Quality Assurance Protocol
Our AI-powered quality system reviews every patient call — not the industry-standard 2–5%. Here's exactly what it flags, and what we do about it.
Your patients hear our voice. That means our call quality is your call quality.
When Brook handles enrollment, onboarding, and outreach for your practice, every patient interaction reflects on your care team. Our QA system monitors every call, and if issues arise, we have a protocol in place to address them that same day.
AI-Powered Scoring
Every patient interaction is automatically analyzed by AI across three dimensions — how the rep handled the call, how the patient felt, and whether every required step was completed. Together they cover behavior, experience, and compliance.
Did the rep handle the conversation well?
0–100
Measures rep behavior on the call — the structural dynamics that decide whether a patient feels heard. Behavior patterns are scored independently of what was actually said.
How did the patient feel on the call?
0–100
Measures the patient's emotional state through tone, language, and emotional cues across the full transcript. A sentiment score is not a direct grade on the rep — a patient can arrive frustrated before the rep even speaks.
Did the rep follow the script and process?
PASS/FAIL
Every call is checked against the required script and procedural steps. This is the compliance leg — rules that must be followed on every single call, with no room for interpretation.
100%
of calls scored by AI
2–5%
industry standard for manual QA review
Source: AmplifAI, 2026
Most call centers manually review a handful of calls per rep per month. Our system scores every interaction automatically — so issues don't go unnoticed.
When any Score Drops Below 80%
Scoring calls is only useful if you act on what the scores reveal. Here's the escalation protocol that runs automatically when any call falls short.
Same Day
The system posts an alert with rep name, score, call link, and failure tags — tone, script deviation, compliance — to the QA channel. No human has to notice the problem. The system catches it.
Within 24 Hours
Every issue is diagnosed — not just documented. Was this a rep delivery issue? A confusing script? A system problem like a bad list or number reputation? We don't coach a rep for a script problem.
24–48 Hours
If it's a rep issue: targeted coaching with specific notes on what was wrong and what good looks like, plus increased QA monitoring on their next calls. If it's a script or system issue: the script gets updated or the queue gets adjusted. Every plan includes a measurable improvement indicator.
48–72 Hours
QA scores and conversion rates are re-checked. Every case is marked resolved, monitoring, or escalated. No issue sits in an ambiguous state.
Ongoing
If a rep has two or more scores below 80% in a single week, or shows no improvement after coaching, they're removed from the queue or retrained. We don't let underperformance persist on calls to your patients.
Weekly
Total alerts, top issue categories, repeat offenders, systemic patterns — and the most important section: "What We Changed This Week Based on QA." Scripts updated. Queues adjusted. Reps retrained. Documented proof that the system is learning.
Diagnose Before Acting
Every flagged call is categorized into one of three root causes. This prevents misdiagnosis — and ensures the fix actually addresses the problem.
A
Delivery issue, tone or clarity problem, missed steps in the call flow. Fix: targeted coaching, increased monitoring on next 10 calls.
B
Confusing explanation, poor wording, high "scam perception" from the patient. Fix: script updated, wording refined, tested on next cohort.
C
Wrong queue, bad list, number reputation issue. Fix: queue adjusted, campaign paused if needed, list source corrected.
How We Think About Quality
01
If it's not documented, it didn't happen. Every call, every score, every action is tracked and visible.
02
Detection happens same day. Action within 48 hours. A slow QA loop is almost as bad as no QA loop.
03
We don't coach a rep for a script problem. Root cause classification prevents wasted effort and ensures the right fix.
04
Every issue is tracked, has a documented action, and must show measurable improvement. Open issues don't stay open.
Examples
Anonymized reviews from our QA dashboard. Each one triggered coaching, a script update, or a queue adjustment — usually within 48 hours.
Script Drift · Automated Evaluation
Rep: "You'll definitely be approved for this — I'll get you set up right now."
Script requires: "Let's check your Medicare eligibility first."
Sentiment Drop · Sentiment Score
Patient (2:45): "Wait — is this one of those scam calls?"
Sentiment score dropped 64 → 28 in a single exchange.
Missed Disclosure · Automated Evaluation
Patient: "Will this cost me anything?"
Rep: "Let me get you enrolled and we'll figure that out later."
Low Engagement · Engagement Score
6-min call. Patient gave one-word answers. Discovery questions skipped.
Dismissive Tone · Sentiment + Engagement
Patient: "I'm nervous about getting confused by the device."
Rep: "Oh it's super easy, don't worry about it."
Patient went quiet; sentiment dipped; no re-engagement.
Review a real flagged call, walk through the QA dashboard, or talk through the escalation protocol with your Partner Success Manager.