AMD Accuracy Under the FTC's 3% Rule: Why Misclassifying Humans as Machines Threatens TCPA Compliance

The 3% rule: how AMD accuracy keeps you TCPA-compliant

The FTC’s 3% abandoned call rule is gamed by operators who aggressively tune AMD to drop calls fast. Stock heuristic AMD in Asterisk’s app_amd misclassifies 5–15% of live humans as machines, hiding real dropped calls from compliance dashboards.

Why This Matters

The FTC’s Telemarketing Sales Rule caps abandoned calls at 3% of live-person answers, but stock AMD solutions like Asterisk’s app_amd achieve only 70–85% accuracy, misclassifying 5–15% of live humans as machines. This allows operators to report low abandonment rates while silently dropping real people—defeating the rule’s purpose. Accurate detection (e.g., 99% accuracy in under 200 ms) removes this tradeoff, keeping operations genuinely compliant instead of cosmetically so.

Key Insights

• FTC Telemarketing Sales Rule caps abandoned calls at 3% of live-person answers, per campaign over 30 days (FTC).
• Stock AMD in Asterisk’s app_amd achieves 70–85% accuracy; misclassifying 5–15% of live humans as machines, per developer analysis (AMDY IO, 2026).
• AMDY classifies acoustic signature of answer audio at 99% accuracy in under 200 ms, avoiding human drop misclassification (AMDY IO, 2026).
• Aggressive AMD tuning lowers reported abandonment rate but increases actual human drops—a compliance gray area noted in article (AMDY IO, 2026).

Practical Applications

• Use case: Operators using AMDY achieve 99% accuracy, keeping agent-efficient pacing without aggressive tuning; reduces human misclassification from 10% to near zero, saving hundreds of people per 10,000 daily dials. Pitfall: Tuning AMD aggressively to drop ‘machines’ fast (common anti-pattern); consequence: real people get dropped as machines, creating hidden non-compliance even as dashboard shows green.
• Use case: AMDY’s sub-200 ms decision leaves nearly all of the two-second window for connecting a live rep, supporting timely connection requirements. Pitfall: High detection latency (e.g., stock AMD delays) eats into the window; consequence: reps cannot connect within 2 seconds, increasing true abandoned calls.
• Use case: AMDY logs per-call classifications with timing, providing queryable evidence for FTC safe-harbor recordkeeping. Pitfall: Relying on a single ‘machine’ flag without per-detection logs; consequence: weak proof of compliance during audits.

References:

• https://dev.to/amdy/the-3-rule-how-amd-accuracy-keeps-you-tcpa-compliant-1k80