Dialer Channel Concurrency Planning: How Many SIP Lines Does Your Campaign Need?

The Channel Demand Formula

The number of simultaneous SIP channels your campaign needs at peak is:

`` Peak channels = (agents × pacing_ratio) + (agents × expected_concurrent_connected) ``

Breaking that apart:

• Agents × pacing_ratio is the number of outbound dials in flight at any moment (calls that have been initiated but not yet answered or abandoned).
• Agents × expected_concurrent_connected is the number of active connected calls at any moment (agents who are mid-conversation).

If you have 20 agents, a pacing ratio of 2.0, and a 35% connect rate with an 8-minute average call:

• Outbound dials in flight: 20 × 2.0 = 40 channels
• Connected calls: 20 agents × 100% (all agents should be busy at steady state) = 20 channels

But dials in flight and connected calls overlap: connected calls are a subset of dials in flight. At any given second, you have:

• 40 dials in progress (dialing phase: ringing or connecting)
• Of those 40, 20 are actively talking (the connected agents)
• The remaining 20 are still in the ringing/connecting phase

Peak simultaneous channels = 40. The 20 connected calls use 20 of those 40 channels; the other 20 are in the dialing phase.

The ceiling is the pacing-ratio-driven number. At 20 agents with a 2.0 ratio, you need 40 concurrent SIP channels as a floor. Add 20% headroom for spikes: 48 channels.

How Connect Rate Changes Channel Demand

A lower connect rate does not reduce peak channel demand — it changes the distribution of how channels are used. On a 20% connect rate list with a 2.0 ratio:

• 40 dials in flight
• Of those, 8 are connected (20% connect rate × 40 dials)
• Only 8 agents are active; 12 agents are waiting for connects

The platform will keep firing calls to fill idle agents, which means the dialing volume stays high even as the effective throughput (agents in conversation) drops. Channel demand at the network level stays at 40.

This is why connect rate does not reduce your peak channel requirement in predictive mode. If anything, on low connect rate lists the platform dials harder to fill agents, which can momentarily push channel demand above the theoretical peak while the algorithm tries to catch up.

Planning for Session Start Spikes

The highest channel demand in any session occurs in the first 3–5 minutes. The system is starting from zero — all agents are available, no calls are in flight. The algorithm fires calls aggressively to fill all agent slots simultaneously.

In a 20-agent predictive session at 2.0 ratio, the first dial burst can touch 40 channels within 10 seconds of session start. If your SIP trunk or network path is provisioned for steady-state capacity (say, 30 channels based on an average-across-session estimate), this burst will queue or drop calls in the first minute.

Always provision for burst capacity, not average. The difference between average and burst in a predictive campaign is typically 40–60%.

Inbound Channel Headroom

If you are running a blended operation with simultaneous inbound queuing, inbound channels are additive. An inbound spike of 15 concurrent calls while your outbound campaign is at peak channel demand means:

`` Total peak = outbound_peak_channels + inbound_peak_channels = 40 + 15 = 55 channels ``

This is the number your SIP gateway and network path need to support without call drops. Inbound spikes are hard to predict; add a 25–30% buffer on the inbound estimate when planning blended operations.

SIP Trunk Licensing vs Flat-Rate Seat Models

On legacy per-trunk provisioning models, channels are licensed in blocks (typically blocks of 10 or 25). You buy a block, you get that many channels, and any demand above that block results in fast-busy signals or call rejection.

On platforms with per-seat flat-rate pricing, the channel capacity question is different. UnlimCall's model provisions SIP channel capacity based on seat count, not on a separate channel license. A 20-seat floor does not need to separately provision a 40-channel SIP trunk — the seat count determines the concurrent channel envelope.

The operational implication: you do not need to manage a separate channel license inventory. You manage seat count. If your peak channel demand grows because you moved from a 1.5 pacing ratio to a 2.5 ratio, no channel upgrade license is required.

The Daily Dial Capacity Reality Check

Beyond peak concurrency, operators frequently miscalculate daily dial volume. A 20-agent predictive operation does not dial 20 calls per second for 8 hours. It runs in sessions, with agent breaks, system ramp-up, and list penetration decay.

A realistic planning model for a 20-agent, 6-hour active session:

• First 30 minutes: ramp-up, lower effective throughput (60% of peak rate)
• Hours 1–4: full throughput at peak connect rate
• Hour 5–6: list penetration decline, 70% of peak throughput

Total daily dials (rough estimate at 25% connect rate, 2.0 ratio):

• Dials attempted: 20 agents × 2 calls/agent/minute (approximate, considering AHT) × 360 minutes = 14,400 dials
• Live connects: 14,400 × 25% = 3,600 human answers
• Agent conversations: 3,600 contacts × 8 minutes / 20 agents = 24 hours of agent conversation time across 6 hours of session — roughly at 100% agent utilization

These are planning estimates, not guarantees. Your actual numbers depend on list quality, time-of-day patterns, and platform implementation.