Commercial insurance agencies often mark deals closed-lost with a guess and move on. An AI analysis reads call transcripts and CRM notes at every loss, tags the true reason and reports where deals die, typically lifting win rate around six points.
Typical results
+6 pts
Win rate across two quarters
22%
Losses actually about price
100%
Lost deals analyzed automatically
₪900K
New annual premium won back
Figures reflect typical results for this kind of automation.
The challenge
The CRM's loss-reason field told the agency nothing: 80% of lost deals were tagged 'price' because that's what producers clicked to close the record. Management suspected slow quoting and coverage gaps were the real killers but had no evidence, so every quarterly strategy debate ran on anecdotes and the loudest voice in the room.
Before and after
Win rate
+25%Lost deals with a verified reason
10×Quote turnaround
−60%The automation
We trigger the flow the moment a deal moves to closed-lost: an OpenAI step reads the full record — call transcripts, email threads, CRM notes, quote timestamps — and assigns a loss reason from a taxonomy the partners defined, with a one-line evidence quote attached. Tags roll into a live dashboard and a quarterly report that breaks losses down by reason, producer, carrier, and deal size, so patterns surface instead of hunches.
How it works
- 01
Loss triggers analysis
The moment a deal moves to closed-lost, the flow picks up the full record automatically.
- 02
AI reads everything
Call transcripts, email threads, CRM notes, and quote timestamps are reviewed end to end.
- 03
Reason tagged with evidence
A loss reason from the partners' taxonomy is assigned, with a one-line evidence quote attached.
- 04
Patterns roll up
A live dashboard breaks losses down by reason, producer, carrier, and deal size.
- 05
Quarterly playbook
The quarterly report shows exactly where deals die, so fixes target the real killers.
Starts from
- Call transcripts
- CRM notes and timeline
Delivers
- True loss reason tagged
- Quarterly loss report
- Pattern alerts to partners
The impact
The first quarterly report often shows price as the real reason in only around 22% of losses, with slow quote turnaround and missing coverage lines behind far more. Fixing those can lift win rate around six points across two quarters, worth roughly ₪900,000 in new annual premium for an agency of this size.
Questions about this automation
What does the AI read when a deal is lost?
The moment a deal moves to closed-lost, the flow collects the full record: call transcripts, email threads, CRM notes and quote timestamps. An OpenAI step reviews all of it and assigns a loss reason from the agency's own taxonomy, with a one-line quote from the record attached as evidence.
Who defines the loss reasons?
The partners do. The taxonomy typically covers reasons such as price, slow quote turnaround, missing coverage lines, carrier appetite and loyalty to the current broker. The AI picks from that list only, and when the evidence is weak, it marks the deal for a manager to review instead of guessing.
How are the results reported?
Tags feed a live dashboard that breaks losses down by reason, producer, carrier and deal size. Partners receive a quarterly report with trends and examples, and an alert fires when one reason starts rising, for example several losses in a single month tied to slow quotes.