Zendesk to Freshdesk Migration: What to Move, Rebuild or Retire

Coding Liquids tutorial cover featuring Sagnik Bhattacharya for Zendesk to Freshdesk Migration: What to Move, Rebuild or Retire.
Coding Liquids tutorial cover featuring Sagnik Bhattacharya for Zendesk to Freshdesk Migration: What to Move, Rebuild or Retire.

When you switch from Zendesk to Freshdesk, move the tickets, contacts and help centre articles, rebuild the macros, triggers, SLA policies and views by hand because they rarely port as they are, and retire anything nobody can explain, plus your Zendesk AI agent history, which cannot be exported. Then retrain Freshdesk's AI on the cleaned-up content and run both desks in parallel before cutover.

The Complete Excel Course with AI Integration

From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.

By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner

Buy now

The order matters more than the tool. Teams that start by copying tickets discover three weeks later that their replies, routing and customer-facing wording were all in Zendesk configuration rather than in the data. This checklist sorts every object into one of three piles first, and each section below tells you how to check the pile is done.

How do you sort Zendesk objects into move, rebuild or retire?

Sort every Zendesk object into move, rebuild or retire by opening a spreadsheet, before you touch Freshdesk, with one row per object. Columns: object, count, owner, verdict (move, rebuild or retire), done. Count the objects in the Zendesk admin area: macros, triggers, automations, views, SLA policies, ticket forms and fields, help centre articles, groups, and any apps installed from the marketplace.

Here is a filled-in version for an illustrative HVAC contractor with six support agents. The numbers are invented for the example, but the verdicts show the reasoning you should copy.

ObjectCountVerdictReason
Tickets, last 18 months9,400MoveCustomers write back about old jobs; agents need the thread
Tickets older than 18 months21,000Export and archiveKeep a copy, do not clutter the new desk
Contacts and companies5,200MoveMatching tickets to people needs them
Help centre articles86Move, then edit 20Fix outdated prices and model names on the way
Macros74Rebuild 25The rest are duplicates or have not been used in a year
Triggers and automations38Rebuild 12Several only compensated for old workarounds
Views19Rebuild 7Agents live in about five
SLA policies3RebuildBusiness-hours targets matter to customers
Zendesk AI agent historyn/aRetireCannot be exported; keep a summary only

The point of the table is the "rebuild 25 of 74" line. A switch is the one moment when everyone agrees that dead configuration can go, so use it. Usage data is the tie-breaker: sort macros by times used in the last 90 days and cut anything at zero without debate.

If keeping that object spreadsheet in order is the slow part, try The Complete Excel Course with AI Integration on Coding Liquids.

Can you export Zendesk tickets, and what does the export leave out?

Yes, Zendesk admins can export tickets, users and organisations as JSON, CSV or XML, but exports are off by default and come with plan conditions that surprise people. Zendesk's own export tools are documented in its help centre. The account owner has to contact Zendesk Customer Support to enable them, quoting the account subdomain. Support plans need Professional or Enterprise; Suite plans need Growth or above; and the export tools are not available on Team plans, though every plan can use the REST API.

The format changes what you get. CSV covers ticket data only and leaves out deleted tickets, ticket comments and ticket descriptions, so on its own it is no use as a record of what customers were told. JSON is the format Zendesk recommends once an account has more than 200,000 tickets. When the export finishes, Zendesk emails a download link to a compressed file, and the link stays valid for at least three days. Do not leave this task until the day before cutover, because enabling exports needs a human at Zendesk.

A raw export is for safekeeping, not for importing. To bring tickets into Freshdesk with threads, notes and attachments intact you will almost certainly use a migration service or app. Treat Freshdesk's own importers as a way to load contacts and companies, and confirm in Freshdesk's help centre whether they can load full ticket history before you rely on them for that.

Questions to ask any migration service

  • Which objects move, in writing: tickets, public and private notes, attachments, CC recipients, custom fields, timestamps, tags, agents, groups.
  • Whether it offers a free demo run on a small sample. If it does, use it to inspect threading and attachments before paying for the full run, and repeat it after you change the field mapping.
  • Whether a delta migration is included, which copies tickets changed in Zendesk while the bulk copy was running. Ask whether it costs extra or depends on the support tier you buy.
  • How inline images in help centre articles are handled, and whether they arrive as images in the body or as attachments. Check this on the demo articles that contain screenshots.
  • What happens to Sunshine custom objects. They have no direct Freshdesk equivalent, so export them yourself first.

