Over-provisioning wastes money. Under-provisioning drops calls. Getting this right requires understanding your actual traffic patterns, not guessing.
The basic formula:
Concurrent channels needed = (calls per hour × average call duration in minutes) ÷ 60
Example: 200 calls per hour × 3 minutes average = 600 call-minutes. 600 ÷ 60 = 10 concurrent channels as a baseline.
But this is the average, not the peak. And peaks are what break systems.
The Erlang B model:
Telecom engineers use Erlang B to calculate the number of channels needed to handle your peak traffic with an acceptable blocking rate (calls that can't connect). A blocking rate of 1% is standard for business use.
For most businesses, the practical rule of thumb:
Take your peak hour call volume. Add 20–30% as a buffer for unexpected spikes. That's your minimum channel count for that peak period.
Common mistakes:
Calculating for average traffic instead of peak. The average hour handles fine. The Monday morning or post-campaign rush reveals the real requirement.
Not accounting for inbound and outbound simultaneously. If your team makes outbound calls while receiving inbound, both use channels. Plan for the combined peak.
Treating channel count as fixed. SIP channels should scale dynamically. If your provider charges you for 30 fixed channels when you only need 8 for 20 hours a day — you're overpaying. On-demand channels are the right model.
Start with your CDR data. The answer is already there.
→ zvonex.com
📩 t.me/zvonex_support