Blog · #intercom
10 min

Intercom Replacement Playbook: How to Migrate from Intercom in 2026

Step-by-step Intercom replacement guide for 2026. Export, import, widget cutover, AI training, contract timing - the full migration playbook in one focused afternoon.

Ori Lev avatarOri LevFounder, KalTalk
kaltalk

Intercom Replacement Playbook: How to Migrate from Intercom in 2026

Step-by-step Intercom replacement guide for 2026. Export, import, widget cutover, AI training, contract timing - the full migration playbook in one focused afternoon.

Migrating from Intercom is one of those projects founders put off for the same reason teams stay on tools they've outgrown - the perceived cost of switching is bigger than the actual cost of switching. The actual cost, for a team running web chat plus email at 5-10 seats, is one focused afternoon plus a week of shadow-running.

This playbook walks the migration end-to-end: what to export, how the import works, what to do with the AI training data, when to swap the widget, and how to time the contract cancellation so you don't double-pay.

Before you start: the real reason teams switch

Most Intercom migrations are triggered by one of three numbers:

  1. Per-seat invoice growth. Hiring increases the bill linearly at $85/seat on Advanced. The seventh hire feels expensive in a way the second one didn't.
  2. Fin AI per-resolution billing. $0.99 per resolution adds up fast at modest volume. 800 resolutions = $792/mo before any seat fees.
  3. The renewal renegotiation. Annual contracts get re-quoted, often with surprises. The "compare alternatives" project is usually born here.

If none of those apply, the switch isn't urgent. If two or three do, the math is straightforward: the KalTalk vs Intercom comparison shows a 7-operator + 800-resolution scenario at $95.99/mo flat versus $1,486/mo on Intercom Advanced + Fin. $16,680/yr saved at the same workload.

This guide assumes you've made the decision to switch. Now the question is execution.

Step 1 - export everything from Intercom

Intercom's export gives you everything you need in standard formats. From the Intercom workspace:

  1. Settings → Data → Export.
  2. Select the data types: conversations, contacts (companies and people), help center articles, tags, custom attributes.
  3. Pick the date range. For most migrations, "all data" is the right call - you'll filter on the receiving side.
  4. Submit the export. Intercom processes it asynchronously and emails a download link, usually within an hour for small workspaces, several hours for larger ones.

The export ships as a zip with JSON for conversations and contacts, plus the help center articles as separate files. Keep this archive somewhere stable - you'll want it as the source of truth during the cutover.

Sanity check before you proceed. Open one conversation in the JSON. Verify the threading, the tags, the customer link, and the attachments are all present. If something is missing, re-export with broader scope before you import anywhere.

Step 2 - set up the receiving workspace

Provision the new workspace before you touch the export. For KalTalk, the flow is:

  1. Sign up at kaltalk.com (Starter is free forever, no card).
  2. Configure the email subdomain so customer email starts routing.
  3. Add team members at the seat count you expect to need - Starter is 2, Pro is 3, Team is 7.
  4. Connect the chat widget in dev mode on a staging environment so you can see it without affecting production traffic.

The reason to provision before importing: you want a working empty workspace to validate the import against. If something looks wrong after import, you need to know whether it's an import issue or a config issue.

Step 3 - run the import

KalTalk's import tool ingests Intercom's export zip directly. Drop it in the workspace, the tool reconstructs:

  • Conversations with threading preserved, tags carried over, attachments attached.
  • Contacts as customer records with custom attributes mapped to KalTalk's customer field schema.
  • Help center articles ingested into the knowledge base so the AI agent can ground answers in your existing content from day one.
  • Open vs closed state preserved so the inbox queue accurately reflects work-in-progress.

The import for a mid-sized workspace (low thousands of conversations, hundreds of contacts) runs in 30-60 minutes. Larger archives take longer, mostly because of the help-center re-indexing for the AI agent. Run it overnight if your data is heavy.

Validate the import. Pick five recent conversations at random, verify threading and customer links. Pick five help center articles, verify they appear in the KB. Pick five customers, verify their attributes are present. If anything is off, re-run with verbose logging.

Step 4 - train the AI agent

Once the help center articles are in the KB, the AI agent has a starting source of truth. Two things to do here:

  1. Add anything not in Intercom's help center. Internal runbooks, product specs, recent FAQs. The agent answers from whatever you ground it in - if a doc isn't in the KB, the agent can't cite it.
  2. Run a small eval set. Pick 20 historical Intercom conversations where Fin handled it. Re-ask those questions to KalTalk's agent. Compare the answers. Look for hallucinations (confident wrong answers) and refusals (the agent says "I don't know" when it should know).

The first eval pass catches both KB gaps (the agent should know but doesn't) and KB conflicts (two articles say different things). Fix the KB, re-run the eval. Most teams iterate two or three times before the agent feels production-ready.

Export from Intercom
Provision receiving workspace
Import
Train + evalActive
Shadow week
Cutover
Migration phases: export → import → train → shadow → cutover. Each phase is a small, reversible step.

Step 5 - shadow run

Don't cut everything over on day one. Pick a subset of traffic and route it to the new tool while Intercom stays live for the rest. Three approaches teams use:

  • Behind a feature flag. New users land on KalTalk's widget, existing users stay on Intercom. Clean separation, easy rollback.
  • By landing page. /pricing and /docs go to KalTalk first; the rest of the marketing site stays on Intercom for a week.
  • By inbox. A specific email subdomain (like support-beta@) routes to KalTalk; the main address stays on Intercom.