The general method for choosing and testing a migration route is covered in the help desk migration tutorial, which also looks at what staff must relearn.

How do you map Zendesk statuses and fields to Freshdesk?

Map each Zendesk status, priority and field to a Freshdesk value before the copy runs, because a bad match is the commonest cause of tickets that look wrong afterwards. Every migration tool asks you to match them. Zendesk uses the statuses New, Open, Pending, On-hold, Solved and Closed. Freshdesk's default set is typically Open, Pending, Resolved and Closed, so one Zendesk status has nowhere obvious to go. Check the exact list in your own Freshdesk admin area, because both can be customised.

Zendesk valueFreshdesk target (check yours)Decision to make
NewOpenNone, they behave alike
PendingPendingNone
On-holdPending, or a custom status such as "Waiting on supplier"Add the custom status first if agents rely on the difference
SolvedResolvedNone
Custom field "Job type" (drop-down, 14 values)Ticket field of the same typeCreate the field and values before the copy, spelled identically
Custom field "Notes for engineer" (multi-line text)Multi-line text fieldExport JSON as well, since Zendesk CSV leaves multi-line fields out

A realistic example: the six-agent contractor skips the On-hold decision, the tool maps it to Open by default, and 140 tickets that were waiting on a part suddenly appear in the live queue on cutover morning. Agents spend a day closing them. Creating the "Waiting on supplier" status first would have taken ten minutes.

How do Zendesk macros become Freshdesk canned responses and scenarios?

A Zendesk macro is rebuilt in Freshdesk as a canned response for the reply text and a scenario automation for the ticket changes, because Freshdesk splits a macro's two jobs. Zendesk macros insert reply text and change ticket fields (status, priority, tags, assignee) at once. A canned response inserts text. A scenario automation applies a set of changes to a ticket on demand. Freshdesk's help centre describes a canned response as a reusable reply template, and a scenario automation as a bundle of actions an agent runs on a ticket, such as setting the status, setting the priority, adding a tag or pre-filling a reply. Check the exact options on your own plan when you build.

So each Zendesk macro needs a verdict on both halves. Here is one macro taken apart:

ZENDESK MACRO: "Refund approved"
Actions:
  Set status: Solved
  Add tags: refund_issued
  Set priority: Normal
  Comment (public): "Hi {{ticket.requester.first_name}}, we have approved your refund of {{ticket.ticket_field_123}}..."

FRESHDESK REBUILD
  Canned response "Refund approved" (shared, folder: Billing)
    "Hi [first name], we have approved your refund of [refund amount]..."
  Scenario automation "Refund approved"
    Set status = Resolved
    Add tag = refund_issued
    Set priority = Low

Two things go wrong here. The placeholder syntax differs, so any text with curly-brace variables needs rewriting, and hard-coded links inside a reply move across exactly as written, so they must be checked by hand. Look at the macros that mention opening hours, phone numbers or prices, because those are the ones that go stale.

Triggers, automations, views and SLAs

Triggers and time-based automations in Zendesk map loosely onto Freshdesk's automation rules, which are grouped by when they run: on ticket creation, on ticket updates and on time triggers. The concepts line up; the objects do not port. SLA policies, business hours, views, dashboards and email notification templates should all be treated as hand-built. Write each rule as one plain sentence before building it ("When a ticket is tagged billing and has no reply after 4 business hours, notify the billing group") so you can test the rule against the sentence afterwards.

Budget time honestly. A team with 12 triggers, 25 macros and 7 views is looking at a few working days of careful rebuilding, not an afternoon. Do it while Zendesk is still live, so that nothing is built under cutover pressure.

A quick macro audit with an AI assistant

If you have more than 50 macros, an assistant can speed up the sorting, because macro text is template wording rather than customer data. Paste in ten macros at a time with the customer-specific placeholders left in, and ask:

Here are 10 support macros. For each one, tell me: (1) what it does in one sentence,
(2) whether it duplicates another in this list, (3) whether it contains a price, phone
number, opening time or link that may be out of date, and (4) your suggested verdict:
rebuild, merge or retire. Do not rewrite them yet.

