs public note visibility. Open a sandbox ticket as a customer in the portal. Confirm internal notes are not visible and public replies are. If visibility leaks, fix the public/private routing before production.
Risk if skipped: Internal-note leakage exposes triage commentary, escalation discussion, or competitive analysis to customers.
Test relationship preservation for contact-company links. Verify that imported contacts are associated with the right company in the sandbox and that 'view all this company's tickets' permissions work as expected.
Test Custom Object lookups on tickets (Enterprise). Open a ticket that should have a Custom Object lookup populated. Confirm the lookup shows the correct related record and the data displays in the ticket pane.
Time the sandbox loads against the production volume. Freshdesk imports via managed-migration apps on the Freshworks Marketplace typically process around 2,000 tickets per hour. Extrapolate to the full source volume and confirm the cutover window is realistic.
Risk if skipped: Discovering a 100k-ticket load needs 50 hours during cutover weekend leaves no buffer for diagnosis.
Test Solutions article import including translations. Load one full Category > Folder > Article chain in the sandbox, then upload a translation .yml and verify the localized labels appear for agents in that language.
Get a domain expert to spot-check sandbox data. A long-tenured support manager (someone who knows what 'right' looks like) reviews 25–50 sandbox tickets and signs off on field-by-field accuracy.
Risk if skipped: Without a domain expert's eye, subtle data wrongness ships to production.
Sandbox throughput sign-off. Record final sandbox numbers: rows/hour, total runtime, retry count. Confirm production can absorb the load at the chosen plan tier within the cutover window.
Sign off the sandbox test results in writing. The DRI plus a sponsor email attestation that sandbox results meet the success criteria from Phase 0. Production cutover starts only after written sign-off.
Risk if skipped: Without written sign-off, scope creep continues into production and the window slips.

Phase 5

Production Cutover

The migration itself — tightly sequenced execution.

Confirm the source-system write freeze. Verify no new tickets, no field edits, no automation changes in the last hour of the source system. Pause inbound email forwarding from any in-scope mailbox.
Risk if skipped: A leaked write during freeze means the delta never matches and reconciliation runs forever.
Take the final delta export from source. Re-export everything created or updated since the Phase 2 export timestamps. This delta represents the gap between baseline export and freeze.
Confirm Dispatch'r, Observer, and Supervisor rules are disabled. Re-verify under Admin > Workflows that all three engines are paused. Confirm Scenario Automations are not auto-triggered (they shouldn't, but double-check).
Confirm outbound notifications are off. Under Admin > Notifications, verify 'New Ticket', 'Agent Reply', and 'Status Change' emails are suppressed. Customer email is the loudest failure mode of a bad cutover.
Risk if skipped: One enabled notification rule emails thousands of customers about 3-year-old resolved tickets.
Load agents first. Agents are referenced by every other object via agent_id. Confirm the full agent roster exists in production before any ticket lands.
Load companies, then contacts, in that order. Contacts reference company_id, so companies must exist first. Use idempotent upserts keyed by external_id where available.
Load Custom Object records before tickets that reference them. Tickets with Custom Object lookup fields need the target record to exist. Load Custom Object data immediately after companies/contacts and before tickets.
Load tickets next, in chronological order. Create each ticket keyed on the Source Ticket ID custom field. Loading chronologically keeps conversation thread reassembly straightforward.
Load ticket conversations after tickets. For each ticket, load public messages and internal notes in original chronological order. Preserve the author by mapping source agent_id to Freshdesk agent_id.
Risk if skipped: Mixed-order conversation loads scramble the thread display and confuse agents reading historical context.
Load attachments mapped to their parent conversation. Upload binaries against the parent conversation. Rewrite inline image src attributes in conversation bodies to the new Freshdesk attachment URLs.
Load time entries and CSAT ratings last. These reference tickets that must already exist. Load time entries against their parent tickets and CSAT ratings against the matching ticket IDs.
Load Solutions content category-by-folder-by-article. Categories first, folders second, articles third. Upload translation .yml files only after each article exists.
Capture import job IDs, row counts, and error logs at every step. Log each batch's response with timestamp, count submitted, count succeeded, and count failed. This is the audit trail for Phase 6.
Risk if skipped: Lost error logs mean failures get rediscovered weeks later by users instead of caught in validation.
Cross-check destination counts against source counts at each step. After each object's load, compare the Freshdesk count to the source baseline. Investigate any delta > 0.1% immediately, not at end of cutover.
If a load step fails mid-batch, stop and diagnose. Do not push through partial data. Fix the transform, re-run idempotently keyed off the Source Ticket ID custom field, and continue.
Risk if skipped: Pushing through partial data creates orphan tickets, missing attachments, and a multi-week cleanup project.
Re-enable automation engines and notifications after all loads complete. Turn Dispatch'r, Observer, and Supervisor back on, re-enable notifications, and reconnect email channels and Freshcaller. Confirm the first inbound test ticket flows correctly.

