How Multiple Agents Increase Campaign Capacity

Multiple agents increase a MYLO human campaign’s capacity by creating several independent opportunities for a live customer conversation. If five agents are eligible, available, and supported by sufficient trunk channels and system resources, the campaign may operate up to five active human call flows rather than one. That is not a promise of five simultaneous calls at every moment. Each flow depends on the selected customer-first or agent-first connection type, the agent’s current availability, caller-ID eligibility, provider capacity, pacing, and other active calls. MYLO’s deterministic scheduler should admit work only when the required resources exist. It should not dial extra customers merely because a lead list is large. People and telephony resources jointly define safe capacity.

One Active Conversation per Available Human

In a human campaign, one person can normally handle one active customer conversation at a time. An agent should not be treated as available for another lead until the prior flow has ended and any configured recovery or wrap period has passed. This is different from an automated IVR that may serve several independent calls within channel and system limits.

Availability must come from current operational state, not from a static employee list. A logged-out, paused, unreachable, or already active agent should not increase admitted campaign capacity. The exact states supported by the current product should be documented and tested.

A Five-Agent Example

Consider a campaign with five trained agents, each assigned an office extension and authorised caller IDs. If all five are available and the provider and appliance can support the required legs, the scheduler can run five independent lanes. Each lane claims a lead, follows its configured connection order, connects one customer to one agent, records attempt evidence, and waits for that agent’s pacing clock before the next eligible lead.

If two agents pause and one endpoint becomes unavailable, only two lanes remain ready. The campaign should reduce new work automatically rather than continue dialling for five. When an agent becomes eligible again, that lane can resume according to deterministic state and pacing.

Connection Order Changes Resource Use

In an agent-first flow, MYLO reaches the agent extension or mobile before originating the customer leg. This can reduce the risk that a customer answers while nobody is ready, but it occupies the first leg before the customer call begins. In a customer-first flow, the customer is called first and the agent leg follows after the configured answer condition.

Both are two-leg human flows when connected. They have different experience and failure implications, so capacity is not only a count of agents. The scheduler must reserve and release the appropriate resources for the chosen flow without confusing a theoretical slot with an active channel.

Trunk Channels Can Be the Real Limit

A provider contract may limit simultaneous external legs. A mobile-agent-to-customer flow can require two external legs per connected conversation, while an extension-to-customer flow may use one external customer leg plus an internal extension leg. The exact topology must be counted correctly.

Five agents do not create ten provider channels. If the trunk supports only four external legs and each active mobile-agent flow needs two, provider capacity may allow only two such conversations. Inbound calls and other authorised workloads also need headroom under the deployment’s capacity policy.

Unavailable Agents and Reassignment

When an agent cannot be reached, the result should be classified as an agent-side failure, not automatically recorded as a customer failure. MYLO should not retry the customer merely because the assigned person or device failed. It can return the lead to an eligible state or reassign it under the campaign’s deterministic policy without consuming a customer retry allowance.

Reassignment must avoid two agents calling the same lead. Claiming needs atomic ownership and idempotent attempt creation. If the appliance restarts, reconciliation should inspect live and stored evidence before deciding that a lane is free.

Per-Agent Pacing Protects the Team

Each agent can have an independent pacing clock after a completed or failed flow. This gives time for notes, recovery, or another required task. Global batching—calling several leads every fixed interval regardless of which agents just finished—can overload some people while others remain idle.

Adding agents therefore adds paced lanes, not one undifferentiated dialling rate. The scheduler should calculate eligibility per lane and admit only the next safe unit of work. Business owners choose the pacing value within current product constraints and employment practices.

Practical Example: Mixed Office and Remote Agents

A service campaign has three office agents on extensions and two remote agents on mobiles. The office flows require one external customer leg each, while remote mobile flows require an external agent leg and an external customer leg. The provider supplies eight channels, and the capacity policy keeps headroom for inbound service.

MYLO calculates admission from current agents, caller IDs, channel requirements, active calls, and pacing. It does not assume that five names mean five immediate conversations. When a remote phone is unreachable, that lead is not marked customer-no-answer and can be reassigned under policy. Operators can see which resource limited progress.

Quality Does Not Scale Automatically

More agents can increase throughput only when training, scripts, supervision, devices, and business processes also scale. Rapidly adding unprepared people may produce more connected calls but worse outcomes. Campaign reporting should distinguish telephony connection from the business result.

Access should follow role and purpose. Agents need only the lead context required for the conversation. Supervisors need clear procedures for pause, reassignment, complaints, suppression, and incident response. The customer remains responsible for lawful outreach, staffing, provider terms, and data protection.

Functional Testing at Real Capacity

A one-call test proves only one path. Before increasing capacity, test each agent type, caller-ID assignment, connection flow, no-answer case, agent failure, recording choice, and two-way audio. Then run a controlled concurrency test within provider and appliance limits.

Observe inbound impact, channel use, pacing, duplicate prevention, reporting, and pause behaviour. A configuration that works sequentially may expose resource errors under concurrency. Increase in controlled stages and preserve evidence of the accepted level.

Multi-Agent Capacity Checklist

• Count agents who are currently eligible and available, not merely assigned.
• Allow one active human conversation per available agent.
• Calculate external legs for the selected connection flow.
• Confirm provider channels, appliance limits, and inbound headroom.
• Validate each agent’s endpoint or mobile and caller-ID pool.
• Use per-agent pacing rather than uncontrolled global batches.
• Classify agent failures separately from customer failures.
• Use atomic lead claiming and safe reassignment.
• Test each lane and controlled concurrency before scaling.
• Scale training, privacy controls, and supervision with the team.

Multiple agents add capacity only when each complete call lane has a ready person and the required telephony resources. The scheduler’s job is to respect the smallest current limit.

Calculate and Accept Capacity in Stages

Create a capacity worksheet for each agent type. Record the number of eligible people, external legs per active flow, provider channels, reserved inbound headroom, appliance limit, caller-ID availability, and any recording or media constraint. The safe concurrent total is governed by the tightest resource after reservations, not by whichever figure looks largest.

Begin with one lane and verify the complete customer and agent experience. Add lanes in controlled steps while observing channel use, endpoint state, audio, recording, call events, pacing, and inbound service. Stop increasing when a defined threshold or error appears. Preserve the highest tested and accepted level instead of publishing a theoretical maximum.

Fairness and lead ownership

Capacity controls should not let one fast lane monopolise all eligible leads if the approved business policy requires fair distribution. At the same time, fairness must not create duplicate ownership or hold leads indefinitely for unavailable agents. Deterministic claiming, expiry, and reassignment rules need precise state transitions and evidence.

Changing the team

When agents join, leave, or move between extension and mobile flows, revalidate devices, caller IDs, permissions, training, and channel calculations. Removing an agent should not erase historical attribution. Adding a person to the roster should not increase live capacity until that lane passes its functional test.

Operational Monitoring During the Campaign

Supervisors should see which lanes are active, pacing, paused, agent-unavailable, channel-blocked, caller-ID-blocked, or awaiting work. These states help distinguish a staffing issue from a provider or configuration limit. Aggregate counts alone can make a campaign look slow without revealing the responsible resource.

Define thresholds for repeated agent failures, channel exhaustion, audio faults, or duplicate-risk reconciliation and pause new admission when required. Do not allow pressure for throughput to override the one-conversation-per-person rule, inbound headroom, suppression updates, or the campaign’s approved pace.