VICIdial Optimization: What the First 2 Weeks Look Like (Day by Day)

You signed the engagement. You paid the $1,000 deposit. Now what?

This article is the day-by-day walkthrough of ViciStack’s 14-day VICIdial optimization engagement. Every client gets this timeline during onboarding so there are no surprises about what happens, when it happens, and what you should expect to see in your metrics.

The engagement costs $5,000 total ($1,000 down, $4,000 on completion). The completion payment is due when we deliver measurable, documented improvement in your key metrics. We have completed over 100 of these engagements. The structure below is what has worked.

Before Day 1: The Free Audit

Before the paid engagement starts, you already have our 247-point audit report. That report identified the specific issues in your system and prioritized them by revenue impact. The engagement is about fixing what the audit found.

If you have not done the audit yet, that happens first. It takes 24 hours and is free regardless of whether you proceed with the paid engagement.

Day 0: Kickoff and Access Setup

Duration: 30-60 minutes of your time

What happens:

1. Kickoff call (30 minutes). We review the audit findings with you and your team. We agree on priorities, establish communication channels, and set expectations for the 14 days. You tell us what matters most — connect rate, compliance, cost reduction, all of the above. We build the work plan around your priorities.

2. Access provisioning. We need:

   • SSH access to all VICIdial servers (read-write for this engagement — we need to make configuration changes)
   • Admin panel access to VICIdial
   • Carrier portal access (read-only — to verify DID health and STIR/SHAKEN status)

3. Baseline metrics snapshot. We pull 30 days of historical data and establish the before numbers. These become the benchmark we measure improvement against. We document: connect rate, AMD accuracy, drop rate, agent utilization, wait time, DID answer rates, sales per agent, and any other KPIs you track.

What you see in metrics: Nothing yet. Day 0 is setup.

Day 1: AMD Initial Tuning

Duration: 3-4 hours of our time. Zero disruption to your operations.

What happens:

AMD tuning is the first change we make because it is the lowest-risk, fastest-impact optimization. We are adjusting software parameters, not changing infrastructure.

We analyze your recent call data to establish your current AMD false positive rate. This involves comparing calls dispositioned as “answering machine” against call duration patterns:

Calls classified as AM with duration 1-4 seconds = likely false positives
Calls classified as AM with duration 15+ seconds = likely correct

Then we adjust CPD parameters based on your specific demographics:

• initial_silence: How long the system waits for any sound after the call is answered. Default 2500ms is too short for older demographics.
• after_greeting_silence: How much silence after the first words before deciding it was a greeting, not a live person. Default 800ms produces false positives on people who pause after “Hello?”
• greeting: Maximum greeting length. Needs to match your target population’s typical answering patterns.
• silence_threshold: How loud a sound must be to count as speech. Background noise on cell phones requires a lower threshold than the default.

We make changes to one campaign first, run it for 2-3 hours, compare results against a control campaign, then roll out to all campaigns if the improvement is confirmed.

What you see in metrics:

Metric	Typical Day 1 Change
Connect rate	+0.3 to 0.8%
AMD false positive rate	Drops 3-5 percentage points
Agent connects/day	+3 to 8 more per agent

The improvement is immediate but modest. AMD tuning is about recovering connections that were being thrown away. It is not going to double your numbers — it is going to stop bleeding.

Day 2: AMD Refinement and Monitoring Setup

Duration: 2-3 hours of our time.

What happens:

We review Day 1’s call data in detail. The initial AMD parameters are a starting point based on demographic data, but every call center’s audio characteristics are slightly different depending on carrier routing, codec selection, and call flow.

Common refinements on Day 2:

• Adjust per-carrier parameters if you have multiple SIP trunks with different audio characteristics
• Fine-tune silence_threshold based on actual background noise levels in your call recordings
• Set up automated AMD accuracy tracking that runs daily and alerts if false positives exceed 5%

We also install our monitoring stack during this window:

# Server health monitoring - checks every 60 seconds
# Asterisk channel count, MySQL connections, disk space,
# SIP registration status, screen process count

# Alert thresholds:
# - CPU load > 8 (sustained 5 min)
# - Disk usage > 85%
# - Asterisk channels > 80% of carrier limit
# - MySQL slow queries > 10/min
# - Any SIP trunk unregistered
# - Fewer than 6 AST_ screens running

This monitoring runs for the remainder of the engagement and becomes part of ongoing management if you continue with us.

