Intercom + Koji: From Support Conversations to Deep Customer Insight

Answer first: Intercom is where your customers talk to your team. Koji is where the same customers go deeper in an AI-moderated conversation that produces structured themes, sentiment, and quotes you can actually act on. The integration pattern is bidirectional: (1) trigger a Koji interview link from any Intercom conversation — after a ticket closes, after a successful onboarding, or as an outbound message to a segment — and (2) pipe each completed Koji interview's themes, sentiment, quality score, and a link back to the transcript into Intercom as custom user attributes and a conversation note. End-to-end this takes about 30 minutes via Intercom's outbound messages + Koji's headless API, or under an hour with no code via Zapier. With tools like Koji, the team that already lives in Intercom inbox can finally hand customers a 5-minute conversational interview that does the work of a 30-minute scheduled call.

If your CS, support, or product team already trusts Intercom as the customer surface, this is the integration that turns every closed ticket into a research data point without asking the customer to do anything more than tap a link.

Why combine Intercom and Koji

Intercom is the most common customer messaging surface in B2B SaaS — chat, email, in-app messages, ticketing, all in one inbox. Intercom's built-in surveys are fine for a one-question NPS or CSAT, but they hit a wall the moment you want depth:

• They cannot follow up on a vague answer. "It's fine" is the end of the data.
• They do not produce themes — you get a list of free-text responses and a human has to read every one.
• They do not score quality, so a one-word junk response weighs the same as a thoughtful paragraph.
• They cannot conduct a 5-10 minute conversational interview without losing the participant midway through.

Koji solves all four. AI-moderated interviews probe vague answers (AI probing guide), generate themes automatically (understanding themes and patterns), score each conversation for quality (understanding quality scores), and routinely sustain 70-85% completion rates on 5-10 minute conversations because the AI keeps the conversation natural.

The integration plugs Koji into the moments Intercom already owns: the end of a support ticket, a successful onboarding, a feature-discovery message, a churn signal in chat.

What flows in each direction

Intercom → Koji (trigger interviews from messaging moments)

• An Intercom event fires: a conversation closes with a CSAT score, a user enters a segment, an outbound message is sent.
• An Intercom webhook (or a Series step) calls your forwarder with the user's email and the Intercom conversation ID.
• The forwarder calls Koji's headless API to start an interview for that respondent (starting interviews via API).
• The forwarder also posts a follow-up message in Intercom containing the personalized Koji interview link — so the participant taps it without leaving the Intercom messenger surface they already trust.

Koji → Intercom (pipe insights back as user attributes + notes)

When each Koji interview reaches analysis_ready, the webhook payload includes themes, sentiment, quality score, and a transcript URL. Your forwarder updates Intercom:

• As custom user attributes — last_interview_at, last_interview_sentiment, last_interview_top_theme, last_interview_quality, last_interview_url.
• As a note on the originating conversation — a short summary with the top theme, sentiment, and link to the full transcript, so the next CS rep to open the ticket has the qualitative context one click away.
• Optional: as a tag — apply Intercom tags like koji_negative_sentiment or koji_pricing_friction so the inbox auto-routes future conversations from that user to the right human.

Once that data lands, Intercom segments can use the new attributes — for example, "all users with last_interview_sentiment = negative in the last 30 days" becomes a save-the-account playbook your CS team can run weekly.

Step 1 — Pick the moments worth triggering an interview

Not every Intercom interaction deserves a research follow-up. The highest-leverage moments are:

• Post-resolution on a major issue. After a 24-hour bug or billing escalation closes, send a Koji interview that asks: "Walk me through what you were trying to do when this went wrong. What were you about to do next?" That is a churn-prevention conversation worth running.
• End of onboarding. After a user completes the onboarding flow, ask them about their expectations vs. reality. See user onboarding research.
• Detected churn signal. When a high-value account stops responding or downgrades, route a Koji interview before they cancel. See cancel-flow exit interviews and win-back customer interviews.
• Feature adoption nudge. When a user starts using a new feature, queue a Koji interview 7 days later asking what they think. See feature adoption research.
• Post-demo or trial midpoint. Outbound an interview at the trial midpoint to surface friction before the conversion deadline.

For each pattern, the customer success interview guide and B2B customer research docs cover the methodology side.

Step 2 — Trigger Koji interviews from Intercom

Three paths, in order of complexity:

Path A: Zapier or Make (no code, ~30 minutes)

1. Trigger: Intercom → Conversation Closed (or any of Intercom's native triggers — new user, tag applied, segment entered).
2. Action: Koji → Start Interview, passing the user's email.
3. Action: Intercom → Send Outbound Message, containing the Koji interview link.

See Zapier research automation for the connector reference.

Path B: Intercom Series + serverless forwarder (~1 hour)

Intercom Series lets you build multi-step automations triggered by user behavior. Add a custom HTTP step that POSTs to your forwarder, which calls Koji's start interview API and returns the interview URL. A subsequent Series step uses that URL in an outbound chat message.

This path keeps the participant entirely inside Intercom's messenger — the interview link opens Koji in a tab, but the invitation feels native to the customer's existing conversation thread.

Path C: Intercom Apps Framework (production-grade)

For teams investing heavily in Intercom inbox workflows, build a small Intercom Canvas app that adds a "Send Koji interview" button to the inbox sidebar. A teammate handling a conversation taps the button to queue an interview without leaving the inbox. This is the cleanest pattern for CS-led research programs.

