The angry caller
A customer whose last job went wrong and who wants a real answer before they will hear anything else.
Chime Coach · AI call centre performance coaching
Rehearse before the call, not after it.
Your people practise the awkward calls against an AI customer. Their real calls come back scored within about twenty minutes, on the same rubric. No listening in, no live monitoring.
The mistakes land on real customers, and they cost real money.
Today a new starter sits next to your best operator for a week, listens in, then goes live and you hope. The customer is the practice material.
Coach flips it. Your people rehearse the awkward calls against an AI first, so the fumbles happen in rehearsal instead of on a paying customer.
•How practice works
A real call, without the risk of a real customer
- 01
They take the call
An AI customer with a real problem, a real accent and a real reason to be short.
- 02
It gets hard on purpose
The caller pushes back, changes their mind and asks the awkward question.
- 03
The call is scored
Against your rubric, applied the same way on every attempt.
- 04
They go again
As many runs as it takes, with no paying customer on the other end.
23% to 33%
More of their leads booked by new starters who trained on Coach, against untrained new starters working the same leads.
Measured across one multi-site service business over roughly four and a half weeks, with booking outcomes read from their own job management system. Observational, not a controlled trial. Results vary.
Scoring
Scored against a rubric you decide
You set the criteria and the weighting. Greeting, discovery, empathy, booking, close, whatever matters on your phones.
Mandatory criteria are absolute. A missed booking cannot be averaged away by a warm greeting.
You decide what matters, and how much
15%
Greeting & Intro25%
Needs Discovery20%
Empathy25%
Booking Confirmed15%
Close & Next StepsMandatory criteria must score 100% — no averaging around them.
Non-negotiables
A high average never hides a missed booking
Two calls can score the same on the soft criteria and still land in different places, because one of them did not book the job.
The rubric fails the call outright when a mandatory criterion is missed, so coaching goes where it is needed.
A high average never hides a missed booking
94
88
91
79
100
96
92
90
84
60
88%Passed
86%Failed
Repetition
The same call, over and over, until it is right
A rehearsal can be run again immediately. Same scenario, same rubric, a fresh attempt.
You can see the movement attempt by attempt, so readiness is a trend rather than a hunch.
Practise again as many times as it takes
Attempt 1
58
64
52
66
61
57
60%
Attempt 2
74
79
68
77
82
71
75%
Attempt 3
91
94
86
88
100
89
91%
AI customers
Callers who behave like real ones
Forty-two AI customers across three difficulty tiers, from the polite first timer to the caller who has already had a bad week.
New starters begin easy and work up, so the hard calls are familiar before a paying customer makes one.
Forty-two AI customers, three difficulty tiers
Beginner
15 personas
ARBMCLDSEJFTGNHP
Intermediate
16 personas
IKJBKWLOMCNFODPR
Hard
11 personas
QHRMSVTEULVGWAXY
1 in 6 becomes 1 in 14
Leads lost to price objections before and after objection handling practice, bringing new starters level with the veteran benchmark.
Same customer, same period. Objection handling was the criterion that separated a booked call from a lost one most sharply.
The calls you would rather they had practised first
The confused first timer
Someone who has never booked a tradie before and does not know what to ask for or what it should cost.
The demanding commercial client
A facilities manager who wants a price, a time and an ETA in the same breath, and will not wait to be called back.
The patient regular
A long standing customer with a simple booking who still expects to be recognised and treated like a regular.
The emergency call
Water is coming through the ceiling right now and the caller needs calm, clear instructions before anything else.
The price shopper
A caller ringing five businesses in a row who leads with cost and will hang up if the answer sounds rehearsed.
The property manager
Managing six sites at once, wants everything logged against the right address and will forward the transcript to an owner.
The rushed trade partner
Another trade on site who talks fast, uses shorthand, and needs a slot confirmed before they lose signal.
The careful decision maker
Takes notes, asks about warranty, licensing and what happens if something goes wrong after the job.
The distracted caller
Half listening, background noise, keeps changing the details and needs the key facts confirmed back to them.
Course authoring
Write the course in plain English
You describe the course in plain English and the AI writes the scenarios. Write it once, and every new hire runs the same path.
Add a new scenario whenever a new type of call starts showing up, at a practice volume no trainer could ever staff.
Describe it once, the AI writes the scenarios
Course brief
“Train new starters to handle after hours plumbing calls. They must show empathy, confirm the address, quote the callout fee and always book the job.”
Generated scenarios
01Angry caller, third no-show
Hard02Price objection on a callout fee
Moderate03Confused first timer, blocked drain
Easy04Commercial client wants it today
HardEvery new hire runs the same path
70.9% versus 42.0%
Booking rates on the phones for agents who cleared their practice scenarios quickly, against those who needed the most attempts.
How many attempts a scenario takes predicts floor performance better than a single practice score does. Small sample from one customer, offered as a signal rather than a guarantee.
For the manager
You can see who is ready
Rehearsal scores and real call scores land in the same view, so at a glance you can see who is still learning, who is ready to take live calls, and who is drifting on the ones they already take.
Scoring runs automatically against the rubric you set, so no one has to sit and review recordings.
Practice readinessExample view
- Alex M.New starter62Practising
- Jordan T.Week 278Nearly ready
- Priya S.Week 391Ready for live
One scoreboard
Practice and real calls, scored the same way
Rehearsals run against AI customers, so new starters can fumble without a real caller on the line.
Real calls are pulled in through your phone system or ServiceTitan and scored on the same rubric within about twenty minutes of the call ending. Same day feedback, without a manager sitting through recordings.
71
88
65
82
99
76
73
90
67
84
62
79
96
73
70
87
64
81
98
75
93
70
67
68
61
78
95
72
69
86
64
81
98
75
92
69
66
83
60
77
95
72
89
66
63
80
97
100
91
68
86
63
60
77
94
76
88
65
62
79
Empathy · lowest column
Tom Fletcher 68
Booking Confirmed ★
Priya Raman 100
Objection Handling
Team average 76
Where the calls come from
Plugs into the call systems you already run
Coach pulls real calls out of your existing stack for scoring, and drops AI agents onto the same lines your human team runs on. It is not locked to one platform.
ServiceTitan
Primary integration. Coach reads real call recordings and transcripts out of ServiceTitan, scores them against your rubric, and returns feedback the same day, within about 20 minutes of the call ending.
Simpro
The Australian platform that actually runs at call centre scale. Coach connects the same way: pull the call, score against your rubric, hand back same day feedback.
Open architecture. If your calls live somewhere else, Coach is built to plug into any call data source, not just the two above.
Need the overflow answered too? See Chime Overflow