Written by: Matt Beucler, CEO, Plura AI
Key Takeaways
- VICIdial agent attrition often costs $10,000–$21,000 per replacement and usually traces back to dialer configuration, not hiring quality.
- Over-aggressive Auto Dial Levels, AMD false positives, and cluttered UIs create nonstop call pressure that pushes agents out.
- Setting Auto Dial Level to 1.5–2.0, enabling Available Only Tally, and capping Drop Percentage at 2.0–2.5% reduces stress and regulatory exposure.
- Simplifying the agent UI with screen pops, hotkeys, and custom labels cuts 90-day “job shock” attrition by removing interface friction.
- For teams ready to move beyond configuration patches, AI predictive dialers that handle AMD and qualification can remove these root causes at the architecture level.
Why VICIdial Agents Quit: Operational Pressure Points You Can Control
ContactBabel’s US Contact Center Decision-Makers’ Guide puts US contact center attrition at a 27% mean, with a third of respondents above 30%.4 Verint’s State of Agent Experience 2026 found that 31% of agents say they are likely to leave within six months.4 For a 1,000-agent operation, that translates to roughly $6.2 million per year in replacement costs before customer churn.3
In VICIdial environments, the operational causes of attrition cluster around five pressure points:
- Over-aggressive predictive dialing produces high drop rates, so agents pick up calls from customers who already experienced dead air and start the call frustrated.
- AMD (Answering Machine Detection) false positives hang up on live contacts, wasting the cost of reaching them and delivering dead air or abrupt disconnects to agents.
- Manual dialing frustration appears when the dialer does not feed calls fast enough, leaving agents idle and disengaged.
- Limited supervision and coaching tools leave agents stranded during difficult calls or feeling watched by metrics they cannot see themselves.
- A cluttered agent UI overwhelms new hires and drives 90-day “job shock” attrition, which can account for 30-40% of total turnover.
Contact centers allocate 60-70% of operating costs to agent labor, so attrition becomes the largest lever on the P&L. Gartner’s 2025 customer service and support workforce survey found that 61% of agents who voluntarily resigned cited “constant call pressure” or “never getting a moment to breathe between calls” as a primary reason.4 That pattern points to configuration as the root cause rather than hiring.
Dialer Settings That Drive Burnout and How to Fix Them
Auto Dial Level: Control Call Velocity to Protect Occupancy
The Auto Dial Level is the multiplier that controls how many lines open per available agent. When the level sits too high, the dialer places more calls than agents can handle, which produces dropped calls and back-to-back pressure with no recovery time. Calabrio’s 2025 State of the Contact Center report found that voluntary attrition climbs from 22-26% at occupancy below 80% to 42-55% above 90% occupancy. The dial level is the primary lever that controls that occupancy band.
ViciStack’s VICIdial training guide recommends setting Auto Dial Level to 1.5-2.0 for RATIO mode campaigns as a starting point. VICIfast Engineering recommends the same range for small teams of three to four agents, with quarter- or half-point adjustments and real-time monitoring during the first half hour before further changes.
For adaptive modes, ViciStack’s dial level tuning guide recommends starting at 1.5 and setting the Maximum Adapt Dial Level as a hard ceiling. Raise that ceiling only when performance data supports it. Use the Auto Dial Level Threshold as a safety guard so that when agent count dips below a set number, the dialer automatically falls back to a dial level of 1.0 and prevents a thin shift from running hot.
Available Only Tally: Count Only Agents Who Are Truly Ready
The Available Only Tally setting controls which agents the dialer counts when calculating how many lines to open. When set to “N,” the system counts agents who are mid-call or wrapping up, which inflates apparent availability and causes the dialer to open more lines than agents can receive.
ViciStack recommends setting Available Only Tally to “Y” so the dialer only counts agents who are actually in READY status. This reduces dialing efficiency slightly but acts as a reliable guard against over-dialing, especially on campaigns with longer average talk times. For campaigns with average talk times under 90 seconds, “N” can improve efficiency, while for talk times of three minutes or more, “Y” works better as the default.
Drop Percentage Limit: Stay Inside the FTC TSR Ceiling
A dropped call is a call answered by a live person that could not be connected to an agent before the line was abandoned. The FTC’s (Federal Trade Commission) Telemarketing Sales Rule (TSR) describes a ceiling of 3% of human-answered calls over a 30-day rolling average per campaign.2 Operating above that level increases regulatory risk and delivers dead air to customers who often call back angry.