For any path, the user research API guide, API authentication doc, and managing research participants doc cover the credential and audience basics.

Step 3 — Pipe Koji results back into Intercom

Subscribe to Koji's interview.analysis_ready event (full reference in webhook setup). Your forwarder:

1. Verifies the Koji HMAC signature.
2. Calls Intercom's API with three writes:
   • PUT /contacts/{id} — set or update custom attributes like last_interview_sentiment.
   • POST /conversations/{id}/reply (admin note) — add a brief private note to the originating conversation with the top theme and a transcript link.
   • POST /conversations/{id}/tags (optional) — apply tags based on theme or sentiment for inbox routing.
3. Handles anonymous interviews. When respondent_email is null, skip the Intercom write entirely — there is no contact to attach to.

The full round-trip — interview completes, Koji analyzes, webhook fires, Intercom updates — typically lands in under 5 minutes from the participant tapping "submit."

What you can build in Intercom once the data lands

• Sentiment-driven inbox routing. Conversations from users with last_interview_sentiment = negative auto-route to senior CSMs instead of front-line support.
• Save-the-account segments. A dynamic segment "users whose last_interview_top_theme = pricing and last_interview_sentiment = negative" becomes a weekly outreach list.
• Onboarding personalization. Outbound onboarding messages branch based on last_interview_top_theme so the next message addresses the friction the customer just told you about.
• Inbox context for every rep. The Koji interview summary lives as a note on the originating conversation, so any rep who opens it sees the qualitative context without leaving Intercom.
• Tag-based reporting. Intercom's reporting can show conversation volume by Koji-generated tag, giving CS leadership a view of which interview themes are tied to the most support load.

Comparison: Koji + Intercom vs. Intercom Surveys alone

Intercom's native Surveys product is fine for a one-question CSAT or a quick NPS. The moment you want more, you hit limits:

• Intercom Surveys cannot probe. A free-text question gets one shot at an answer. Koji's AI moderator probes vague answers and routinely turns "it's fine" into a paragraph of specifics. (AI probing guide.)
• Intercom Surveys do not generate themes. You get a list of responses. Koji surfaces themes, sentiment, and representative quotes automatically. (Understanding themes and patterns.)
• Intercom Surveys do not score quality. Every response is treated equal. Koji scores each conversation 1-5 and you can keep only quality 3+ from flowing back into Intercom attributes. (Understanding quality scores.)
• Intercom Surveys lock you into form logic. Koji's six structured question types (open_ended, scale, single_choice, multiple_choice, ranking, yes_no — see structured questions guide) plus open-ended conversation flow give you depth and structure in the same conversation.
• Voice mode. Intercom Surveys are text-only. Koji supports voice interviews with the same AI moderator pattern when you want richer responses (voice vs text interviews).

Intercom Surveys remain the right tool for a one-question pulse. The integration above adds Koji for anything that needs depth without losing the Intercom workflow your team already runs.

Plan requirements and cost

Webhooks and the headless API are included on the Interviews plan (€79/month, 79 credits) and Enterprise. The Insights plan (€29/month) doesn't include webhooks — for that tier, the Zapier path works on any plan. Text interviews cost 1 credit, voice interviews cost 3, and only conversations scoring 3 or higher on Koji's quality gate consume credits. See plan comparison guide.

A typical post-ticket research program — every closed conversation with CSAT ≤ 7 gets a Koji interview, ~50/week for a mid-stage SaaS — uses around 30-50 credits per week if running text. That fits inside the Interviews plan with room to spare.

Identity, consent, and what stays in Intercom

For a closed-loop CS research program, identity flows naturally: the Intercom contact's email is the Koji respondent_email, the round trip uses email as the join key. For sensitive research topics, run the study in anonymous mode and skip the Intercom write — see anonymizing customer interview data.

What lives in Intercom after the integration runs is intentional and minimal: themes, sentiment, quality, an attribute timestamp, and a link to the full transcript. The transcript itself stays in Koji. That keeps Intercom out of scope for anything that classifies raw interview transcripts as sensitive data, while still giving the CS team enough context to act.

For regulated industries, see GDPR-compliant AI user research and HIPAA-compliant AI user research.

A 30-minute first run

The fastest way to test the loop:

1. In Koji, create a 5-question study targeting "customers whose last support ticket was resolved in the past 48 hours." Publish.
2. In Intercom, pull a list of contacts who match that filter (about 20-30 for a mid-stage SaaS in a typical week).
3. Use Koji's CSV participant import to load them — no integration code yet.
4. Send the personalized Koji links via an Intercom outbound message.
5. Within 48 hours you'll have themes, sentiment, and quotes in the insights dashboard. If the signal is useful, wire up the webhook → Intercom forwarder so future studies are automated.

This manual first run is usually enough to convince a skeptical CS lead that the Intercom-Koji loop is worth the engineering cost of full automation.

Related Resources

• Structured Questions Guide — the six question types that ride along on every Intercom attribute update.
• Salesforce Research Integration — sister guide for CRM-side closed-loop research.
• HubSpot Research Integration — same pattern, mapped to HubSpot CRM.
• Webhook Setup — full reference for Koji webhook events and HMAC signature verification.
• Zapier Research Automation — no-code path for the same integration.
• Customer Success Interview Guide — methodology for the CS-led research programs this integration powers.
• Cancel-Flow Exit Interviews — what to ask the customers about to leave.
• Real-Time Research Insights — how themes appear in Koji as conversations complete.