What you see in metrics:

Metric	Typical Day 2 Change
Connect rate	+0.1 to 0.3% over Day 1
AMD false positive rate	Drops another 1-2 percentage points
AMD accuracy	90-93% (up from typical 80-85%)

Day 3: AMD Finalization and DID Audit

Duration: 3-4 hours of our time.

What happens:

Morning: final AMD adjustments based on two full days of data. At this point, AMD tuning is typically complete. False positive rate should be in the 3-5% range, down from the typical 15-25% starting point.

Afternoon: complete DID audit. We check every DID in your system against the four major spam databases:

1. Hiya (used by Samsung, AT&T)
2. First Orion (used by T-Mobile)
3. TNS Call Guardian (used by multiple carriers)
4. Nomorobo (used by many smaller carriers)

We also check STIR/SHAKEN attestation level for each DID and verify that your carrier is signing calls at the A (full) attestation level.

The DID audit produces a report that looks like this:

DID	Hiya	First Orion	TNS	Nomorobo	Answer Rate	Status
(555) 123-4567	SPAM	Clean	SPAM	Clean	2.3%	Replace
(555) 234-5678	Clean	Clean	Clean	Clean	8.1%	Keep
(555) 345-6789	SPAM	SPAM	SPAM	Clean	1.1%	Replace

In the median audit, 65% of DIDs need replacement.

What you see in metrics: AMD gains are stabilized. DID report is delivered. No DID changes yet — we are ordering replacements.

Running cumulative change by end of Day 3:

Metric	Baseline	Day 3	Change
Connect rate	3.8% (typical)	4.6%	+21%
AMD false positive rate	19%	4.5%	-76%
Agent connects/day	28	34	+21%

Days 4-5: DID Replacement and Rotation Setup

Duration: 4-6 hours of our time. Your carrier may need 24-48 hours to provision new DIDs.

What happens:

This is the single highest-impact change in the entire engagement.

Day 4: We order replacement DIDs from clean providers. Key requirements:

• Numbers with no prior outbound history
• Local area codes matching your target geography
• Numbers spread across multiple underlying carriers
• Numbers with confirmed A-level STIR/SHAKEN attestation

We typically replace the DID pool with 3x the previous count. If you had 20 DIDs, we set up 60. The larger pool allows lower per-DID volume, which keeps each number below carrier flagging thresholds.

Day 5: We configure VICIdial’s CID group rotation:

CID Group Configuration:
  Rotation Method: ROUND_ROBIN
  Max Calls Per CID Per Hour: 25-35 (depends on pool size)
  Max Calls Per CID Per Day: 120-180

The per-DID caps are critical. Carrier spam algorithms flag numbers exceeding roughly 200 calls/day. By spreading volume across more DIDs, each number stays well below the threshold.

We also set up the daily DID health monitoring script that checks every DID’s answer rate against a 5% threshold. DIDs that drop below 5% get flagged for investigation and potential replacement.

What you see in metrics:

This is where the hockey stick starts.

Metric	Day 3	Day 5	Change
Connect rate	4.6%	5.8-6.2%	+26-35%
DID average answer rate	3.8%	8.5-9.5%	+124-150%
Agent connects/day	34	44-48	+29-41%

The DID improvement shows up within hours of deploying clean numbers. Customers’ phones display a normal caller ID instead of “Spam Likely,” and they answer. It really is that straightforward.

Days 6-7: DID Stabilization and Algorithm Prep

Duration: 2-3 hours of our time.

What happens:

Day 6: We monitor the new DID pool performance and make calibration adjustments. Some DIDs from the batch may underperform due to number history we could not detect in advance. We replace any underperformers.

Day 7: We analyze the calling patterns now that AMD and DIDs are optimized. This data informs the algorithm changes coming in week 2. Specifically, we need to understand:

• What is the current connect-to-agent ratio?
• How much agent idle time exists?
• What is the actual drop rate with the new DID performance?
• Are there time-of-day patterns in connect rates?

We pull this data from vicidial_log and vicidial_closer_log and build the basis for predictive algorithm tuning.