A typical illustrative reply flags "Call-out fee" as containing a fixed price, "Thanks for waiting" and "Sorry for the delay" as near-duplicates to merge, and "Winter promo 2024" as a retire. Treat the suggestions as a first pass: you still decide, and you still check every flagged price against the current one yourself.

What happens to the help centre and its article URLs?

Help centre articles move with a migration service, usually with categories, folders and tags, but their URLs change, so links in old tickets, email signatures and your website will break. The work you cannot delegate is the customer-facing side. Make a list of the 15 most visited articles, find every place they are linked, and update those links on cutover day. If the old help centre can stay up as a read-only copy for a month, say so on your website and give agents a short script for pointing people at the new address.

This is also the moment to correct content. Say the HVAC contractor's articles still quote last year's call-out fee. Read all 86 articles against current prices, and rewrite the 20 that are wrong before they are used to train an AI.

What AI retraining does Freshdesk need after the move?

Freshdesk's AI needs your cleaned-up help centre and your past human-written replies to learn from, because nothing about Zendesk's automated resolutions carries across. You are changing AI along with the desk. Zendesk's export documentation says AI agent tickets cannot be exported, so your automated-resolution history stays behind. That is why the content clean-up above comes first.

The costs differ and are worth stating next to each other. At the time of writing, Zendesk lists Support Team at $19 per agent a month paid yearly ($25 monthly), Suite Team at $55 yearly and Suite Professional at $115 yearly, a Copilot add-on at $50 per agent a month, and automated resolutions at $2 each on pay-as-you-go. Freshdesk lists Growth at $19, Pro at $55 and Enterprise at $89 per agent a month, billed annually. Its Freddy AI Agent comes with 500 free sessions, once per account, then $49 per 100 sessions, and the Freddy Copilot add-on is $29 per agent a month listed on Pro and Enterprise. A Freshdesk email session lasts 72 hours from the customer's first message, so one thread can be one session however many AI replies it contains. For the plan-by-plan breakdown see the Freshdesk AI plans tutorial, and for what you were buying before, the Zendesk AI allowances tutorial.

A 30-question test set before you switch the AI on

Take 30 real customer questions from the last month, spread across your top ticket types, and write the correct answer for each from the current help centre. Ask the new AI the same 30 questions in a test inbox or sandbox where customers cannot see it, and score each as correct, partly right or wrong. A filled-in line looks like this:

Q: "Can you come on a Saturday for a no-heat call-out?"
Expected: Yes, weekend emergency visits are available at the weekend rate; book by phone.
AI answer: "We are open Monday to Friday, 8am to 5pm."
Score: Wrong (article "Opening hours" is outdated and omits emergency cover)
Fix: Edit the article, add an emergency-cover paragraph, re-ask.

A wrong answer is nearly always a content problem, not an AI problem, and it is far cheaper to fix in the article than to argue with the tool. Aim for 27 of 30 correct or partly right before the AI answers any live customer, and keep it in suggestion mode for agents first if the plan allows it.

Retire: what to leave behind on purpose

  • Automations nobody can explain. If the person who built it left, and nobody has noticed it firing in 90 days, do not rebuild it.
  • Duplicate macros. Teams often hold three versions of "Thanks for waiting". Keep the best, and name it clearly.
  • Old tickets as live data. Move recent history into Freshdesk, and archive the rest as an export. A smaller desk is quicker to search and cheaper to migrate if the vendor charges per record.
  • Marketplace apps you stopped using. Check each for a Freshdesk equivalent, or for whether the job still needs doing.
  • Zendesk AI settings. Write down which topics the old AI handled well and which it escalated, so you can design the same boundaries in Freshdesk.

Mark the retired items in your spreadsheet instead of deleting the rows. Three months later, when someone asks where a certain automation went, the sheet answers the question.

Parallel run: a three-week schedule that leaves no gap

Migration services copy data over the API while both platforms stay live, so there is no technical outage. What fails is human: SLA clocks, automations and canned responses will not exist in Freshdesk on day one unless you built them during the parallel phase. This schedule assumes a team of six and a customer inbox of around 60 tickets a day.

