Capacity Planning for Outbound: How to Size Your SIP Infrastructure Before You Need It

Why Outbound Capacity Is Different From Inbound

Inbound capacity planning is a queuing problem. You model arrival rates, service times, and acceptable wait durations, then solve for minimum agents and channels. The math is well established.

Outbound capacity planning is an optimization problem. You have a list, a campaign window, and a pacing goal. The question is not "how many channels do we need to avoid queue overflow" but "how many channels and agents produce the target result within the window, at the optimal cost structure."

Those are different questions and they require different infrastructure thinking.

The Three Capacity Variables

Every outbound capacity plan involves three interacting variables:

Agent count — the number of simultaneous agents on the dialing floor. This drives labor cost, which is typically 70 to 80% of total outbound operating cost.

Channel concurrency — the number of simultaneous SIP calls the infrastructure supports. For predictive dialing, channels need to run 2x to 4x the agent count to maintain pacing ratios. A 50-agent predictive floor might need 150 to 200 channels of SIP capacity.

Carrier termination model — how the underlying telephony charges for those channels. Per-minute models charge for every second of every connected call. Flat-rate models charge per agent seat regardless of channel utilization.

The carrier model is the only one of these three variables that is a pure cost decision rather than a performance decision.

Channel Over-Provisioning Is Free on Flat-Rate

On a per-minute carrier, every channel you provision that generates connected time costs money. There is a natural incentive to run tight on channel ratios — not so tight that pacing suffers, but tight enough that idle channels do not generate bills.

That incentive is wrong. Pacing quality improves as you give dialers more channels to work with. A predictive dialer running at 2.5x channel ratio leaves call quality on the table compared to 3.5x or 4x. But on per-minute billing, the extra channel capacity is a cost you can see on the invoice.

On flat-rate SIP, you pay per agent seat, not per channel. Provisioning 200 channels for a 50-agent predictive floor costs the same as provisioning 100. You can optimize channel ratios for campaign performance without calculating the carrier cost of doing so.

For operations teams running predictive campaigns, this removes an optimization constraint that should not have existed in the first place.

Forecasting the Fleet for 12-Month Capacity Plans

A standard outbound capacity plan covers 12 months with quarterly adjustments. The inputs are:

• Projected headcount by quarter
• Campaign mix (predictive vs. power vs. preview ratios)
• Target list sizes and expected campaign lifetimes
• Seasonal peaks and troughs
• New market entries planned

Under per-minute billing, translating those inputs to a carrier cost projection requires estimating talk time, contact rates, and average call duration for each campaign type. The estimate carries significant variance, and the 12-month projection is often wrong by 15 to 30%.

Under flat-rate billing, the projection is headcount by quarter times the per-seat rate. A 12-month projection with quarterly headcount steps produces an exact carrier cost forecast. The variance is zero, conditional on headcount accuracy — which is a much more controllable input than call intensity estimation.

SIP Trunk Sizing Rules for Common Outbound Configurations

For teams building capacity plans, these channel-to-agent ratios are standard starting points:

• Preview dialing: 1.0 to 1.2 channels per agent — agents control the dial pace, near-simultaneous usage is rare.
• Power dialing: 1.5 to 2.0 channels per agent — system dials slightly ahead, some concurrent connections.
• Predictive dialing, conservative: 2.5 to 3.0 channels per agent — FTC-compliant abandon rate management.
• Predictive dialing, aggressive: 3.5 to 5.0 channels per agent — maximum pacing, higher abandon rate, requires careful compliance management.

On a flat-rate model, you provision for the peak ratio you intend to use — not for the minimum ratio you can tolerate before pacing quality degrades.

The Capacity Buffer Problem

Every capacity plan should include a buffer — capacity above the current projected need to handle unexpected volume growth without reprovisioning. On per-minute billing, maintaining that buffer costs money even when it is not used.

On flat-rate, the buffer is seat-count-based. If you want the ability to add 10 agents to a floor on 24 hours notice, you maintain 10 additional provisioned seats. Those seats cost $99 per month each as a capacity insurance premium. For a floor that needs to respond to client volume requests quickly, $990 per month for instant surge capacity is a trivial cost.

Sales and SDR teams that run variable-intensity prospecting campaigns often maintain a 10 to 20% seat buffer specifically for this reason. On flat-rate it is affordable. On per-minute, the buffer concept does not translate the same way.