What you see in metrics by end of Day 7:

Metric	Baseline	Day 7	Change
Connect rate	3.8%	5.9-6.3%	+55-66%
AMD accuracy	82%	93.5%	+11.5 points
DID answer rate	4.2%	9.2%	+119%
Agent connects/day	28	48-52	+71-86%
Daily sales	100	165-180	+65-80%

At this point, you have already seen a major improvement. The remaining week is about squeezing out the last 15-25% of gains through algorithm optimization and fine-tuning.

Day 8: Predictive Algorithm Switch

Duration: 3-4 hours of our time.

What happens:

This is the biggest single configuration change of the engagement. We switch the dial method from whatever you are running (usually RATIO with a fixed dial level) to ADAPT_TAPERED.

RATIO mode is like cruise control locked at 65 mph. ADAPT_TAPERED is like adaptive cruise control that adjusts speed based on traffic. At shift start when all agents log in simultaneously, ADAPT_TAPERED ramps up gradually to prevent drop rate spikes. During steady-state operation, it dials more aggressively when agents are available and backs off when they are busy.

Specific settings we configure:

Parameter	Typical Before	What We Set	Why
Dial Method	RATIO	ADAPT_TAPERED	Dynamic pacing based on real-time conditions
Auto Dial Level	2.5-3.0 (fixed)	1.0 (starting point)	System auto-adjusts from here
Adaptive Maximum Level	N/A	6.0-10.0	Upper bound depends on agent count
Adaptive Dropped Percentage	N/A	3.00%	FCC compliance ceiling
Adaptive Target Drop Rate	N/A	1.50-2.00%	Keeps you safely below the 3% limit
Adaptive Intensity	N/A	25-35	How fast the system adjusts

We make this change during off-hours or at the start of a shift and monitor closely for the first 2 hours.

What you see in metrics:

Metric	Day 7	Day 8	Change
Agent wait time	14-16 sec	9-11 sec	-30 to -37%
Drop rate	2.6-3.0%	2.0-2.4%	-20 to -23%
Connect rate	5.9-6.3%	6.1-6.5%	+3-5%

The algorithm switch immediately reduces agent wait time and improves drop rate compliance. Connect rate improvement from the algorithm alone is modest — the big gains came from AMD and DIDs. The algorithm’s value is in agent efficiency and compliance.

Days 9-10: Hopper and Lead Management Optimization

Duration: 3-4 hours of our time.

What happens:

Day 9: Hopper optimization. The hopper is VICIdial’s lead queue — it pre-loads leads so the dialer always has numbers ready to dial. An undersized hopper causes the dialer to stall between loading cycles. An oversized hopper loads leads that expire before they are dialed.

Common fixes:

• Increase hopper level (most operations run 200, we set 400-600 depending on agent count)
• Optimize hopper loading frequency
• Set appropriate lead filters at the hopper level instead of relying on campaign filters

Day 10: Lead recycling and dispositioning. This is about getting more value from your existing leads. We configure recycling directly in campaign settings:

Campaign > Lead Recycling:
  NA (No Answer): recycle after 2 hours
  B (Busy): recycle after 1 hour
  A (Answering Machine): recycle after 4 hours
  DROP: recycle after 30 minutes
  CALLBK: follow scheduled callback time

Campaign > Dispositioning:
  Wrapup Seconds: 5 (was 30)
  Max Pause Time: 120
  Max Dead Time: 180

We also verify the /usr/share/astguiclient/AST_update process is feeding the hopper correctly by checking vicidial_hopper counts:

mysql -e "SELECT campaign_id, COUNT(*) AS hopper_leads,
    MIN(priority) AS min_priority, MAX(priority) AS max_priority
FROM asterisk.vicidial_hopper
GROUP BY campaign_id;"

Additional tuning:

• Implement max pause timers to catch agents hiding in pause codes
• Optimize lead ordering (fresh leads first from vicidial_list, recycled leads fill gaps)

What you see in metrics:

Metric	Day 8	Day 10	Change
Agent utilization	68-72%	74-78%	+6-8 points
Agent wait time	9-11 sec	7-9 sec	-18 to -22%
Connect rate	6.1-6.5%	6.5-6.9%	+4-8%
Connects per agent/day	50-54	54-60	+8-11%

Days 11-12: Compliance Hardening and Security

Duration: 3-5 hours of our time.

