Forethought to Fin migration takes 2–4 weeks for most teams. The core difference: Forethought used a trained intent classifier on your historical tickets; Fin uses knowledge source retrieval + AI synthesis. Teams with well-structured help centers migrate faster. The three things that break most often: implicit intents with no knowledge source match, custom CRM actions that need rebuilding in Intercom's workflow engine, and a 2–4 week CSAT dip during the Fin learning period.
Why Teams Are Migrating Now
Forethought was acquired by Zendesk in March 2026. Teams not on Zendesk — or on Zendesk but unhappy with the new pricing and roadmap direction — are evaluating exits. Fin is the most common landing spot because Intercom has aggressive migration incentives and Fin's resolution rates are well-documented.
The push factors vary: some teams see Forethought's roadmap stalling as the Zendesk integration deepens. Others are concerned about pricing changes post-acquisition. A smaller group is proactively moving because their platform contract is coming up for renewal and they want to evaluate options before being locked in.
What Changes in Your Stack
| Component | How Forethought Handled It | How Fin Handles It | Migration Effort |
|---|---|---|---|
| Intent classification | Trained ML classifier on your historical tickets — learned from past volume | Knowledge source retrieval + GPT-4 synthesis — no pre-training needed | Medium — must map Forethought intents to knowledge sources |
| Knowledge base | Ingested your help center articles + trained on ticket history | Reads connected knowledge sources (Intercom Articles, Zendesk Guide, Confluence, etc.) | Low — if your help center is current |
| Escalation routing | Rule-based + intent-triggered handoff to Zendesk/Salesforce queues | Fin routes to Intercom Inbox; custom routing via Intercom workflows | Medium — routing logic must be rebuilt in Intercom |
| CRM/API actions | Custom actions via Forethought's integration layer (order lookup, account status) | Custom actions via Fin Actions (REST API calls from within the conversation) | High — each action needs rebuilding and testing |
| Agent assist | Forethought Triage: auto-tags, suggests responses to agents in Zendesk sidebar | Fin is customer-facing; Intercom Copilot is the agent-assist equivalent | Medium — separate product, separate evaluation |
| Analytics | Deflection rate, containment, intent breakdown, ticket volume by category | Resolution rate, CSAT, handoff rate, topic breakdown in Intercom Reports | Low — different labels, same underlying metrics |
| Pricing model | Per-contact or per-seat, depending on plan | Per-resolution (Fin charges only for conversations it resolves) | N/A — model change may reduce cost at high volume |
What Breaks Most Often
1. Implicit intents with no knowledge source match
Forethought could classify and respond to intents it learned from historical tickets — even if you had no help article written for them. Fin needs a knowledge source to retrieve from. If your Forethought was handling questions like "my account was double-charged" or "why is my payment showing as pending" based on ticket training alone, and you have no help article for those topics, Fin will escalate those to a human instead of resolving them.
Fix before cutover: Export your top 50 Forethought intents by volume. Map each to an existing article. For intents with no article, write one before launch.
2. Custom CRM actions that need rebuilding
If Forethought was wired to look up order status, check account standing, or pull customer data from your CRM mid-conversation, those integrations need to be rebuilt as Fin Actions. Each action requires a REST API endpoint on your end, authentication setup, and testing. This is the highest-effort part of the migration and should be scoped before you commit to a cutover date.
3. CSAT dip in the first 2–4 weeks
Fin's performance improves as it indexes your knowledge base and accumulates conversation data. The first two weeks typically show lower resolution rates and CSAT than your Forethought baseline. This is expected — plan for it in your stakeholder communication and don't judge the migration on week-one numbers.
