A real step-by-step guide to leaving Intercom. What to export, what doesn't port, what to expect from Fin lock-in, and how to avoid downtime.
TL;DR:
- Intercom's billing combines per-seat fees ($29-$132 per seat per month) plus $0.99 per Fin AI outcome. Teams growing past 5 seats and 1,000+ AI resolutions per month typically pay $2,500-$8,000 per month.
- The hard parts of leaving Intercom are not exporting articles. They are rebuilding custom bots, workflows, and saved replies, plus migrating in-flight conversation history.
- The Intercom Articles API and Conversations API can pull what you need. Plan for a 90-day rolling export window: the Export API caps each request at 90 days of data.
- Run Intercom and your new platform in parallel for 2-4 weeks. Do not cancel Intercom on the day you launch the new tool.
- Skip the migration if you depend heavily on Product Tours, Series, or deep Salesforce-Intercom integrations. Those do not port cleanly.
Intercom is a great product. It is also expensive in a way that punishes growth: a starter team paying $400 a month at year one quietly becomes a $4,000 a month line item by year three as the team adds seats and Fin handles more resolutions. For some teams that is worth it. For many, by 2026, it is not.
This is the playbook I wish I had when I helped three different teams migrate off Intercom in the last 18 months. It is not a sales page. Intercom does things very few competitors match. You should know what you are giving up.
Three patterns drive most migrations.
Intercom's public pricing lists Essential at $29 per seat per month (annual), Advanced at $85, and Expert at $132. A 12-person support team on Advanced is over $12,000 a year before AI usage.
The friction is not the price per seat. It is what scales with seats: every new hire adds to the bill, every part-time CX-adjacent person you want viewing tickets either costs a full seat or gets locked out.
Fin is good. The pricing is the problem. Intercom charges $0.99 per Fin AI outcome (a resolution or workflow handoff) on top of seats. Per Intercom's own pricing page, escalations triggered by Fin's default fallback are not billed, but anything that resolves is.
The math: if Fin handles 1,500 outcomes a month, that is $1,485 in AI charges on top of seats. The better Fin gets, the more you pay. Volume discounts exist on enterprise contracts but the basic shape stays the same.
Compare to flat-rate AI plans where 1,500 resolutions cost the same as 500. For teams whose AI handles a large share of volume, the per-resolution model gets uncomfortable quickly.
Intercom's AI features (Fin Agent, Fin Tasks, Copilot) live inside Intercom. You cannot run Fin against a Zendesk knowledge base. You cannot easily A/B test Fin vs another AI vendor on the same conversation surface. If Fin is not solving your specific intent mix well, your options inside Intercom are limited.
This is the real driver in many migrations: teams want to control their AI stack, evaluate models, and switch when something better ships. Intercom's bundle does not allow that.
This section saves migrations. Skip it and you will lose data you did not realize was load-bearing.
Easiest piece. Intercom's Articles API returns each article as HTML with metadata. Loop through paginated calls and dump to disk.
Two gotchas:
Pull from the Admin app under Settings or via the Conversations API for canned responses. Map each one to whatever the new platform calls them (Macros, Snippets, Quick Replies). The text translates; the keyboard shortcuts will not.
This is the painful one. Intercom's bots and Workflows are stored as proprietary JSON and there is no documented export format you can replay against another vendor's API.
What this means in practice: rebuild from scratch. Take screenshots of every bot flow in Intercom. Document each trigger, branch, and action. Plan to rebuild in your new platform's builder, which will use different concepts (triggers vs intents vs actions).
This is the single biggest migration cost. A team with 30 custom workflows should plan a week of CX-engineering work to rebuild them.
The Intercom Conversations API returns conversation objects with full message threads. The Export API caps each request at a 90-day window, so a multi-year archive takes multiple paginated jobs.
Browser-based exports cap at 10,000 rows per file; larger exports get emailed to you, which can take an hour. Plan to use the API for any serious volume.
Customer and Company objects export cleanly via the API. Segments are stored as filter definitions and have to be rebuilt in the new platform because the filter languages differ.
Tags export with conversation data. Team membership and routing rules need to be recreated by hand. If you have round-robin assignment, business hours rules, or escalation routing, document them before you start, not during cutover.
This sequence assumes you have decided on the destination platform and have admin access to Intercom.
Pull every article. Check which ones have been updated in the last 12 months. Archive or delete anything stale before you migrate.
This is the highest-leverage step. The single biggest reason post-migration AI chatbots underperform Intercom Fin is not the model; it is that the team imported a junk knowledge base full of outdated articles and the new bot grounded itself in noise.
A 200-article KB with 60 truly accurate, current articles will outperform a 200-article KB where half are out of date. Cut before you migrate.
Create the new platform account. Import the cleaned article set. Configure team members. Install the chat widget on a staging or beta surface (a /help-beta route, a single product page, an internal Slack channel for testing).
Do not point your live customers at the new chat yet. The point of step 2 is to find broken article references, missing categories, and chat widget styling issues before traffic sees them.
Take your screenshots from the documentation phase and build the equivalent flows in the new platform. Some platforms (Chatsy, eesel AI, Ada) lean heavily on AI intent recognition and let you skip building explicit decision trees for many flows. Others (Botpress, custom Rasa setups) keep the explicit flow paradigm.
Test each flow with a real test message before you ship.
There are two strategies, pick one before you start:
Strategy A: Full history migration. Use the Conversations API to export everything and use a migration tool like help-desk-migration.com or build your own importer to push into the new platform. Cost: 1-3 weeks of engineering plus possible service fees.