What happens:

With performance optimized, we turn to compliance and security. This protects the gains we have made.

Day 11: TCPA compliance audit and fixes:

• Verify drop rate stays below 3% under the new algorithm settings (it should — we set the target at 1.5-2.0%)
• Confirm time zone enforcement is working correctly for every list
• Verify DNC scrubbing is current (both federal and state lists)
• Check consent management workflow against current FCC rules
• Review state-specific requirements for all states you dial into (12 states have their own telemarketing laws, and Connecticut fines up to $20,000 per violation)

Day 12: Security hardening:

• Apply all pending VICIdial SVN patches (particularly CVE-2024-8503 and CVE-2024-8504 if not already addressed)
• Verify firewall rules for all public-facing ports
• Check SSH configuration (disable root login, require key-based auth)
• Configure fail2ban for SIP and web interface
• Verify STIR/SHAKEN certificate validity and renewal schedule
• Audit admin accounts and remove unused credentials

What you see in metrics: No change in revenue metrics. These days are about risk reduction. The compliance and security work protects against liabilities that can be existential — a single TCPA class action averages $6.6 million in settlement costs.

Day 13: Fine-Tuning and Edge Cases

Duration: 2-3 hours of our time.

What happens:

We review all metrics from the past 12 days and make targeted adjustments:

• Time-of-day analysis: Are there hours where connect rates spike or drop? We adjust dial intensity by hour if patterns exist.
• Per-carrier analysis: Do some SIP trunks produce better connect rates than others? We weight routing accordingly.
• Campaign-specific tuning: If you run multiple campaigns, each gets individually calibrated settings.
• Agent-level analysis: We identify agents with significantly different performance patterns and flag them for management review (not our job to manage agents, but we surface the data).

What you see in metrics:

Metric	Day 10	Day 13	Change
Connect rate	6.5-6.9%	6.8-7.3%	+3-6%
Drop rate	1.8-2.2%	1.5-1.8%	-15 to -18%
Agent utilization	74-78%	76-80%	+2-3 points

Day 14: Documentation, Handoff, and Final Report

Duration: 2-3 hours of our time. 30-minute review call with you.

What happens:

1. Final metrics pull. We document the complete before/after comparison across all tracked metrics.

2. Configuration documentation. Every setting we changed is documented: what it was, what we set it to, and why. If you ever need to rebuild or migrate your VICIdial installation, this document lets you or any admin replicate the optimized configuration.

3. Monitoring handoff. If you are continuing with ongoing management ($1,500/month), the monitoring stays in place and we transition to steady-state management. If not, we provide the monitoring scripts and alert thresholds so your team can run them.

4. Final review call. 30 minutes where we walk through the results, answer questions, and discuss next steps.

The Final Numbers

Here is what the complete before/after typically looks like at Day 14:

Metric	Baseline	Day 14	Change
Connect rate	3.8%	7.1%	+87%
AMD accuracy	82%	93.6%	+11.6 points
Drop rate	3.4%	1.8%	-47%
Agent wait time	19 sec	8 sec	-58%
Agent utilization	58%	76%	+18 points
Connects/agent/day	28	52	+86%
DID answer rate	4.2%	9.8%	+133%

These are median results across 100+ engagements. Individual results vary based on starting configuration, list quality, and market conditions. But the pattern is consistent: AMD tuning in week 1 stops the bleeding, DID replacement mid-engagement creates the biggest jump, and algorithm optimization in week 2 maximizes throughput.

The $5,000 Question

Is it worth it?

Take the median 50-agent operation. The daily sales improvement from 100 to 185 units, at $300/sale, represents $25,500 in additional daily revenue. Over 22 working days, that is $561,000/month.

The engagement costs $5,000.

The $5,000 pays for itself before lunch on Day 5.

Every day you wait is a day your agents are making 28 connects instead of 52. A day your DIDs are showing “Spam Likely” on half the phones they dial. A day your AMD is killing one in five live connections.

Get your free audit to see exactly what is fixable in your system. Or call 343-204-2353 and we will get started this week.