  1. Week 1, build. Freeze configuration changes in Zendesk (ticket work carries on). Connect the source and destination, map statuses, priorities, groups and custom fields, run the demo, and rebuild the 25 macros, 12 automations, 7 views and the SLA policies.
  2. Week 2, rehearse. Run the full copy in a quiet window. Have two agents work real tickets in Freshdesk from a duplicated mailbox or internal test addresses while the rest stay in Zendesk. Run the 30-question AI test and fix the articles it exposes.
  3. Cutover day. Run the delta copy for anything changed during the bulk run. Repoint email routing to Freshdesk, which is the actual cutover, and update help centre links on your site and signatures.
  4. Week 3, watch. Keep Zendesk read-only. Check for tickets that went to the old address, replies that lost their thread, and automations that did not fire. Hold a 15-minute stand-up each morning to log problems.
  5. End of week 3, close. Final export from Zendesk, stored somewhere safe. Cancel the Zendesk subscription only once the export is checked, remembering that a monthly plan costs more than a yearly one if you timed the notice badly.

A realistic mistake from this stage: the email forwarding rule is switched on a Friday evening, a customer replies to an old Zendesk ticket over the weekend, and the reply lands in an inbox nobody watches. Prevent it by keeping the old support address forwarding for at least a month, and by checking the Zendesk side each morning of the watch week.

Cutover-day checklist

Print this and tick it live. Every line has a way to check it worked.

  • Test email sent to the support address arrives in Freshdesk within a minute, and an auto-acknowledgement goes back.
  • One migrated ticket opens with its full thread, notes and attachments, and a reply to it reaches the customer.
  • Each of the five busiest macros has a canned response and scenario automation that behaves correctly on a test ticket.
  • SLA timers start on a new ticket and show against the right business hours.
  • The AI test set still scores 27 of 30 after the final content edits.
  • Website contact form, phone-menu messages and email signatures point to the new help centre.
  • Every agent can log in, sees the right views, and knows where to find canned responses.
  • A named person owns the old Zendesk account until the final export is stored.

If a line fails, record the failure in the spreadsheet and keep the old desk live a little longer rather than patching live. For a sense of which plan to pick before you start building, the Freshdesk against Zendesk comparison covers the decision from a small team's side. If you would rather sort your own objects with someone else, that is the sort of question my AI implementation consultation covers.

Zendesk to Freshdesk migration: questions teams ask

Can I export my Zendesk data without a paid migration tool?

Often yes, for the raw records. Zendesk admins can export tickets, users and organisations to JSON, CSV or XML, but the account owner must first ask Zendesk Customer Support to enable exports, and the tools are not available on Team plans, which can use the REST API instead. Importing the result into Freshdesk with threads intact is the hard part.

Will my Zendesk AI agent history come across?

No. Zendesk's export documentation states that AI agent tickets cannot be exported in any format. Treat your automated-resolution history as something to screenshot or summarise for your own records before cutover, not as data you will carry into Freshdesk.

How long should the parallel run last?

Usually two to four weeks for a small team, long enough to see one full cycle of your busiest period. The aim is that every recurring ticket type has been handled at least once in Freshdesk, with the new automations and AI answers checked against real customer messages.

Do I need to rebuild my Zendesk triggers one for one?

No. Use the move wave as a chance to retire rules nobody can explain. List each trigger and automation, write down what it does in one line, and rebuild only those that still match how the team works. Fewer, clearer rules are easier to trust in the first month.

What should I move, rebuild or retire when switching from Zendesk to Freshdesk?

Move the tickets, contacts and help centre articles. Rebuild macros, triggers, SLA policies and views by hand, because they rarely port as they are. Retire automations nobody can explain, duplicate macros, unused marketplace apps and old tickets as live data. Zendesk AI agent history cannot be exported, so it stays behind.

Further reads

Sources: Zendesk's help article on exporting data to JSON, CSV or XML; the Zendesk and Freshdesk pricing pages; Freshdesk's support article on Freddy AI Agent sessions; a migration-tool vendor's published checklist for what moves and what is rebuilt. Read in November 2026.

Rather not work this out alone?

An AI implementation consultation is a one-to-one call with me, not an agency or a sales team. You leave knowing what AI you can use in your business and where to start.

Book a 1:1 call with me