Phase 6

Validate

Prove the migration is correct before users get in.

6a. Reconciliation

Compare total ticket count per status: source vs Freshdesk. Counts of Open, Pending, Resolved, Closed must match the source within the agreed delta threshold from Phase 0.
Risk if skipped: Count mismatch caught a week after cutover surfaces as 'where did my ticket go?' agent escalations.
Compare contact and company counts. Pull total counts of Contacts and Companies from Freshdesk and compare to source baseline. Investigate any delta.
Compare KB article counts per category and folder. Walk Solutions in Freshdesk and produce per-folder article counts; compare to source. Article-level mismatches are easier to debug than aggregate ones.
Compare open-ticket count for current workload. Operationally critical: confirm agents see the same Open + Pending ticket count in Freshdesk as they did in source. Any miss creates immediate workload distortion.
Risk if skipped: An open-ticket count delta of even 1% breaks first-response SLA reporting from day one.
Cross-check 25 randomly selected tickets field-by-field. Pull a random sample stratified across priorities, sources, and date ranges. Compare subject, description, status, priority, type, source, every custom field, requester, agent, group, tags, resolved_at.
Cross-check 25 randomly selected contacts field-by-field. Verify primary email, secondary emails, phone (E.164), company association, custom fields, and 'view all company tickets' flag.
Verify Source Ticket ID custom field populated on 100% of tickets. Query Freshdesk for tickets where Source Ticket ID is null. Any null result means an import batch failed silently and needs re-run.
Risk if skipped: Tickets without a Source Ticket ID cannot be re-imported idempotently and orphan future delta runs.

6b. Relationship validation

Verify contact-to-company associations. Sample 25 companies and confirm the expected contacts roll up. Confirm the 'one contact, multiple companies' configuration matches source if used.
Verify conversation threads attach to the right ticket. On a sample, confirm public replies and internal notes appear in chronological order under the parent ticket with correct author attribution.
Verify attachments open and render. Sample 25 attachments across PDF, image, and Office formats. Confirm each downloads cleanly and inline images render inside conversation bodies.
Verify Custom Object lookups resolve (Enterprise). Open tickets with Custom Object associations and confirm the related record displays in the ticket pane with all expected fields.
Verify Solutions visibility settings. Log in as an end user via the portal and confirm logged-in-only folders, company-specific folders, and public folders are exposed correctly.
Risk if skipped: A visibility misconfiguration exposes internal KB articles to the public portal.

6c. Audit and compliance

Verify Source Timestamp custom fields preserved original dates. Sample 25 tickets and confirm Source Created At, Source Updated At, Source Resolved At match the values from the source export.
Confirm data residency obligations are met. Verify the Freshdesk account is hosted in the agreed region (US/EU/AU/IN) and any DPA or contractual region requirements are satisfied.
Confirm HIPAA BAA execution if PHI is present. Verify the executed BAA covers the Freshworks entity hosting the account and that PHI handling matches the BAA's terms.
Risk if skipped: Discovering an unsigned BAA after PHI is loaded creates a notifiable breach.
Capture audit-log evidence from Admin > Audit Log. Export the Freshdesk audit log covering the import window. It captures agent creation, automation changes, and configuration edits as evidence for SOC2/ISO27001 controls.
Verify SLA clocks calculate correctly on a migrated open ticket. Pick a migrated ticket in Open status, manually re-trigger SLA evaluation, and confirm the response/resolution clocks reflect the original created_at against the business-hours calendar.
Risk if skipped: Wrong SLA clocks mean SLA breach reports are wrong for the entire migrated cohort.

6d. User-acceptance check

Have 5 power-user agents validate their daily workflows. Each agent runs through: open inbox, pick a ticket, reply with canned response, change status, add private note, log time. Capture issues in a shared doc.
Have a manager verify reports and views. Confirm built-in reports (Helpdesk in-depth, Tickets, Agents) show the migrated data with sensible counts. Confirm any custom views from source are recreated.
Risk if skipped: Without manager UAT, reporting gaps surface in the first board meeting after cutover.
Test inbound email arrival end-to-end. Send a test message to each support address and confirm a ticket lands in Freshdesk with the right group, agent, and product (if multi-product).
Test Freshcaller / Freshchat / WhatsApp inbound flows. For each enabled channel, place a test inbound interaction and confirm it produces a Freshdesk ticket with correct attribution.
Document every UAT issue and assign owners. Even minor issues (a missing tag on 10 tickets, a wrong CSAT rating) get logged with a fix owner and a due date.

