Estimate contact-center network bandwidth by combining simultaneous voice traffic with an allowance for agent application traffic and overall capacity headroom. The voice portion is derived from staffed agents, an expected simultaneous-call percentage, and an on-wire per-call bitrate; application traffic is added separately for CRM, desktop, messaging, and other business systems.
This is a planning estimate for site links, virtual desktop environments, and WAN or internet access supporting contact-center operations. It separates voice and non-voice demand so assumptions can be adjusted independently. Actual requirements can be higher during updates, screen sharing, video, large downloads, or cloud desktop bursts, so those workloads should be reflected in the application allowance or modeled separately.
1. Enter staffed agents
Use the number of agents concurrently logged in or working during the planning peak.
2. Set call concurrency
Enter the percentage expected to be on active calls at the same time.
3. Enter voice bitrate
Use the on-wire kbps per active voice call.
4. Add application traffic
Estimate average simultaneous Mbps per staffed agent for CRM, desktop, messaging, and related tools.
5. Apply headroom
Add reserve for variability and review the combined Mbps requirement.
The model assumes the voice bitrate is a per-direction network rate and application traffic is an average simultaneous allowance.
The result is an estimated peak planning bandwidth for the combined voice and agent-application workload.
If upstream and downstream application patterns differ, run separate directional estimates rather than assuming symmetry.
Given: 250 agents, 70% on calls, 100 kbps voice bitrate, 0.25 Mbps application traffic per agent, and 25% headroom.
Calculation: Active calls = 250 × 0.70 = 175. Voice = 17.5 Mbps. Applications = 250 × 0.25 = 62.5 Mbps. Base total = 80 Mbps. With headroom: 80 × 1.25 = 100 Mbps.
Result: Plan for about 100 Mbps under these assumptions.
Contact-center agents use CRM, knowledge bases, messaging, desktops, and other services whose bandwidth can exceed the voice media itself.
Not necessarily. Occupancy and concurrency are related operational measures but can be defined differently. Enter the simultaneous on-call share that best fits your workload.
Only if the entered per-agent application allowance or voice bitrate has been adjusted for video. Video should usually be modeled explicitly because it can add substantial bandwidth.
Yes, but use a value representative of the planning peak, not a quiet-period average that understates simultaneous demand.
Applying reserve to the total protects against variability in both voice and application demand and allows some room for growth.