Pipedrive + Koji: Turn Deal Stages into AI Win-Loss and Post-Demo Interviews

Connect Pipedrive to Koji and every won or lost deal becomes an interview instead of a guess. When a deal changes stage, gets marked won, or slips to lost, Pipedrive fires an event that starts a Koji AI interview — voice or text — with the buyer, then writes the reason, themes, and structured scores back onto the deal. No rep chasing feedback weeks later, no win-loss study that happens once a year and is stale on arrival.

Pipedrive is where your pipeline lives: stages, deals, lost reasons, and workflow automations. Koji is the AI-native research layer that turns those pipeline events into moderated buyer interviews at scale — automatically, the moment the outcome is fresh and the buyer still remembers why they chose you or didn't.

What data flows between Pipedrive and Koji

The integration runs both directions:

• Pipedrive → Koji (trigger): When a deal moves to a stage, is marked won or lost, or an activity completes, Pipedrive's workflow automation fires an event that launches a Koji interview and sends the contact a unique link.
• Koji → Pipedrive (write-back): When the interview finishes, Koji writes an interview status, a short AI summary, the top themes, and typed scores from your structured questions back onto the deal — as custom fields and notes you can report on across the whole pipeline.

Because Koji's outputs come from structured questions with typed values — a single_choice loss reason, a scale rating of the sales experience, a yes/no "did we make the shortlist" — they land in Pipedrive as clean custom-field data you can filter, group, and chart, not a paragraph a rep has to read and re-key.

Three ways to connect

1. No-code with Zapier or Make (recommended to start). Pipedrive exposes instant triggers like Updated Deal Stage, New Deal, and Deal Matching Filter in both Zapier and Make. Wire "deal marked lost" to Koji's start-interview action. On the return trip, use Koji's "interview completed" trigger to run a Pipedrive Update Deal or Create Note action so the reason lands on the record. Most teams have a working loop in an afternoon.

2. Real-time with webhooks. Pipedrive supports webhooks (v2) and a full REST API. Subscribe to change.deal events, filter for a won/lost status change, and call Koji to launch the interview. When Koji fires its completion webhook, your handler updates the deal's custom fields and adds a note via the Pipedrive API. This is the path for high-volume sales teams that want feedback captured within minutes of the outcome.

3. AI-driven with Koji's MCP and API. Koji ships a Model Context Protocol integration (15 tools) plus a developer API, so you can drive the workflow from Claude or your own code: create a win-loss study, import lost deals from a Pipedrive filter, launch interviews, and pull structured results back for reporting — all programmatically.

The best Pipedrive moments to add interviews to

Lost-deal interviews. Pipedrive captures a lost_reason when a rep marks a deal lost — but that field is the rep's guess, not the buyer's truth. Trigger a Koji interview on every lost deal and ask the buyer directly: who did you choose, what tipped it, what would have changed your mind. Koji's AI follow-ups push past "price" to the real decision criteria, giving you a defensible loss-reason dataset instead of pipeline folklore.

Won-deal "why us" interviews. The mirror image is just as valuable. Interview new customers right after the deal closes to capture the message that landed, the objection you overcame, and the moment they decided. That is verbatim fuel for your positioning, your case studies, and your next rep's pitch.

Post-demo and mid-pipeline checkpoints. Trigger a short interview when a deal stalls in a mid-funnel stage. Understanding why a promising deal went quiet — before it is marked lost — is the difference between a save and a post-mortem.

What syncs back to Pipedrive

After each completed interview, Koji can write to the deal:

• Interview status as a custom field (invited, completed)
• A short AI summary of the conversation
• Top themes across the interview
• Structured question scores — a single_choice verified loss reason, a scale rating of the buying experience, a yes/no shortlist signal — each a typed custom field you can report on
• A note linking to the full transcript and Koji report

Now your pipeline reports can group lost deals by buyer-verified reason and segment won deals by the message that closed them — analysis you cannot run on a rep-entered dropdown alone.

Why a conversation beats a rep-entered loss reason

Structured win-loss analysis is one of the highest-leverage habits in sales, yet widely cited Gartner research suggests only a minority of organizations run it with real rigor — and teams that do report meaningfully higher win rates. The blocker has always been effort: interviewing every lost buyer by hand does not scale, so most teams fall back on a rep guessing a reason from a dropdown.

Koji removes the effort. Its AI interviewer runs the conversation in voice or text, asks adaptive follow-ups so a one-word answer becomes a real explanation, and supports all six structured question types in a single interview — open_ended, scale, single_choice, multiple_choice, ranking, and yes_no — so you capture a clean, codable loss reason and the story behind it at once. See the structured questions guide for how to combine them into a win-loss template.

Koji's quality gate keeps the dataset trustworthy: each session is scored for effort and coherence, so low-effort responses are flagged, excluded, and never consume credits. Koji's free tier includes 10 one-time credits to pilot a win-loss loop; paid plans start at €29/mo (Insights) and €79/mo (Interviews), with text interviews costing 1 credit and voice 3.

Common setup mistakes to avoid

Only interviewing losses. Won-deal interviews are where your best positioning comes from — wire up both status changes.

Letting the rep pick the reason and stopping there. Keep the Pipedrive lost_reason for speed, but treat the Koji single_choice result as the buyer-verified truth, and report on that field. When the two disagree, you have found a coaching opportunity.

No suppression window. Add a filter so a contact tied to multiple deals is not interviewed repeatedly. A "interviewed in last 90 days" exclusion keeps the outreach respectful.

Start with lost deals — the highest-signal, most-neglected moment in the pipeline — prove the write-back, then add won-deal and mid-funnel interviews once the loop is landing cleanly on your records.

A worked example: a continuous win-loss loop

Say your team closes 40 deals a month, winning 12 and losing 28. Wire a Pipedrive workflow so that any deal moving to Won or Lost sends a webhook to Koji. Koji invites the buyer to a five-minute voice or text interview built from four structured questions: a single_choice "primary reason for your decision," a scale rating of the sales experience, a yes/no "did we make your shortlist," and one open_ended "what would have changed your decision." The AI probes each answer with an adaptive follow-up.

Within a day, the completed interviews write back to each deal: a buyer-verified reason in a dropdown custom field, a satisfaction number, and a note linking to the transcript. At month end you filter the pipeline by that dropdown and discover that "missing integration" — not price — drove a third of your losses. That is a product-roadmap decision you could never have reached from the rep-entered field alone, produced with zero manual interviewing. Repeat it every month and win-loss stops being an annual project and becomes a living signal.

Related Resources

• Structured Questions Guide — the 6 question types that make win-loss data codable
• Win-Loss Analysis Guide — how to run a rigorous program
• HubSpot + Koji Integration — the same pattern for HubSpot deals
• Salesforce + Koji Integration — for Salesforce opportunities
• Zapier Research Automation — build the no-code loop
• Webhook Setup — the real-time path
• How the Quality Gate Works — why low-effort responses never pollute your data