Strategy B: Cut over fresh with a read-only Intercom archive. Export historical conversations to S3 or a database. Keep an internal-only Intercom workspace at the cheapest seat tier for 30-60 days so agents can reference historical context. Cancel after the read-only window expires.
Strategy B is what most teams I have helped chose. Full migration is expensive and most agents do not need a 3-year-old conversation thread on day one of the new platform.
Route 10 percent of new chats to the new platform. Watch:
Compare to the equivalent Intercom numbers from the prior month. If the new platform is materially worse, fix it before increasing the percentage. If it is comparable or better, ramp to 50 percent the next week, 100 percent the week after.
Once you are at 100 percent traffic on the new platform and the metrics are stable for at least two weeks, cancel Intercom.
Two specific cancellation steps:
| Intercom feature | How to replicate elsewhere |
|---|---|
| Fin AI Agent | Any modern AI chatbot platform with RAG: Chatsy, eesel, Ada, custom RAG built on Anthropic API |
| Custom Bots / Workflows | Rebuild in destination platform's bot builder. No automated migration exists. |
| Saved Replies | Macros, Canned Responses, or Quick Replies in destination |
| Help Center | Most platforms include built-in help center; alternatively use HelpDocs, Document360, or GitBook with a chatbot pointed at it |
| Inbox + ticketing | Native inbox in Chatsy, Help Scout, Freshdesk, or Front. Self-hosted: Chatwoot. |
| Series (drip campaigns) | Customer.io, HubSpot, or Iterable. This is not a "support" tool replacement; you need a marketing automation product. |
| Product Tours | Userpilot, Appcues, Pendo. Intercom's tours are best-in-class; expect a real product evaluation here. |
| Surveys (in-app) | Sprig, Refiner, or native survey features in the new helpdesk |
| Salesforce integration | Most platforms have one, but the depth of Intercom's Salesforce sync is unusual. Test specifically before committing. |
| Phone (Intercom Calls) | Aircall, Dialpad, or a CCaaS like Talkdesk |
The takeaway: Intercom is a five-product bundle. Replacing it often means choosing two or three replacement vendors, not one. That can still be cheaper, but plan the stack first.
These come from migration retros, not theory.
Migrating the full knowledge base unfiltered. Importing 800 articles including drafts and deprecated content. The new AI grounds itself on noise and gives bad answers. Clean ruthlessly before import.
Cancelling Intercom on launch day. Inevitably something is still pointing at Intercom: a Stripe webhook, a status-page integration, a retention email. Keep Intercom alive for 30-60 days at the cheapest seat tier.
Underestimating workflow rebuild time. Teams budget two days and spend two weeks. Sophisticated branching logic takes longer to replicate than the screenshots suggest.
Migrating history nobody needs. Spending three engineering weeks moving 4 years of conversation data that no agent will ever open. If usage is below 1 percent, archive to S3 instead.
Not telling customers anything. A short proactive note prevents a wave of "is this real or hacked" tickets when the widget changes appearance.
These are the patterns where staying on Intercom is the right call.
Intercom Product Tours and Series are genuinely best-in-class for in-app onboarding. Replacing them means buying Appcues, Userpilot, or Pendo on top of your new helpdesk. The combined cost can exceed Intercom.
If your team has 50+ active Intercom Workflows that are mission-critical and stable, the migration cost is real. A reasonable rule of thumb: each complex workflow takes 4-8 hours to rebuild and test elsewhere. 50 workflows is 200-400 hours of work.
If Fin is resolving 50 percent or more of your support volume well, and your team has internalized Intercom's UI, the productivity hit of switching is real. The economic case has to clear that switching cost too, not just be cheaper on a per-month basis.
Intercom's CRM sync is unusually deep. If your sales team relies on Intercom conversation data flowing into Salesforce leads in a specific way, recreating that sync elsewhere is not a weekend project.
Five concrete things that prevent painful migrations:
How long does a typical Intercom migration take? For a 10-person support team with a moderate knowledge base, plan 4-6 weeks calendar time including content audit, parallel run, and cleanup. Engineering effort is 2-3 weeks if you have one engineer on it. Shorter is possible but rushed migrations cause the failures above.
Can I keep using Fin AI without Intercom seats? No. Fin is licensed only as part of the Intercom platform. You cannot run Fin against an external helpdesk. Alternatives that work cross-platform include Chatsy, eesel AI, and any RAG you build directly on Anthropic, OpenAI, or Google APIs.
Will I lose my customer data when I cancel Intercom? You will lose access to it via the UI, but Intercom keeps data per their retention policy and you can usually request a final export. The safer path is to export everything yourself via the API before downgrading, not to rely on a final export request after cancellation.
Is there a tool that automates the migration? Help-desk-migration.com supports Intercom as a source and many destinations. It works well for conversations, contacts, and articles. Custom bots and workflows always require manual rebuild. No tool migrates those.
Migrating from Intercom is not technically hard. It is operationally heavy, and the way most teams break a migration is by underestimating the workflow-rebuild and content-cleanup pieces. Do those well and the actual cutover is uneventful.
If you want to evaluate Chatsy as the AI-first destination, start a free account and import a subset of your help center. Run it in parallel against Intercom for two weeks before you commit to anything bigger. Or see pricing to model the cost difference at your team size and resolution volume.
Intercom's Fin charges $0.99 per AI resolution on top of seats. Chatsy is flat. Honest comparison with real cost math at 500, 1,500, and 5,000 resolutions.
