AI Call Coaching
Guide to configuring your AI call coaches across different call types
Step 1: Define Your Call Types
Call types determine how conversations get classified and reported. Common examples:
For insurance agencies:
New Quote Call
Policy Review
Claims Discussion
Renewal Conversation
Objection Handling
Follow-up Check-in
For sales and customer success teams:
Discovery Call
Demo/Presentation
Pricing Discussion
Customer onboarding/implementation
To add a call type:
Click "Add Call Type"
Name it clearly (this appears in reports and to your reps)
Write a brief description of when a call should be classified this way
Save
Pro tip: Start with 3-4 core types. You can expand later, but too many types dilutes your coaching insights.
Step 2: Enable Coaching Per Call Type
Not every call type needs coaching. Some conversations are purely informational or administrative.
For each call type, toggle "Enable Coaching" on or off.
Should have coaching enabled:
Revenue-generating calls (discovery, demo, closing)
Customer retention calls (renewals, check-ins)
Critical support interactions (claims, complaints)
Any call where performance directly impacts outcomes
Can skip coaching:
Internal team calls
Quick administrative check-ins
Calls where the goal is just information gathering
Step 3: Write Your Coaching Prompts
This is where the magic happens. For each call type with coaching enabled, you'll write a prompt that tells the AI what "good" looks like.
What Makes a Great Coaching Prompt
Be specific about your methodology:
"Coach reps to use the SPIN questioning framework"
"Evaluate whether the rep identified the prospect's pain points in the first 5 minutes"
"Flag when reps don't mention our 24-hour claims turnaround guarantee"
Define your scoring criteria:
"Great calls have a 40/60 talk-to-listen ratio, with the prospect talking more"
"Every call should cover x, y, and z criteria"
"Reps should ask at least 3 discovery questions before presenting solutions"
"Always verify coverage amounts and deductibles during policy reviews"
Include your competitive differentiators:
"Emphasize our same-day quote turnaround vs. industry standard 48 hours"
"Reference our 98% claims satisfaction rate when handling objections"
"Position our bundling discounts early in pricing conversations"
Address common pitfalls:
"Flag when reps jump to pricing before understanding needs"
"Catch when reps miss obvious upsell opportunities"
"Identify calls where next steps weren't clearly defined"
Use AI to generate a prompt starting point
When creating a new call type you can always select "Generate prompt" to have AI make a first-pass at the call coaching prompt. This AI will consider what you have entered in the call type and description along with context of your company to generate a call coaching prompt.
Step 4: Test with Example Calls (Highly Recommended)
You can test and iterate on your prompts by testing with previously recorded calls
To test:
Select the play button on any call type
Select a call to test
Review the coaching notes as it returns
Best Practices
Start with Your Top 2-3 Call Types
Don't try to configure coaching for every scenario on day one. Focus on your highest-impact conversations first.
Be Prescriptive in Prompts
"Coach reps on discovery" is too vague. "Coach reps to ask about current provider, coverage gaps, claims history, and budget in the first 10 minutes" is specific and actionable.
Use Your Own Language
If you say "gap selling" instead of "needs analysis," use that term. If you call it "overcoming concerns" instead of "objection handling," write that. The AI will mirror your coaching style.
Include Both Positive and Negative Guidance
Tell the AI what to look for that's good ("celebrate when...") and what to flag as areas for improvement ("catch when...").
Review Coaching Output in Week One
Check how the AI is coaching your reps on actual calls. Refine your prompts based on whether the feedback is helpful and accurate.
Common Mistakes to Avoid
Generic prompts that could apply to any business "Make sure the rep builds rapport" isn't useful. "Flag when reps don't ask about family members who might need coverage" is specific and valuable.
Forgetting your differentiators Your prompts should reinforce what makes your agency unique. If your 24-hour response time is a competitive advantage, the AI should coach reps to mention it.
Over-focusing on metrics vs. outcomes Talk-to-listen ratios matter, but closing deals matters more. Balance quantitative criteria with qualitative assessment of deal progression.
Not iterating based on feedback The first prompts you write won't be perfect. Watch how reps respond to coaching and adjust your prompts to be more helpful.
Creating call types that overlap If a call could be classified as both "Discovery" and "New Quote," make the distinction clearer or combine them into one type.
What Happens Next
For your reps: After every call, they'll receive personalized coaching feedback based on the prompts you've written. They'll see what they did well and specific areas to improve.
For you: Your Performance Reports dashboard will show coaching trends across the team—common strengths, recurring gaps, and opportunities for training.
Learning curve: The AI gets better at understanding your standards over the first 10-20 calls of each type. Early coaching might feel slightly generic, but it becomes more nuanced quickly.
Iteration is Normal
Expect to refine your coaching prompts over the first 1-2 weeks:
Adjusting what gets flagged vs. celebrated
Adding criteria you initially overlooked
Removing scoring dimensions that aren't meaningful
Rewriting prompts to match how you'd actually coach in person
This is how you build a virtual coach, not just a call analyzer. Every refinement makes the coaching more valuable to your team.
Last updated