Migration Timeline
| Phase | Duration | Key Work |
|---|---|---|
| Audit | 3–5 days | Export Forethought intent volume report. Map top 50 intents to knowledge sources. Identify gaps. |
| Knowledge base prep | 3–7 days | Write missing articles. Update outdated ones. Connect knowledge sources to Fin. |
| Actions rebuild | 3–10 days | Rebuild each custom CRM action as a Fin Action. Test each individually. |
| Routing rebuild | 2–4 days | Rebuild escalation routing in Intercom workflows. Map Forethought rules to Intercom equivalents. |
| Parallel run | 7–14 days | Run Fin on 10–20% of traffic. Compare resolution rate and CSAT to Forethought baseline. |
| Full cutover | 1 day | Flip traffic to Fin. Monitor for 48 hours. Keep Forethought available for rollback. |
Pre-Cutover Checklist
- Exported Forethought's top 50 intents by resolved volume
- Every top intent maps to at least one knowledge source Fin can retrieve from
- All custom CRM/API actions rebuilt and tested as Fin Actions
- Escalation routing rebuilt in Intercom workflows — tested with sample conversations
- Fin connected to all knowledge sources (Intercom Articles, Zendesk Guide, Confluence, etc.)
- Fin tested on 10–20% of live traffic for at least 7 days
- Fin resolution rate on test traffic ≥ Forethought baseline (or documented gap with plan)
- CSAT monitoring in place — alerts set for drops >5 points
- Stakeholders notified of expected 2–4 week learning dip
- Forethought account kept active for rollback window (30 days recommended)
How to Run Parallel
The safest migration runs both tools simultaneously on split traffic before full cutover. The practical approach:
- Route 10–20% of new conversations to Fin while Forethought handles the rest. In Intercom, this is a conversation routing rule by percentage.
- Track three metrics for 7+ days: resolution rate (Fin vs. Forethought baseline), escalation rate, and post-resolution CSAT.
- Identify gaps — any topic category where Fin's resolution rate is materially lower than Forethought's indicates a knowledge source gap or a missing action.
- Close gaps, then increase Fin traffic to 50% for another week.
- Full cutover only once Fin matches or exceeds Forethought on test traffic.
If You're Not on Intercom
Fin is optimized for Intercom's messenger. If your support platform is Zendesk, Salesforce Service Cloud, Freshdesk, or a custom setup, you have two options:
- Migrate to Intercom + Fin together. This is the full stack change — more work upfront, but you get the deepest Fin integration. Plan 6–10 weeks total.
- Use Fin API with your existing platform. Fin's API allows custom channel integrations, but you lose the native Intercom workflows, routing, and analytics. The integration overhead is significant.
If you're on Zendesk and looking to stay there, the better path may be evaluating Forethought alternatives that are Zendesk-native rather than forcing a Fin integration. Decagon and Cleverly are worth evaluating in that scenario.
Frequently Asked Questions
How long does a Forethought to Fin migration take?
2–4 weeks for most teams doing a clean cutover. Teams with complex custom actions (CRM lookups, account management flows) should plan for 4–6 weeks. The knowledge base audit and actions rebuild are the longest phases.
Does Fin work without Intercom?
There's a Fin API for custom integrations, but the native product is built for Intercom's messenger. Teams not on Intercom will face significant integration work and lose Intercom's native routing and analytics. If you're staying on Zendesk, there are better-fit alternatives.
What does Forethought do that Fin doesn't?
Forethought's intent classifier trained on your historical ticket data — it was strong at handling sparse, ambiguous queries because it learned from your specific patterns. Fin relies more on knowledge source retrieval, which means it needs well-written articles to be effective. Forethought also had deeper Zendesk-native features (Triage, auto-tagging, agent assist sidebar) that Fin doesn't replicate directly.
Should I run Forethought and Fin in parallel?
Yes. Run Fin on 10–20% of traffic for at least 7 days before full cutover. Compare resolution rate and CSAT against your Forethought baseline. Only cut over completely when Fin matches Forethought on the test segment.
What breaks most during the migration?
Three things: (1) intents Forethought handled that have no matching knowledge source for Fin, (2) custom CRM actions that need rebuilding in Fin's action framework, and (3) an expected CSAT dip in the first 2–4 weeks. All three are addressable — the checklist above covers the pre-launch work to minimize them.
Need Help Scoping the Migration?
We've mapped this migration for multiple support teams. Tell us your current Forethought setup — intents, custom actions, platform — and we'll scope the cutover honestly.
Related: Forethought alternatives for non-Zendesk teams · Intercom Fin alternatives · AI implementation cost guide