VICIfast recommends setting the Drop Percentage Limit to 3% as the ceiling and running tighter at 2.0-2.5% for additional margin. ViciStack’s reporting guide recommends automated alerts when any campaign exceeds 2.5%, which gives operators time to adjust before reaching the TSR threshold.
AMD False Positives: The Hidden Attrition Driver
AMD false positives occur when VICIdial’s answering machine detection misclassifies a live human as a machine and hangs up on them or routes them to a recorded message. From the agent’s perspective, this means picking up a call expecting a live conversation and getting dead air or a disconnected line multiple times per hour. That pattern directly drives the stress and disengagement that often precede resignation.
VICIdial’s default Asterisk AMD settings drop 15-25% of live calls before they reach an agent. The default parameters were tuned for generic North American English landline calls from the early 2000s and do not match modern VoIP and cell phone networks. A Scribd technical document on VICIdial AMD false positive reduction identifies four core CPD (Call Progress Detection) parameters that control AMD behavior and recommends these starting values:
- cpd_amd_maximum_word_length: 2000ms (default ~5000ms)
- cpd_amd_minimum_word_length: 120ms (default ~100ms)
- cpd_amd_between_words_silence: 70ms (default ~50ms)
- cpd_amd_maximum_number_of_words: 3
Apply these changes one parameter at a time so you can see which adjustment improves results. Native fixes within VICIdial include routing through extension 8368 for normal carrier routing, adjusting the parameters above in /etc/asterisk/amd.conf, and setting a dial timeout of 22-26 seconds as an alternative to AMD entirely. VICIfast notes that a tuned dial timeout around 22-26 seconds lets most voicemail systems time out before answering, adds no processing cost, and produces zero false positives.
For operations running more than 300 dials per hour, third-party AI-powered AMD solutions such as amdify.io can reduce false positive rates to 1-3%. They replace Asterisk’s threshold-based detection with a neural model trained on current call recordings.
Simplify the Agent UI to Cut 90-Day Job Shock
ViciStack’s agent screen customization guide reports that poorly designed agent screens waste 10-15 minutes per agent per hour due to interface friction. Unnecessary fields, missing data, and multiple tabs for a single disposition all contribute. For new hires in their first 90 days, that friction compounds into job shock and early exits.
Use these steps to reduce UI complexity in VICIdial and make the job feel manageable:
- Disable manual dialing for new hires in User Group settings so novices stay in structured queues instead of managing their own dial cadence.
- Use screen pops via VICIdial web forms to auto-fill customer data on connect so agents do not search for lead information mid-call.
- Rename agent screen fields using Screen Labels (Admin > Screen Labels) to match your campaign’s terminology, such as changing “Address 1” to “Service Address,” with no-code configuration.
- Set up disposition hotkeys (Campaign > Disposition Hotkeys) to map keyboard shortcuts to common dispositions, saving a few seconds per call across hundreds of calls per shift.
- Configure custom transfer lists (Campaign > XFER Group) to replace typing extension numbers with named quick-transfer buttons, which reduces misrouted transfers.
- Use Campaign > Detail settings to hide non-essential buttons and panels for new teams so the screen feels focused and training time shortens.
The goal is to remove friction so agents spend more time talking and less time clicking.
Supervision and Transparency: Turn Monitoring into Support
Gallup research consistently shows that the quality of the direct manager accounts for up to 70% of variance in employee engagement scores. VICIdial provides tools that can make supervision feel supportive instead of punitive when configured with that intent.
- Start by training team leads to use the Real-Time Report (vicidial/realtime_report.php) to identify agents under severe load. Watch the agent_wait_avg column in vicidial_campaign_stats; if average wait time between calls is under 5 seconds during peak hours, agents are saturated.
- From there, use Barge-In or Whisper features to guide struggling agents mid-call without the customer hearing so coaching happens in the moment instead of only after the fact.
- Enable Agent Call Log View in User Group options so agents can see their own history and metrics, which frames performance tracking as transparency rather than surveillance.
- Add dedicated COOLDOWN pause codes for agents after escalations or abusive calls, giving 3-5 minutes of recovery time before the next contact.
ViciStack’s burnout guide summarizes this clearly: “Call centers with single-digit turnover rates run operations where agents feel like professionals and supervisors support rather than surveil.”
See Plura AI handle the repetitive calls that burn out human agents in a live demo.
How to Measure Agent Attrition in VICIdial
The standard attrition formula is: (Number of agents who left in a period / Average number of agents during that period) x 100. Tracking this monthly against your VICIdial headcount gives you a baseline. The more actionable work involves spotting early warning signals before agents resign.