A week is the sweet spot. Long enough to catch real customer conversations and edge cases, short enough that operators don't context-switch between two tools forever. During the shadow, run weekly reviews: AI accuracy, queue ergonomics, missing features, surprises.

If the shadow surfaces a blocker, fix it before the full cutover. If everything looks fine, schedule the cutover.

Step 6 - the cutover

Cutover day is anticlimactic when the prior steps were done well. The actual change is small:

  1. Swap the widget script. Replace the Intercom embed snippet with the KalTalk one across your domains.
  2. Update DNS for email. If you're moving to KalTalk's custom-domain email, the MX record swap routes new mail to the new workspace.
  3. Notify the team. Operators need to log out of Intercom and into KalTalk. Bookmarks, saved searches, slack notifications - all need pointing at the new tool.
  4. Set Intercom to read-only. Don't cancel the contract yet - keep Intercom available so historical conversations remain accessible during the audit period.

Most teams keep Intercom alive for 30-60 days after cutover for that audit window. The new tool has the data via import; Intercom is the backup.

Step 7 - cancel Intercom

Once the audit window closes and no one is opening Intercom anymore, cancel the contract. Two things to remember:

  1. Annual contracts. If you're mid-term, you may not get a refund for the remaining months. Some teams time the migration to the renewal date for this reason; others swallow the lost months because the savings on the new tool more than compensate.
  2. Final export. Pull a fresh export from Intercom right before cancellation as a final archive, even if you imported earlier. Belt-and-suspenders - data after import is the new tool's responsibility, but a clean Intercom export sitting in cold storage is cheap insurance.

Timeline expectations

For a typical 5-10 seat SaaS team running web chat plus email:

  • Export: 1-3 hours including Intercom's processing.
  • Provision + import: 1-2 hours.
  • Train + eval: 1-3 days, mostly KB iteration.
  • Shadow week: 7 days.
  • Cutover: 2-4 hours.
  • Audit window: 30-60 days.

Total real-time engagement: under a week of focused work. Total clock time including the audit window: 6-10 weeks. The savings start the day you cancel.

What to do if you're not switching to KalTalk

The same playbook applies for Help Scout, Front, Crisp, Tidio, and Zendesk - the export side is identical and each tool has its own import utility. The phases (export, provision, import, train, shadow, cutover, cancel) hold regardless of destination.

The destination choice is a separate question - the Intercom alternatives comparison walks the six options side-by-side with pricing, AI billing, and where each one fits.

Bottom line

Migrating off Intercom is mechanically simple - the export is clean, the import utilities on the receiving side are mature, the cutover is one widget snippet swap. The shadow week catches the remaining surprises before the full cutover.

The decision to migrate is the hard part. The execution is an afternoon plus a week.

If KalTalk is the destination, the KalTalk vs Intercom switch log has the per-feature comparison and the pricing math at your scenario. Starter is free forever for the shadow phase.

Intercom replacement FAQ

  • How do I migrate from Intercom?

    Export conversations, contacts, and help-center articles from Intercom (Settings, Data, Export). Provision the receiving workspace. Run the import utility. Train the AI agent on your help-center content. Shadow-run for a week on a subset of traffic. Swap the widget script across your domains. Cancel Intercom after a 30-60 day audit window.

  • How long does Intercom migration take?

    Active engineering time is roughly one afternoon plus a one-week shadow period. Total clock time including the post-cutover audit window is 6-10 weeks. The export takes 1-3 hours including Intercoms processing. Provisioning and import takes 1-2 hours. Train and eval takes 1-3 days. Shadow runs for 7 days. Cutover is 2-4 hours.

  • What does Intercom export include?

    Intercoms export ships as a zip containing JSON for conversations (with threading and tags) and contacts (people + companies + custom attributes), plus help-center articles as separate files. Most receiving tools import utilities ingest this format directly. Validate one conversation, one customer, and one article after import to confirm fidelity before proceeding.

  • Can I migrate from Intercom without losing data?

    Yes. Intercoms export is comprehensive and standard formats. The receiving tool reconstructs conversations with threading preserved, customers with custom attributes mapped, and help-center articles ingested into the knowledge base. The data fidelity question is whether the receiving tools import utility maps every field cleanly - validate on a small sample before committing.

  • Whats the best Intercom replacement for SaaS startups?

    For SaaS teams running web chat plus email at modest-to-rising AI volume, KalTalk is the cost-effective replacement because AI resolutions are bundled into a flat tier instead of metered per resolution. The 7-operator + 800-resolution scenario is $95.99/mo on KalTalk Team versus $1,486/mo on Intercom Advanced + Fin. For voice + SMS, Zendesk remains a stronger pick despite the higher bill.

  • Do I have to cancel Intercom on cutover day?

    No. Set Intercom to read-only on cutover day so historical conversations remain accessible during the audit window. Most teams keep Intercom alive for 30-60 days after the widget swap, then cancel. Annual contracts may not refund mid-term, so timing the migration to the renewal date saves the unused months.

  • Can I run Intercom and a replacement tool in parallel?

    Yes - this is the recommended shadow run. Route a subset of traffic (feature flag, specific landing pages, or a beta email subdomain) to the new tool while Intercom stays live for the rest. A week is enough to surface AI accuracy issues, missing features, and operator friction before the full cutover.