6e. Sign-off

Get written sign-off from the migration sponsor. Sponsor confirms the Phase 0 success criteria are met and authorizes opening the account to end users. Document the sign-off email.
Risk if skipped: Without sign-off, users get access while issues are still open and post-cutover firefighting begins immediately.

Phase 7

Post-Migration Cleanup

Wrap up so the team can move on.

Communicate go-live to all agents and end users. Send login instructions, the portal URL, a link to the where-to-get-help contact, and the top three workflow differences from the old system.
Run role-segmented training sessions. Separate sessions for agents (ticket lifecycle), team leads (reports and Scenario Automations), and admins (Dispatch'r / Observer / Supervisor / SLAs). Record each.
Set the source system to read-only. Revoke write permissions for all but two admins. Announce the read-only date and the planned full-decommission date.
Risk if skipped: An accidental source-system write post-cutover muddies the historical record and confuses agents who still log in by habit.
Rebuild reports and dashboards in Freshdesk. Freshdesk Analytics does not import source reports. Recreate each report using Freshdesk's report builder, validated against source numbers for the last full month.
Re-enable and tune Dispatch'r rules. Walk each rebuilt rule, enable it, and watch the first 24 hours of triggered tickets to confirm correctness. Disable any rule firing unexpectedly.
Re-enable and tune Observer rules. Observer fires on ticket update events. Watch the first day of update-triggered actions (status change emails, group reassignments) for misfires.
Re-enable and tune Supervisor rules. Supervisor runs hourly on all tickets. Confirm the first hourly pass does not bulk-update or close historical tickets that should stay in their migrated state.
Risk if skipped: An over-broad Supervisor rule can bulk-close thousands of migrated open tickets in a single hour.
Re-publish Scenario Automations for agents. Confirm every macro from the source library has an equivalent Scenario Automation visible to the right agent groups.
Reconnect integrations to other systems. Re-enable connectors to Freshsales, Slack, calendar sync, BI tools, and any custom apps from the Freshworks Marketplace. Test each end-to-end before declaring done.
Enable Freddy AI features now that data is loaded. Freddy Copilot, Triage, and predictive prioritization need migrated history to train. Enable them after the data has settled (typically 48 hours after final load) and monitor the first week of suggestions.
Schedule a 30-day post-migration retrospective. Capture what worked, what bit the team, and what would be done differently. Update internal runbooks for the next migration.
Document the source-system decommission date. Typically 90 days post-cutover once the retention window has fully cleared. Confirm the legal team has approved the data-deletion plan.
Archive source exports to long-term storage. Keep the Phase 2 baseline export, the Phase 5 delta export, attachment binaries, and the mapping spreadsheet in a versioned bucket. Useful for forensic queries and future migrations.
Risk if skipped: Lost source archives mean any future audit or legal question about migrated data has no source-of-truth to consult.
Close out the migration project formally. Final status report, lessons-learned doc, archive the project channel, release the migration team back to BAU.

Watch list

Risks to track throughout

These risks live across multiple phases — keep an eye on them from kickoff through cutover.

Case-sensitive dropdowns and immutable field types

Freshdesk dropdown values are case-sensitive and several field types cannot be changed after creation. A 'Closed Won' source value fails against a 'closed won' Freshdesk choice silently in batch. Lock the destination schema in Phase 1, validate every dropdown value in Phase 3, and treat field-type changes after import as last-resort destructive operations.
Email reconnect creates duplicate tickets

Freshdesk fetches the last 48–72 hours of inbox history when a support mailbox is connected or reconnected. During cutover, an accidental disconnect-and-reconnect on a previously-connected mailbox pulls in two days of emails as net-new Freshdesk tickets, on top of the migration. Sequence email channel connection deliberately in Phase 5 and avoid reconnecting any mailbox already in scope.
Three automation engines hide silent dual-firing

Freshdesk's Dispatch'r (creation), Observer (update), and Supervisor (hourly time-based) engines are separate. Source-platform triggers often map to multiple Freshdesk engines, and re-enabling all three at once can fire the same action twice on a migrated ticket. Re-enable engines one at a time in Phase 7, watch the first 24 hours of triggered actions on each engine, and confirm no duplicate updates land on historical tickets.