VICIdial’s vicidial_agent_log table records every agent state transition, including login, logout, READY, INCALL, and PAUSED statuses, with columns for talk_sec, wait_sec, pause_sec, and dispo_sec. ViciStack’s reporting guide provides a talk time ratio query that calculates SUM(talk_sec) / SUM(talk_sec + wait_sec + pause_sec + dispo_sec) * 100 per agent over a 7-day window. Top-performing outbound agents typically hit 45-55% talk time, while values below 35% often indicate dialer tuning issues or agent behavior problems worth investigation.
For programmatic access to agent statistics without browser-based reports, ViciStack’s performance tracking guide documents the agent_stats_export Non-Agent API function. It accepts parameters including datetime_start, datetime_end, and campaign_id and returns agent-level statistics in a format suitable for external dashboards.
Early warning signals to monitor in VICIdial data, per ViciStack’s burnout guide:
- AHT (Average Handle Time) drifting upward more than 15% over 4-6 weeks
- Schedule adherence dropping more than 5 percentage points
- Pause frequency increasing more than 50%
- Quality scores declining on two or more consecutive evaluations
- Two or more unplanned absences in 30 days
Agents showing two or more of these signals at the same time merit a supportive conversation before they become a resignation statistic.
Estimate your savings with Plura AI’s ROI calculator.
The Modern Alternative: How AI Predictive Dialers Change the Attrition Math
Every fix in this guide patches a system that was not built for the attrition economics of 2026. Lowering the Auto Dial Level reduces burnout but also reduces throughput. Tuning AMD parameters improves accuracy but rarely eliminates false positives entirely. Simplifying the UI reduces friction but leaves the underlying monotony of back-to-back rejection calls in place.
Plura AI’s AI Predictive Dialer addresses these attrition drivers at the architecture level rather than only at the configuration level. Plura’s AI agents handle the dialing, AMD, qualification, and initial conversation, then connect human agents only to live, interested contacts. That shift removes dead air from AMD false positives, eases drop rate pressure from aggressive dialing, and takes repetitive manual dialing off the human floor.
Plura operates as an FCC-licensed carrier, so calls present with the company’s name instead of “Spam Likely.” That improves connect rates, which removes the frustration of working a floor where half the calls never pick up, a quiet contributor to disengagement. The platform’s AI contact center model carries a 0% turnover rate compared to the 30-45% annual attrition of traditional operations,3 because AI agents do not burn out, do not need recovery time between hostile calls, and do not leave after 13 months.
Plura supports TCPA compliance, DNC compliance, SOC 2, HIPAA, and ISO certification as platform capabilities, with real-time DNC scrubbing applied to every outbound contact before dial.1 Operators retain responsibility for their own compliance obligations, and Plura provides infrastructure that supports those programs.
Plura’s TCO of $700,000 replaces traditional $7M contact-center economics on equivalent volume, with AI agents running at 100% talk utilization versus the 40% typical of human outbound floors.3
Attrition Prevention Settings Checklist
| VICIdial Setting | Recommended Value | Impact on Attrition |
|---|---|---|
| Auto Dial Level | 1.5-2.0 (RATIO mode) | Lowers call velocity and reduces burnout from back-to-back pressure |
| Available Only Tally | Y | Prevents over-dialing and lowers dropped calls and agent frustration |
| Drop Percentage Limit | 3% ceiling; 2.0-2.5% for safety margin | Keeps campaigns within FTC TSR parameters and reduces dead air incidents |
| Agent Manual Dialing | N for new hires | Keeps novices focused on structured queues and reduces UI overwhelm |
| Agent Call Log View | Y | Gives agents metric visibility and reduces the feeling of surveillance |
| Max Calls Per Second | 1-2 (small teams) | Smooths call distribution and prevents burst drops on thin shifts |
Frequently Asked Questions
What is a safe drop percentage in VICIdial?
The FTC’s Telemarketing Sales Rule describes a 3% ceiling of human-answered calls over a 30-day rolling average per campaign. Most operators run at 2.0-2.5% to maintain a buffer. The calculation uses dropped calls divided by calls answered by a live person, not total dial attempts, so it focuses on live answers that get dropped. VICIdial’s Drop Percentage Limit setting enforces this threshold automatically in adaptive modes, and alerts at 2.5% give time to adjust before reaching the ceiling.
How do I lower the auto dial level in VICIdial?
Navigate to Admin > Campaigns > [Your Campaign] > Detail. The Auto Dial Level field sets the fixed multiplier in RATIO mode or the starting maximum in adaptive modes. For RATIO mode, start in the 1.5-2.0 range described earlier and adjust in quarter- or half-point increments. For adaptive modes such as ADAPT_TAPERED or ADAPT_HARD_LIMIT, set the Maximum Adapt Dial Level as the ceiling and let the algorithm tune within that range while you monitor the real-time campaign screen for at least two hours.
What is the difference between VICIdial and an AI predictive dialer?
VICIdial is an open-source predictive dialer that connects human agents to outbound calls, with the agent handling the full conversation from greeting to disposition. An AI predictive dialer like Plura’s handles the dialing, AMD, qualification, and often the full conversation, then connects human agents only to live contacts who have already expressed interest. For attrition, this means human agents on VICIdial absorb the full volume of rejection, dead air, and hostile callers, while AI agents on a platform like Plura absorb that volume and pass only qualified conversations to humans.
How do I measure agent attrition in VICIdial?
The attrition formula is: (Number of agents who left in a period / Average number of agents during that period) x 100. For early warning signals, query the vicidial_agent_log table to track talk time ratios, pause code frequency, and AHT trends per agent over rolling 4-6 week windows. A sustained upward drift in AHT of 15% or more, combined with increasing pause frequency and declining adherence, often signals burnout. The agent_stats_export Non-Agent API function provides programmatic access to these statistics for external dashboards and automated alerts.
Can AMD false positives cause agent burnout?
AMD false positives contribute to burnout when they misclassify a live human as a machine and hang up on them. The agent then experiences dead air or an abrupt disconnect instead of a conversation. At a 15-25% false positive rate on default Asterisk AMD settings, agents on a high-volume outbound floor can encounter this pattern dozens of times per shift. The cumulative effect mirrors other forms of back-to-back call pressure, with elevated stress and reduced confidence in the system. Tuning AMD parameters or switching to a 22-26 second dial timeout as an alternative reduces this pressure directly.
Watch how Plura AI removes AMD false positives and drop-rate pressure in a live walkthrough.
Conclusion: Prioritize Configuration Fixes, Then Consider Architectural Change
VICIdial agent attrition does not have to remain a fixed cost. It responds to specific configuration changes. The Auto Dial Level controls call velocity and occupancy. Available Only Tally and Drop Percentage Limit settings prevent over-dialing and keep you inside the TSR ceiling. AMD tuning or dial timeouts reduce dead air, while UI simplification and transparent supervision make the job feel professional instead of chaotic. Regular analysis of vicidial_agent_log and agent_stats_export data shows whether these changes move your attrition rate.
Each of these fixes reduces attrition at the margin. For teams ready to remove the root causes, Plura AI’s AI Predictive Dialer replaces legacy dialers with AI agents that handle dialing, qualification, and conversation so human agents only talk to live, interested contacts. Your agents spend their time in real conversations instead of dead air, drop pressure, and constant rejection.
See what AI dialing could save your operation with the ROI calculator. Review Plura AI’s pricing to compare plans and rates side by side.
1 Plura AI maintains SOC 2, HIPAA, ISO, and GDPR posture as part of its platform infrastructure. References to compliance frameworks in this article describe Plura’s platform capabilities and do not constitute a guarantee that any customer using Plura will themselves be compliant with applicable laws or standards. Customers remain solely responsible for their own regulatory obligations, certifications, consent management, recordkeeping, and the claims they make to their own end users. Consult qualified legal counsel for guidance specific to your use case.
2 This article describes regulatory frameworks at a general level and does not constitute legal advice. Laws and regulations vary by jurisdiction, change over time, and apply differently depending on facts and circumstances. Readers should consult qualified legal counsel before making compliance decisions.
3 Performance figures, customer outcomes, and industry statistics referenced in this article are drawn from cited third-party sources or Plura customer case studies. Individual results vary based on implementation, use case, industry, audience, and execution. Past or aggregate performance is not a guarantee of future results.
4 References to third-party products, services, companies, or research are made for informational and comparative purposes only. Plura AI is not affiliated with, endorsed by, or sponsored by any third party named in this article unless explicitly stated. Trademarks and product names referenced remain the property of their respective owners.
This article is provided for informational purposes only and reflects Plura AI’s understanding at the time of publication. Product capabilities, integrations, and specifications are subject to change. For the most current information, visit plura.ai.
This article was produced with the assistance of AI tools and reviewed by Plura AI prior to publication.